This document provides an overview of inertial navigation systems. It begins by introducing two-dimensional motion on a flat earth using accelerometers and gyros to measure linear acceleration and angular rates. It then expands the discussion to three-dimensional motion by incorporating rotation of the earth. Key concepts covered include coordinate frame transformations, error sources from sensors, and the use of filtering techniques like the Kalman filter to increase accuracy. Appendices provide more details on topics like sensor errors, coordinate transformations, and modeling the rotating earth.
1) The document presents the implementation of a Generalized Multiplicative Extended Kalman Filter (GMEKF) for state estimation of a quadcopter using real sensor data. The GMEKF uses quaternion representations and accounts for nonlinearities and sensor noise.
2) Testing the GMEKF on recorded quadcopter flight data showed the filter did not converge, likely due to inconsistencies in the initial sensor and system noise matrices used. Further work is needed to properly implement the GMEKF equations and account for quaternion geometry.
3) Proper implementation of invariant extended Kalman filters like the GMEKF has been shown to provide good state estimation for various quadcopter trajectories by exploiting symmetries in the system dynamics.
The document summarizes key concepts from Chapter 2 of a Physics textbook on kinematics of linear motion. It discusses the following in 3 sentences:
Linear motion can be one-dimensional or two-dimensional projectile motion. Equations of motion include relationships between displacement, velocity, acceleration, and time. Uniformly accelerated motion follows equations that relate the initial and final velocity, acceleration, and time to determine displacement and distance traveled.
In this unit we will analyze the plane kinematics of a rigid body
➢The study is very important for the design of gears, cams and
mechanisms, often in mechanical operations,
THE PLANE MOVEMENT. It is when all the particles of a
rigid bodies move along trajectories that are
equidistant from a fixed plane, the body is said to experience
fixed plane motion
This document describes a project to design and build a model that demonstrates kinematics in tangential and normal coordinates. The student aims to apply error calculations, analyze physical variables like position, velocity, and acceleration, and construct the equation that defines the particle's trajectory. Experimental data is collected from the model over 10 trials and calculations are shown to determine values like final position, velocity, and accelerations (normal, centrifugal, tangential, and total). Conclusions state that the model successfully analyzed the tangential and normal components of motion and applied error theory to determine average revolutions. Characteristics distinguishing uniform and accelerated circular motion are also described.
The document contains sections from a physics textbook chapter on two-dimensional kinematics and projectile motion. It includes sample problems, equations, and explanations of concepts such as relative velocity. Sections cover topics like equations of motion in two dimensions, projectile motion under gravity, calculating maximum height and time of flight, and using relative velocity to solve problems involving objects in motion relative to different frames of reference.
Distance is the total path length travelled, regardless of direction. Displacement describes both the distance and direction from the starting point. Speed is a scalar quantity referring to distance travelled per unit time, while velocity is a vector quantity referring to displacement per unit time, including direction. Examples show calculating distances, displacements, speeds and velocities for objects moving in one or more directions. Bearings are used to describe angular displacement from north in a clockwise direction.
This document discusses key kinematic concepts including displacement, speed, velocity, acceleration, average velocity, instantaneous velocity, and uniformly accelerated motion. It defines these terms and discusses how to calculate them using equations of motion. Graphical representations of motion like distance-time graphs and velocity-time graphs are also covered. The effects of air resistance and gravity are summarized.
This document provides an overview of key physics concepts and mathematical tools. It covers units of measurement in the SI system, vector notation and operations like addition/subtraction, trigonometry, and dimensional analysis. Example problems demonstrate various concepts like finding components of vectors and adding multiple vectors. The document concludes with additional mathematical rules and functions important for physics problems.
1) The document presents the implementation of a Generalized Multiplicative Extended Kalman Filter (GMEKF) for state estimation of a quadcopter using real sensor data. The GMEKF uses quaternion representations and accounts for nonlinearities and sensor noise.
2) Testing the GMEKF on recorded quadcopter flight data showed the filter did not converge, likely due to inconsistencies in the initial sensor and system noise matrices used. Further work is needed to properly implement the GMEKF equations and account for quaternion geometry.
3) Proper implementation of invariant extended Kalman filters like the GMEKF has been shown to provide good state estimation for various quadcopter trajectories by exploiting symmetries in the system dynamics.
The document summarizes key concepts from Chapter 2 of a Physics textbook on kinematics of linear motion. It discusses the following in 3 sentences:
Linear motion can be one-dimensional or two-dimensional projectile motion. Equations of motion include relationships between displacement, velocity, acceleration, and time. Uniformly accelerated motion follows equations that relate the initial and final velocity, acceleration, and time to determine displacement and distance traveled.
In this unit we will analyze the plane kinematics of a rigid body
➢The study is very important for the design of gears, cams and
mechanisms, often in mechanical operations,
THE PLANE MOVEMENT. It is when all the particles of a
rigid bodies move along trajectories that are
equidistant from a fixed plane, the body is said to experience
fixed plane motion
This document describes a project to design and build a model that demonstrates kinematics in tangential and normal coordinates. The student aims to apply error calculations, analyze physical variables like position, velocity, and acceleration, and construct the equation that defines the particle's trajectory. Experimental data is collected from the model over 10 trials and calculations are shown to determine values like final position, velocity, and accelerations (normal, centrifugal, tangential, and total). Conclusions state that the model successfully analyzed the tangential and normal components of motion and applied error theory to determine average revolutions. Characteristics distinguishing uniform and accelerated circular motion are also described.
The document contains sections from a physics textbook chapter on two-dimensional kinematics and projectile motion. It includes sample problems, equations, and explanations of concepts such as relative velocity. Sections cover topics like equations of motion in two dimensions, projectile motion under gravity, calculating maximum height and time of flight, and using relative velocity to solve problems involving objects in motion relative to different frames of reference.
Distance is the total path length travelled, regardless of direction. Displacement describes both the distance and direction from the starting point. Speed is a scalar quantity referring to distance travelled per unit time, while velocity is a vector quantity referring to displacement per unit time, including direction. Examples show calculating distances, displacements, speeds and velocities for objects moving in one or more directions. Bearings are used to describe angular displacement from north in a clockwise direction.
This document discusses key kinematic concepts including displacement, speed, velocity, acceleration, average velocity, instantaneous velocity, and uniformly accelerated motion. It defines these terms and discusses how to calculate them using equations of motion. Graphical representations of motion like distance-time graphs and velocity-time graphs are also covered. The effects of air resistance and gravity are summarized.
This document provides an overview of key physics concepts and mathematical tools. It covers units of measurement in the SI system, vector notation and operations like addition/subtraction, trigonometry, and dimensional analysis. Example problems demonstrate various concepts like finding components of vectors and adding multiple vectors. The document concludes with additional mathematical rules and functions important for physics problems.
Ekeeda Provides Online Civil Engineering Degree Subjects Courses, Video Lectures for All Engineering Universities. Video Tutorials Covers Subjects of Mechanical Engineering Degree.
This document discusses kinematics, which is the geometry of motion without considering forces. It defines key concepts like displacement, velocity, acceleration, and their relationships. It presents four kinematic equations and provides examples of using these equations and graphs of position-time and velocity-time to solve kinematics problems for objects undergoing uniform and non-uniform acceleration.
This document provides an overview of key concepts in rotational kinematics covered in Chapter 8, including angular displacement, velocity, and acceleration. It defines these rotational variables and their relationships to linear motion. Examples are given to illustrate calculating angular variables and transforming between rotational and tangential linear motion for objects like rolling wheels or helicopter blades. Formulas for rotational kinematics with constant angular acceleration are also presented.
Mechanics is the branch of physics that deals with the study of motion and forces on objects, and it is classified into statics, dynamics, and kinematics; statics concerns equilibrium, dynamics concerns forces on moving objects, and kinematics concerns motion without forces. Mechanics studies the motion of macroscopic bodies using concepts like position, displacement, distance, vectors, and methods for adding vectors like the component method using trigonometry or the graphical head-to-tail method.
This document provides an overview of Newtonian mechanics and one-dimensional kinematics. It defines key terms like position, velocity, acceleration, displacement, distance, speed, average speed, average velocity, instantaneous velocity, constant acceleration, and the kinematic equations. It includes examples of how to use the kinematic equations to solve problems involving constant acceleration. There are also sample problems assessing understanding of concepts like displacement vs distance, velocity, acceleration, and interpreting graphs of kinematic variables.
Kinematics of a Particle document discusses:
1) Kinematics involves describing motion without considering forces, studying how position, velocity, and acceleration change over time for a particle.
2) Rectilinear motion involves a particle moving along a straight line, where position (x) is defined as the distance from a fixed origin, velocity (v) is the rate of change of position over time, and acceleration (a) is the rate of change of velocity over time.
3) Examples are provided to demonstrate solving kinematics problems using differentiation, integration, and relationships between position, velocity, acceleration graphs. Problems involve determining velocity, acceleration, distance or displacement given various relationships between these quantities.
This document discusses key terms and equations related to rectilinear motion. Rectilinear motion refers to motion along a straight line. Kinematics deals with the motion of bodies without considering forces. Important concepts discussed include displacement, average and instantaneous velocity, acceleration, distance traveled, and equations of motion. Graphical representations of motion using velocity-time graphs are also presented for different scenarios including uniform velocity, variable velocity from rest to a final velocity, and variable velocity between two points.
The document discusses concepts related to motion including speed, velocity, acceleration, and free fall. It defines key terms, provides examples of calculating speed, acceleration, and distance using formulas like the relationship between velocity and acceleration under gravity. Examples include calculating the impact speed of cars moving in the same direction and the acceleration and distance traveled by objects in free fall.
The document discusses particle kinematics and concepts such as displacement, velocity, acceleration, and their relationships for rectilinear and curvilinear motion. Key concepts covered include definitions of displacement, average and instantaneous velocity, acceleration, graphical representations of position, velocity, and acceleration over time, and analytical methods for solving kinematic equations involving constant or variable acceleration. Several sample problems are provided to illustrate applying these kinematic concepts and relationships to solve for variables like time, velocity, acceleration, and displacement given relevant conditions.
This document provides an overview of planar kinematics of rigid body motion. It describes three types of planar rigid body motion: translation, rotation about a fixed axis, and general plane motion. Translation can be rectilinear or curvilinear. Rotation about a fixed axis involves angular position, velocity, acceleration, and the motion of a point on the rotating body. General plane motion is a combination of translation and rotation. Formulas are provided for analyzing velocity and acceleration during these different types of motion. Examples are also given to demonstrate how to apply the kinematic equations.
The document discusses equations for calculating velocity, acceleration, displacement, and time for objects moving with constant velocity or uniform acceleration. It provides the key equations:
1) For constant velocity, average velocity (v) equals displacement (Δx) over time (Δt), and displacement equals velocity times time.
2) For uniform acceleration, final velocity (vf) equals initial velocity (v0) plus acceleration (a) times time (Δt), and displacement equals initial velocity times time plus one-half acceleration times time squared.
3) A single equation relates displacement, initial velocity, final velocity, and acceleration, which can be rearranged to solve for any of those variables.
This document discusses kinetics of particles and Newton's laws of motion. It introduces the concept of equation of motion relating the forces acting on a particle to its acceleration. Equations of motion are developed for rectangular, normal-tangential, and cylindrical coordinate systems. Examples are provided to demonstrate solving equations of motion for particles undergoing accelerated motion under various force conditions in different coordinate systems.
Curvilinear motion occurs when a particle moves along a curved path.
Since this path is often described in three dimensions, vector analysis will
be used to formulate the particle's position, velocity, and acceleration
Chapter 2 introduces the concepts of kinematics including reference frames, displacement, velocity, acceleration, and motion with constant acceleration. Equations are derived that relate displacement, velocity, acceleration, and time for objects undergoing constant acceleration. Near the Earth's surface, the acceleration due to gravity is approximately 9.80 m/s2, so these equations can be applied to analyze falling or projected objects using this value of acceleration.
This document provides an introduction to kinematics, the branch of physics that deals with the motion of objects. It explains key concepts like coordinate systems, equations of motion, and how to use the kinematics equations to solve problems related to projectile motion. Examples are worked through, such as calculating the maximum height and velocity of a ball thrown upward. Practice problems are also provided. The overall purpose is to teach readers the basics of kinematics and show how to apply the equations of motion.
study of yaw and pitch control in quad copter PranaliPatil76
This document describes an experimental study of yaw and pitch control in quadcopters. It discusses how quadcopters use independent variation of rotor speeds to control pitch, roll, and yaw. Yaw control is achieved by varying the net torque on the quadcopter by increasing the thrust of rotors spinning in one direction compared to the other. Pitch control is achieved by varying the net center of thrust. The document provides equations for calculating torque and describes how adjusting rotor thrusts can cause the quadcopter to yaw or rotate in different directions.
Describes displacement, velocity, acceleration as vectors and distance and speed as scalars, Show all needed equations and their use.
**More good stuff available at:
www.wsautter.com
and
http://www.youtube.com/results?search_query=wnsautter&aq=f
This document provides an overview of graphing motion in one dimension. It discusses position versus time graphs, velocity versus time graphs, and acceleration versus time graphs. Key points include:
- The slope of a position-time graph represents velocity, and the slope of a velocity-time graph represents acceleration.
- Straight lines on position-time graphs indicate uniform motion with constant velocity.
- The area under a velocity-time graph represents displacement.
- Kinematic equations allow calculations of variables like position, velocity, and acceleration given information about an object's motion under constant acceleration.
Physics 504 Chapter 9 Uniform Rectilinear MotionNeil MacIntosh
This document discusses uniform rectilinear motion. It defines different types of motion including rectilinear, curvilinear, and random motion. Distance is defined as a scalar quantity representing how far an object has moved, while displacement is a vector quantity that includes both distance and direction. Uniform motion refers to motion at a constant speed in a single direction. Graphs of distance-time and velocity-time relationships are used to analyze motion. The average velocity and speed can be calculated from these graphs by determining slope.
Christian Kasumo gave a presentation at the ZAME Annual Provincial Conference on teaching kinematics. He defined key kinematics concepts like displacement, velocity, acceleration, and discussed common student difficulties. He recommended teaching kinematics through real-world examples, group work, and activities using software like Geogebra to help students understand graphs and equations of motion. He concluded by thanking the organizers and Mulungushi University for their support.
The document repeatedly mentions the names Cleyson Conrado across multiple lines. It provides contact information at the end for Cleyson Conrado, including phone numbers, a Skype account, and email for scheduling contact. The document encourages signing up now on a website for Cleyson Conrado.
Ekeeda Provides Online Civil Engineering Degree Subjects Courses, Video Lectures for All Engineering Universities. Video Tutorials Covers Subjects of Mechanical Engineering Degree.
This document discusses kinematics, which is the geometry of motion without considering forces. It defines key concepts like displacement, velocity, acceleration, and their relationships. It presents four kinematic equations and provides examples of using these equations and graphs of position-time and velocity-time to solve kinematics problems for objects undergoing uniform and non-uniform acceleration.
This document provides an overview of key concepts in rotational kinematics covered in Chapter 8, including angular displacement, velocity, and acceleration. It defines these rotational variables and their relationships to linear motion. Examples are given to illustrate calculating angular variables and transforming between rotational and tangential linear motion for objects like rolling wheels or helicopter blades. Formulas for rotational kinematics with constant angular acceleration are also presented.
Mechanics is the branch of physics that deals with the study of motion and forces on objects, and it is classified into statics, dynamics, and kinematics; statics concerns equilibrium, dynamics concerns forces on moving objects, and kinematics concerns motion without forces. Mechanics studies the motion of macroscopic bodies using concepts like position, displacement, distance, vectors, and methods for adding vectors like the component method using trigonometry or the graphical head-to-tail method.
This document provides an overview of Newtonian mechanics and one-dimensional kinematics. It defines key terms like position, velocity, acceleration, displacement, distance, speed, average speed, average velocity, instantaneous velocity, constant acceleration, and the kinematic equations. It includes examples of how to use the kinematic equations to solve problems involving constant acceleration. There are also sample problems assessing understanding of concepts like displacement vs distance, velocity, acceleration, and interpreting graphs of kinematic variables.
Kinematics of a Particle document discusses:
1) Kinematics involves describing motion without considering forces, studying how position, velocity, and acceleration change over time for a particle.
2) Rectilinear motion involves a particle moving along a straight line, where position (x) is defined as the distance from a fixed origin, velocity (v) is the rate of change of position over time, and acceleration (a) is the rate of change of velocity over time.
3) Examples are provided to demonstrate solving kinematics problems using differentiation, integration, and relationships between position, velocity, acceleration graphs. Problems involve determining velocity, acceleration, distance or displacement given various relationships between these quantities.
This document discusses key terms and equations related to rectilinear motion. Rectilinear motion refers to motion along a straight line. Kinematics deals with the motion of bodies without considering forces. Important concepts discussed include displacement, average and instantaneous velocity, acceleration, distance traveled, and equations of motion. Graphical representations of motion using velocity-time graphs are also presented for different scenarios including uniform velocity, variable velocity from rest to a final velocity, and variable velocity between two points.
The document discusses concepts related to motion including speed, velocity, acceleration, and free fall. It defines key terms, provides examples of calculating speed, acceleration, and distance using formulas like the relationship between velocity and acceleration under gravity. Examples include calculating the impact speed of cars moving in the same direction and the acceleration and distance traveled by objects in free fall.
The document discusses particle kinematics and concepts such as displacement, velocity, acceleration, and their relationships for rectilinear and curvilinear motion. Key concepts covered include definitions of displacement, average and instantaneous velocity, acceleration, graphical representations of position, velocity, and acceleration over time, and analytical methods for solving kinematic equations involving constant or variable acceleration. Several sample problems are provided to illustrate applying these kinematic concepts and relationships to solve for variables like time, velocity, acceleration, and displacement given relevant conditions.
This document provides an overview of planar kinematics of rigid body motion. It describes three types of planar rigid body motion: translation, rotation about a fixed axis, and general plane motion. Translation can be rectilinear or curvilinear. Rotation about a fixed axis involves angular position, velocity, acceleration, and the motion of a point on the rotating body. General plane motion is a combination of translation and rotation. Formulas are provided for analyzing velocity and acceleration during these different types of motion. Examples are also given to demonstrate how to apply the kinematic equations.
The document discusses equations for calculating velocity, acceleration, displacement, and time for objects moving with constant velocity or uniform acceleration. It provides the key equations:
1) For constant velocity, average velocity (v) equals displacement (Δx) over time (Δt), and displacement equals velocity times time.
2) For uniform acceleration, final velocity (vf) equals initial velocity (v0) plus acceleration (a) times time (Δt), and displacement equals initial velocity times time plus one-half acceleration times time squared.
3) A single equation relates displacement, initial velocity, final velocity, and acceleration, which can be rearranged to solve for any of those variables.
This document discusses kinetics of particles and Newton's laws of motion. It introduces the concept of equation of motion relating the forces acting on a particle to its acceleration. Equations of motion are developed for rectangular, normal-tangential, and cylindrical coordinate systems. Examples are provided to demonstrate solving equations of motion for particles undergoing accelerated motion under various force conditions in different coordinate systems.
Curvilinear motion occurs when a particle moves along a curved path.
Since this path is often described in three dimensions, vector analysis will
be used to formulate the particle's position, velocity, and acceleration
Chapter 2 introduces the concepts of kinematics including reference frames, displacement, velocity, acceleration, and motion with constant acceleration. Equations are derived that relate displacement, velocity, acceleration, and time for objects undergoing constant acceleration. Near the Earth's surface, the acceleration due to gravity is approximately 9.80 m/s2, so these equations can be applied to analyze falling or projected objects using this value of acceleration.
This document provides an introduction to kinematics, the branch of physics that deals with the motion of objects. It explains key concepts like coordinate systems, equations of motion, and how to use the kinematics equations to solve problems related to projectile motion. Examples are worked through, such as calculating the maximum height and velocity of a ball thrown upward. Practice problems are also provided. The overall purpose is to teach readers the basics of kinematics and show how to apply the equations of motion.
study of yaw and pitch control in quad copter PranaliPatil76
This document describes an experimental study of yaw and pitch control in quadcopters. It discusses how quadcopters use independent variation of rotor speeds to control pitch, roll, and yaw. Yaw control is achieved by varying the net torque on the quadcopter by increasing the thrust of rotors spinning in one direction compared to the other. Pitch control is achieved by varying the net center of thrust. The document provides equations for calculating torque and describes how adjusting rotor thrusts can cause the quadcopter to yaw or rotate in different directions.
Describes displacement, velocity, acceleration as vectors and distance and speed as scalars, Show all needed equations and their use.
**More good stuff available at:
www.wsautter.com
and
http://www.youtube.com/results?search_query=wnsautter&aq=f
This document provides an overview of graphing motion in one dimension. It discusses position versus time graphs, velocity versus time graphs, and acceleration versus time graphs. Key points include:
- The slope of a position-time graph represents velocity, and the slope of a velocity-time graph represents acceleration.
- Straight lines on position-time graphs indicate uniform motion with constant velocity.
- The area under a velocity-time graph represents displacement.
- Kinematic equations allow calculations of variables like position, velocity, and acceleration given information about an object's motion under constant acceleration.
Physics 504 Chapter 9 Uniform Rectilinear MotionNeil MacIntosh
This document discusses uniform rectilinear motion. It defines different types of motion including rectilinear, curvilinear, and random motion. Distance is defined as a scalar quantity representing how far an object has moved, while displacement is a vector quantity that includes both distance and direction. Uniform motion refers to motion at a constant speed in a single direction. Graphs of distance-time and velocity-time relationships are used to analyze motion. The average velocity and speed can be calculated from these graphs by determining slope.
Christian Kasumo gave a presentation at the ZAME Annual Provincial Conference on teaching kinematics. He defined key kinematics concepts like displacement, velocity, acceleration, and discussed common student difficulties. He recommended teaching kinematics through real-world examples, group work, and activities using software like Geogebra to help students understand graphs and equations of motion. He concluded by thanking the organizers and Mulungushi University for their support.
The document repeatedly mentions the names Cleyson Conrado across multiple lines. It provides contact information at the end for Cleyson Conrado, including phone numbers, a Skype account, and email for scheduling contact. The document encourages signing up now on a website for Cleyson Conrado.
Este documento resume dos cuerpos normativos principales de la legislación venezolana relacionados con la protección, higiene y seguridad ocupacional: la Ley Orgánica del Trabajo y la Ley Orgánica de Prevención, Condiciones y Medio Ambiente de Trabajo. La Ley Orgánica del Trabajo establece la responsabilidad de los empleadores por los accidentes laborales y enfermedades ocupacionales. La Ley Orgánica de Prevención, Condiciones y Medio Ambiente de Trabajo tiene como objetivo garantizar condiciones de seg
El documento define la isometría como una transformación que no altera la forma ni el tamaño de una figura, solo involucra un cambio de posición. Describe tres tipos principales de transformaciones isométricas: traslaciones, que implican un cambio de lugar determinado por un eje; rotaciones, que implican un cambio en la orientación alrededor de un punto fijo; y simetrías, que implican una correspondencia exacta de puntos con respecto a un centro, eje o plano.
El diseño tecnopedagógico (DTP) surge de la necesidad de optimizar los procesos de enseñanza-aprendizaje para garantizar su éxito. Es por eso que, necesariamente, tiene que integrar en su intervención los diferentes aspectos que convergen en la programación de intenciones pedagógicas y en la generación de los recursos materiales. Por lo tanto, el DTP acontece un concepto en el que confluyen aspectos tan aparentemente dispares como la psicología, las teorías del aprendizaje, el diseño o las tecnologías de la información y la comunicación.
El éxito de la vida se trata de lograr las metas personales a través del esfuerzo y la perseverancia, y presentado por Luz Marina Cifuentes con el código 201422742.
Digital Analytics Association Symposium- Research, Analytics and Testing for ...Oxford Tech + UX
Digital Analytics Association Symposium November 13, 2014 in San Francisco, CA
Research, Analytics and Testing for Product Design: Choosing the best Approach for the Job
La pestaña Revisar en Excel proporciona herramientas para revisar la ortografía, buscar sinónimos, traducir texto y agregar comentarios. Se organiza en grupos como Revisión, Comentarios y Cambios. Permite proteger hojas de cálculo con contraseña. La vista Diseño de página muestra cómo se verá el documento al imprimirse.
Résultats enquêtes : DIAGNOSTIC DE LA SITUATION ACTUELLE ET IDENTIFICATION DE...Youssef CHARHABAILI
Le présent travail, qui s’intègre dans le programme de recherche et développement établi entre l’Association Marocaine des Productions Bio (AMABIO) et l’Institut Agronomique et Vétérinaire Hassan II, présente un résumé très succinct des résultats d’une enquête auprès des producteurs.
L’objectif étant de diagnostiquer la situation actuelle et identifier les problèmes liés aux pratiques culturales en agriculture biologique à travers une enquête dans cinq régions différentes du Maroc.
The document describes the rectangular coordinate system. It establishes that a coordinate system assigns positions in a plane using ordered pairs of numbers (x,y). It defines the x-axis, y-axis, and origin at their intersection. Any point is addressed by its coordinates (x,y) where x represents horizontal distance from the origin and y represents vertical distance. The four quadrants divided by the axes are also defined based on positive and negative coordinate values. Reflections of points across the axes and origin are discussed. Finally, it introduces the concept of graphing mathematical relations between x and y coordinates to represent collections of points.
El documento habla sobre las nuevas tecnologías del futuro. Discutirá cómo la tecnología permitirá realizar múltiples tareas simultáneamente en el trabajo y la educación, y ayudará a compartir información a distancia, como conferencias empresariales. También abordará cómo la innovación tecnológica apunta a facilitar el trabajo y pagos en línea con moneda electrónica y una comunicación más directa.
This document discusses Communicative Language Teaching (CLT), an approach that emphasizes using the target language for meaningful communication. It focuses on functions of language, provides opportunities for students to negotiate meaning, and engages students in role-plays, dialogues, and other interactive activities. While CLT had early successes, it was later criticized for excluding grammar instruction and failing to develop student accuracy. The approach also faced issues with cultural adaptation and commercialization. The document concludes by discussing communicative approaches as an evolution of CLT that balances functional language use with different classroom techniques.
El documento describe el comercio electrónico y los medios sociales. Explica que el comercio electrónico permite realizar transacciones en línea de forma rápida y económica. También describe los tipos de comercio electrónico, las ventajas para las empresas y los desafíos de la confianza y seguridad. Además, explica que los medios sociales permiten compartir información de forma bidireccional y que los blogs y redes sociales son herramientas importantes.
El documento explica los conceptos fundamentales de los derechos de autor, incluyendo que son derechos exclusivos que otorgan a los creadores sobre sus obras originales, la historia de cómo se desarrollaron legalmente y los tipos de derechos (morales y patrimoniales). También describe los símbolos utilizados para indicar la protección de derechos de autor, los campos a los que aplican estos derechos y la diferencia entre derechos de autor y copyright.
Este documento propone una campaña de promoción turística en la que los turistas pueden tomar fotografías en kioscos ubicados en sitios turísticos y compartirlas automáticamente en redes sociales con el branding del destino. Se implementó con éxito en Cancún con 10,000 turistas y más de 23,000 fotografías compartidas. Ahora se propone expandirla a otros destinos mexicanos como Taxco con proyecciones de 6000 turistas y más de 7,500 testimonios al mes.
La conferencia de Teherán de 1943 acordó apoyar a los partisanos yugoslavos, que Turquía debía unirse a los Aliados antes de fin de año con apoyo soviético, y que la invasión de Europa Occidental comenzaría en mayo de 1944. También se acordaron las fronteras polacas posteriores a la guerra y la creación provisional de las Naciones Unidas.
Action Trajectory Reconstruction for Controlling of Vehicle Using SensorsIOSR Journals
Abstract: Inertial sensors, such as accelerometers and gyro-scopes, are rarely used by themselves to compute
velocity and position as each requires the integration of very noisy data. The variance and bias in the resulting
position and velocity estimates grow un-bounded in time. This paper proposes a solution to provide a de-biased
and de-noised estimation of position and velocity of moving vehicle actions from accelerometer measurements.
The method uses a continuous wavelet transform applied to the measurements recursively to provide reliable
action trajectory reconstruction. The results are presented from experiments performed with a MEMS accelerometer
and gyroscope.
Keywords: Action trajectory, continuous wavelet transform, inertial measurement unit.
1 Lab 3 Newton’s Second Law of Motion Introducti.docxmercysuttle
1
Lab 3: Newton’s Second Law of Motion
Introduction
Newton’s Second law of motion can be summarized by the following equation:
Σ F = m a (1)
where Σ F represents a net force acting on an object, m is the mass of the object moving
under the influence of Σ F, and a is the acceleration of that object. The bold letters in
the equation represent vector quantities.
In this lab you will try to validate this law by applying Eq. 1 to the almost frictionless
motion of a car moving along a horizontal aluminum track when a constant force T
(tension in the string) acts upon it. This motion (to be exact the velocity of the moving
object) will be recorded automatically by a motion sensor. The experimental set up
for a car moving away from the motion sensor is depicted below.
If we consider the frictionless motion of the cart in the positive x-direction chosen in
the diagram, then Newton’s Second Law can be written for each of the objects as
follows:
T Ma (2)
and
– gT F ma (3)
From this system of equations we can get the acceleration of the system:
2
gF
a
m M
(4)
Because the motion of the car is not frictionless, to get better results it is necessary to
include the force of kinetic friction fk experienced by the moving car in the analysis.
When the cart is moving away from the motion detector (positive x-direction in the
diagram) Newton’s Second Law is written as follows for each of the moving objects
m and M:
1 1– kT f Ma (5)
and
1 1– gT F ma (6)
Since it is quite difficult to assess quantitatively the magnitude of kinetic friction
involved in our experiment we will solve the problem by putting the object in two
different situations in which the friction acts in opposite directions respectively while
the tension in the string remains the same.
When the cart M is forced to move towards the motion detector (negative x-direction
in the diagram), the corresponding Newton’s Second Law equations will change as
follows:
2 2kT f Ma (7)
and
2 2gT F ma (8)
Note that in equations 5, 6, 7, and 8 the direction of acceleration represented by vector
a has been chosen in the same direction as the direction of motion.
We are able to eliminate the force of kinetic friction on the final result, by calculating
the mean acceleration from these two runs:
1 2
2
ave
slope slope
a
(9)
Combing the equations (5) – (8) we derive the equation to calculate the value of
gravitational acceleration:
avea M mg
m
(10)
3
Equipment
Horizontal dynamics track with smart pulley and safety stopper on one end; collision
cart with reflector connected to a variable mass hanging over the pulley; motion
detector connected to the Science Workshop interface recording the velocity of the
moving cart.
Procedure:
a) Weigh the cart (M) and the small mass (m) hanger.
b) Open the experiment file “New ...
The document presents a new approach for dynamic analysis of parallel manipulators based on the principle of virtual work. It illustrates the approach using a simple 4-bar linkage example, calculating the inertial forces and moments, virtual displacements, and input torque. It then generalizes the approach for dynamic analysis of a 6 degree-of-freedom parallel manipulator like a Gough-Stewart platform. The approach leads to faster computation than traditional Newton-Euler methods by not requiring calculation of constraint forces between links.
The document describes two algorithms, gradient descent and Gauss-Newton, for determining location from GPS satellite signals. It outlines the process of linearizing the pseudorange equations and deriving the iterative algorithms to minimize error between measured and calculated ranges. Simulation results on synthetic noiseless data show Gauss-Newton converges much faster than gradient descent, in 4 iterations versus over 50,000 iterations. Gauss-Newton is thus determined to be a more optimal algorithm for GPS positioning.
- The document details a state space solver approach for analog mixed-signal simulations using SystemC. It models analog circuits as sets of linear differential equations and solves them using the Runge-Kutta method of numerical integration.
- Two examples are provided: a digital voltage regulator simulation and a digital phase locked loop simulation. Both analog circuits are modeled in state space and simulated alongside a digital design to verify mixed-signal behavior.
- The state space approach allows modeling analog circuits without transistor-level details, improving simulation speed over traditional mixed-mode simulations while still capturing system-level behavior.
The document discusses cinematica (kinematics) in normal and tangential coordinates. It specifically analyzes intrinsic acceleration components, which are the decomposition of the acceleration vector along intrinsic axes. The tangential component is responsible for changes in velocity magnitude, while the normal component is zero for rectilinear motion. Measurement data is presented to calculate errors for analyzing motion along a curved trajectory using normal and tangential coordinate systems. Conclusions discuss considering all variables for calculations and understanding the reference system of the model.
Ayan Das describes single degree of freedom (SDOF) systems in mechanical engineering. An SDOF system is constrained to move in only one direction or axis. It can be modeled as a mass attached to a spring and damper. The motion of an SDOF system can be described by a second-order differential equation. Solving this equation provides the displacement of the mass over time. Real-life examples of SDOF systems include accelerometers. SDOF models allow engineers to evaluate system behavior under different conditions.
Subha Das is a 7th semester mechanical engineering student studying mechanical vibrations. Their document discusses single degree of freedom (SDOF) systems, which describe systems that can only move in one direction or axis. An SDOF system can be modeled as a mass attached to a spring and damper, with its motion described through a differential equation. Solving this equation provides the system's response over time as it oscillates and comes to rest. Real-life examples of SDOF systems include accelerometers. Engineering models allow evaluating how actual systems will perform under different conditions.
This document discusses attitude determination for spacecraft. It begins by explaining that attitude determination uses sensors to measure vector components in the body frame and mathematical models to determine vector components in the inertial frame. These measurements are then used in attitude determination algorithms to estimate the attitude as a quaternion, Euler angles, or rotation matrix. Common sensors include sun sensors and magnetometers. The document focuses on static attitude determination algorithms that do not consider time.
This paper addresses the problem of estimating measurement biases in optical sensor systems using measurements of common targets of opportunity. It presents a method to estimate the roll, pitch, and yaw biases of multiple passive sensors. The method involves iteratively estimating the target positions and sensor biases by maximizing the likelihood function. It is shown that for bias estimation to be possible, the number of measurements must be greater than or equal to the number of targets and sensor biases. For two sensors, bias observability is not guaranteed, but it is for three or more sensors. The Cramér-Rao lower bound is used to quantify the information available about the biases. Simulations show the method is statistically efficient even for small sample sizes.
The document discusses forced vibrations of damped, single degree of freedom linear spring mass systems. It derives the equations of motion for three types of forcing - external forcing, base excitation, and rotor excitation. It presents the steady state solutions and discusses key features, including that the response frequency matches the forcing frequency. The maximum response occurs at resonance when the forcing frequency matches the natural frequency. Engineering applications include designing systems to minimize vibrations by increasing stiffness/natural frequency and damping.
This paper proposed a nonlinear robust control for spacecraft attitude based on passivity and
disturbance suppression vector. The spacecraft model was described using quaternion. The control law
introduced the suppression vector of external disturbances and had no information related to the system
parameters. The desired performance of spacecraft attitude control could be achieved using the designed
control law. And stability conditions of the nonlinear robust control for spacecraft attitude were given. The
stability could be proved by applying Lyapunov approach. The verification of the proposed attitude control
method was performed through a series of simulations. The numerical results showed the effectiveness of
the proposed control method in controlling the spacecraft attitude in the presence of external disturbances.
The main benefit of the proposed attitude control method does not need angular velocity measurement
and has its robustness against model uncertainties and external disturbances.
The document discusses several integration schemes used to model the time evolution of dynamical systems given their initial conditions and equations of motion. It describes the Verlet, leapfrog, and velocity Verlet algorithms, which are explicit integration methods that require only one force evaluation per time step. The document also outlines the Gear predictor-corrector algorithm, an implicit method that predicts positions and velocities before applying corrections.
Oscar Nieves (11710858) Computational Physics Project - Inverted PendulumOscar Nieves
This document describes a numerical simulation of an inverted pendulum system created in MATLAB using a 4th order Runge-Kutta algorithm. The simulation models an inverted pendulum attached to a horizontally moving cart. Forces like air drag and friction are included, and parameters like mass, pendulum length, and initial conditions can be varied. Small changes to initial conditions can lead to large differences in motion, demonstrating the system's chaotic behavior. The document also outlines the methodology for adapting the Runge-Kutta algorithm to solve systems of coupled differential equations.
The document summarizes an experimental and theoretical study of a mechanical system consisting of a sliding car that bounces on an oscillating piston. The researchers were able to produce period-1, period-2, period-3 orbits, sticking solutions, and chaotic behaviors experimentally that agreed well with the theoretical model. They aim to identify parameters that produce different orbit types and find co-existing orbits by varying the piston frequency and amplitude. The experimental setup uses infrared lights and a Nintendo Wii remote to track the car's position over time.
This document summarizes a lab experiment where students modeled a Furuta pendulum system and designed a state-space controller to control the pendulum's motion. Students first created a mathematical model of the system and calculated theoretical controller gains. They then implemented the controller on the physical pendulum system and manually tuned the gains to improve performance. Manual tuning produced better tracking of the pendulum and swing arm positions compared to using the theoretical gains. This demonstrated the challenges of applying a mathematical model to a physical system that can vary over time.
This document contains instructions for a physics lab experiment on moment of inertia. The experiment has two parts:
Part I measures the moment of inertia of a disk by applying a torque from a hanging mass and measuring the angular acceleration.
Part II measures the moment of inertia of a rod with two movable masses by varying the mass positions and amounts and again measuring angular acceleration from a hanging torque source. Equations are provided to calculate moment of inertia from experimental measurements.
This document presents an adaptive Kalman filter for estimating human body orientation using micro-sensors. The filter uses quaternions to represent orientation to avoid singularities. It includes motion acceleration in the state vector to compensate for interference of body motion on gravity measurements. Additionally, it adapts the process noise covariance based on sensor signal variations to optimize performance under human motion. Experiments showed this algorithm had less error than existing methods and that including motion compensation and adaptive mechanisms improved accuracy of human motion capture.
Attitude Control of Quadrotor Using PD Plus Feedforward Controller on SO(3)IJECEIAES
This paper proposes a simple scheme of Proportional-Derivative (PD) plus Feedforward controller on SO(3) to control the attitude of a quadrotor. This controller only needs the measurement of angular velocity to calculate the exponential coordinates of the rotation matrix. With rotation matrix as an error variable of the controller, the simulation shows that the controller is able to drive the attitude of the quadrotor from hovering condition to desired attitude and from an attitude condition goes to the hovering condition, despite the system is disturbed. When the system is convergent, the rotation error matrix will be a 3 3 identity matrix.
This chapter discusses molecular dynamics (MD) simulations, which allow modeling the behavior of atomic and molecular systems by numerically solving Newton's equations of motion. It describes the Verlet algorithm and its variants commonly used to integrate the equations of motion in MD simulations. Analysis of the trajectory data generated by MD simulations can provide information on system properties like pressure, diffusion, and the radial distribution function.
7. NAWCWPNS TM 8128
3
INTRODUCTION
Navigation is the art of knowing where you are, how fast you are moving and in which
direction; and of positioning yourself in relation to your environment in such a way as to maximize
your chances for survival. Navigation may consist of examining the bark on a tree, reading the
display of a GPS receiver, or something in between. In this document we discuss a particular
method of navigation known as inertial navigation.
Inertial navigation is accomplished by integrating the output of a set of sensors to compute
position, velocity, and attitude. The sensors used are gyros and accelerometers. Gyros measure
angular rate with respect to inertial space, and accelerometers measure linear acceleration, again
with respect to an inertial frame. Integration is a simple process, complexities arise due to the
various coordinate frames encountered, sensor errors, and noise in the system.
TWO-DIMENSIONAL MOTION
To begin our discussion, we reduce the problem to its simplest terms by assuming a flat earth
and confining our motion to the surface. To indicate position we must establish a grid or coordinate
system. We label one direction on the surface as N and another, perpendicular to it, as E. If the
third direction is perpendicular to the surface in the down direction, a north-east-down (NED),
right-handed coordinate frame is established.
For the navigating system (platform), we label the longitudinal axis as x and the cross axis as y
(Figure 1). The z axis is down in order to complete a right-handed coordinate system. This
coordinate frame (xyz) moves with the vehicle and is fixed in the body. In general, this frame is
rotated about the vertical with respect to the NED frame by some azimuth angle, ψ.
N
E
y
x
ψ
FIGURE 1. Coordinate Frames.
Assume that, initially, the x axis is oriented north and the y axis is oriented east. (We will
consider alignment later.) To track our motion with respect to the grid, we mount an accelerometer
on the x axis and another on the y axis. (This is a strapdown configuration in which the inertial
sensors are fixed with respect to the body of the navigating system.) If we assume that the NED
frame is an inertial frame, the accelerometers will measure acceleration with respect to that frame,
8. NAWCWPNS TM 8128
4
but the output will be in the xyz frame. Therefore, we will need to know the azimuth angle to
navigate in the NED frame. We can sense the change in the azimuth angle with a gyro mounted on
the z axis. The output of this gyro is ωz = dψ/dt. Since ψ is initially zero, the integral of ωz gives us
ψ as a function of time. For now we assume perfect sensors. Once we have ψ, we can rotate the
accelerometer outputs to the NED frame.
a
a
a
a
N
E
x
y
=
−
cos sin
sin cos
ψ ψ
ψ ψ
(1)
Then we can easily integrate aN and aE, once to obtain the velocity in the NED frame and again to
obtain position.
In a practical system, we must have some means of determining our initial position, velocity,
and attitude; integrations require initial conditions. If we are at rest, we know our initial velocity
(zero) and we can measure our initial position. The initial azimuth angle also must be measured by
some external means as no system error is produced as a result of an initial misalignment. By
system error we mean position, velocity, and attitude errors. Thus, for our flat earth model, a
stationary alignment must be accomplished by external measurements. If our platform is moving
with a constant velocity, we also must have an external measurement of velocity. If there is no
accelerated motion, there is no observability of our system errors. Observability means that some
condition exists that excites one or more of the system error states and causes a measurable
response of the system. An example is given in the following.
Suppose we have a co-moving platform (i.e., our platform and a reference platform are
attached to the same structure) that has been accurately aligned. We use information from this
reference system to align our platform. This process is transfer alignment. When we refer to
alignment from this point on, we will mean velocity-matching alignment unless otherwise
specified. In a velocity-matching alignment, we compare our system velocity with the velocity of
some external reference. If an alignment error or sensor error propagates into a velocity error, then
the error is observable for this type of alignment. Note that observability may depend on the
conditions under which the alignment is being performed, as well as the type of alignment. We will
touch on other types of alignment in the appendix on Kalman filtering (Appendix A).
If we accelerate the system now, any azimuth error becomes observable. Let
ˆψ ψ δψ= + (2)
be our estimate of azimuth, where ψ is the true azimuth angle and δψ is the error in our estimate.
We measure a change in velocity due to acceleration of
∆
∆
r
r
V
a
a
dt
a
a
dt
V
m
t
t
x
y
t
t
x
y
=
−
≅
−
−
=
−
∫
∫
1
2
1
2
1
1
1
1
cos ˆ sin ˆ
sin ˆ cos ˆ
cos sin
sin cos
ψ ψ
ψ ψ
δψ
δψ
ψ ψ
ψ ψ
δψ
δψ
(3)
9. NAWCWPNS TM 8128
5
where
∆
r
V = true change in velocity in the NED frame
∆
r
Vm = measured change in velocity in NED frame.
We have used Equation (1) and the approximations
cos(ψ+δψ) ≅ cosψ - δψsinψ
and
sin(ψ+δψ) ≅ sinψ + δψcosψ.
Since we know ∆V from our reference system, we have a direct measure of our azimuth error.
If no noise is present in the system, we could determine ψ as accurately as we pleased with one
measurement by taking higher order expansions of the trigonometric functions in Equation (3)
(assuming we could solve the resulting nth-order polynomials in δψ). If noise is present in the
system, we must use filtering techniques to increase the accuracy of our alignment. The most
common method of filtering in navigation systems is the Kalman filter, which is discussed in
Appendix A. Note that Equation (3) is overdetermined; we have two equations in one unknown. If
we were to estimate δψ from a noisy measurement, we would use least-squares techniques.
So far we have assumed perfect sensors. Some common gyro and accelerometer error models
are discussed in Appendix B. In general, different error sources generate navigation errors that are
integral or half-integral powers of time. An accelerometer bias, for example, generates a velocity
error that is linear in time and a position error that is quadratic in time. A bias on the vertical gyro
would generate an azimuth error that is linear in time. The velocity and position errors due to this
heading error would be a function of the accelerations present (Equation (3)). A common
misconception about how inertial navigation systems work is that the heading error translates
directly to position error. We can see from Equation (3) that this is not so. If we know the north
and east components of velocity, we will navigate accurately, no matter how erroneous our
heading estimate is, as long as we do not accelerate. A human operator might use azimuth to steer
by, but that would mean he was heading to the wrong place. The navigation system would know
its own position. Of course, as soon as the system accelerated due to a course change or for some
other reason, the heading error would generate velocity and position errors.
THREE-DIMENSIONAL MOTION
Our two-dimensional model exhibits, to a limited extent, all the characteristics of an inertial
navigation system. The next level of complexity we consider is to remove the constraint of two-
dimensional motion. Once we remove this constraint, we also turn on a gravitational field that is
normal to the north-east plane. This field produces an acceleration of approximately 32.2 ft/s2
in
the down direction.
A short digression is necessary at this point to explain how an accelerometer measures
gravitational acceleration. No matter how an accelerometer is constructed, we may think of it as
shown in Figure 2.
10. NAWCWPNS TM 8128
6
m
+
FIGURE 2. Accelerometer.
The accelerometer consists of a proof mass, m, suspended from a case by a pair of springs.
The arrow indicates the input axis. An acceleration along this axis will cause the proof mass to be
displaced from its equilibrium position. This displacement will be proportional to the acceleration.
The amount of displacement from the equilibrium position is sensed by a pick-off and scaled to
provide an indication of acceleration along this axis. The equilibrium position of the proof mass is
calibrated for zero acceleration. An acceleration in the plus direction will cause the proof mass to
move downward with respect to the case. This downward movement indicates positive
acceleration. Now imagine that the accelerometer is sitting on a bench in a gravitational field. We
see that the proof mass is again displaced downward with respect to the case, which indicates
positive acceleration. However, the gravitational acceleration is downward. Therefore, the output
of an accelerometer due to a gravitational field is the negative of the field acceleration. The output
of an accelerometer is called the specific force and is given by
f = a - g (4)
where
f = specific force
a = acceleration with respect to the inertial frame
g = gravitational acceleration.
This relation is the cause of much confusion. The easy way to remember this relation is to think
of one of two cases. If the accelerometer is sitting on a bench, it is at rest so a is zero. The force on
the accelerometer is the normal force of reaction of the bench on the case or negative g. Or
imagine-dropping the accelerometer in a vacuum. In this case f reads zero and the actual
acceleration is a = g. To navigate with respect to the inertial frame, we need a, which is why, in
the navigation equations, we convert the output of the accelerometers from f to a by adding g.
Now, we return to our consideration of three-dimensional motion over a flat earth. The
platform attitude can no longer be specified by just the azimuth angle. We will specify the
11. NAWCWPNS TM 8128
7
orientation of the platform by azimuth, pitch, and roll, as shown in Figure 3. In this document,
platform refers to the body axes of the system being aligned. These axes are the same as the xyz
coordinate frame previously defined.
PITCH
NORTH
HEADING
EAST
HEADING
ROLL
ROLL
PITCH
Xa
DOWN
Za
Ya
FIGURE 3. Platform Orientation.
The transformation matrix from the NED frame to the xyz frame (body axes) is given by
CN
B
=
−
−
−
1 0 0
0
0
0
0 1 0
0
0
0
0 0 1
cos sin
sin cos
cos sin
sin cos
cos sin
sin cosφ φ
φ φ
θ θ
θ θ
ψ ψ
ψ ψ (5)
where
ψ= azimuth
θ = pitch
φ = roll.
This matrix is to be considered as a coordinate transformation rather than a rotation of a vector.
That is, when the matrix multiplies the components of a vector expressed in the NED frame, it
produces the components of the same vector, expressed in the body frame. The inverse of this
matrix, or the transformation from the body to the NED frame, is the transpose of this matrix.
Transformations and coordinate frames are further discussed in Appendix C.
12. NAWCWPNS TM 8128
8
Since we now have three degrees of freedom, we will need an accelerometer on the z body axis
and gyros on the body x and y axes. We now measure both acceleration and angular rate about
each body axis.
Let us return to the case where we are at rest at some known point in the north-east plane. To
align the platform, we must determine three angles; pitch, roll, and heading. To be clear on this
point, alignment does not mean we physically rotate the platform, we merely determine its attitude
with respect to some reference frame. Fortunately, we now have the means of determining pitch
and roll. Let us assume our platform is rotated by some angle δθ about the y or pitch axis. The
situation appears as shown in Figure 4. The y axis is out of the plane of the paper.
N-E PLANE
x
z
g
D
δθ
→
FIGURE 4. Tilt Errors.
This tilt is observable. The reason is that we are now sitting in a gravitational field. The
accelerometer will measure a component of g in the z direction of
fz = -gcosδθ
and in the x-direction of
fx = gsinδθ
where g is the magnitude of the gravitational field. For small angles, the measured values are
fz = -g
fx = gδθ.
The output of the x accelerometer gives us a direct measurement of the tilt about the y axis.
Similarly, the output of the y accelerometer provides a measure of the tilt about the x axis. We still
have no observability on heading.
This discussion provides a simple example of correlation of errors. If the x accelerometer has a
bias error of δbx, the output of the accelerometer is
fx = δbx + gδθ
If we are trying to level the platform (i.e., determine its attitude, we do not actually rotate anything)
with the accelerometer output, we cannot tell the difference between accelerometer bias and tilt. A
13. NAWCWPNS TM 8128
9
filter will zero out this signal by apportioning the corrections to bias and tilt according to its a priori
estimate of their initial errors.
Note that even though we have residual tilt and bias errors, these errors, once correlated, will
produce no velocity errors. This will be true as long as we remain at rest or proceed without
turning. As soon as we turn, the errors start to become decorrelated and no longer cancel, which is
why it is desirable to rotate the platform before the alignment ends. A rotation of 180 degrees
would allow us to compute both the accelerometer bias and the platform tilt. If we align without
turning, then turn 180 degrees after the alignment ends, the bias and tilt errors will now add rather
than cancel. This is a worst case scenario.
To summarize, we can level our platform for a stationary alignment. This alignment can be
done very accurately as we know the reference velocity accurately (it is zero) and there is little
noise on the velocity measurement. For our flat nonrotating earth, we still cannot estimate heading
in a stationary alignment. As we shall see, this situation will change once we allow the earth to
rotate.
If we now consider a moving base alignment, we find that little new is added to the equation.
Since we can now initialize the system at any arbitrary attitude, the reference must supply us with
an estimate of our attitude as well as position and velocity. Alignment in this case means reducing
the errors in these quantities as much as possible. Gravity still excites the level axis errors and
allows us to estimate pitch and roll errors by comparing our velocity to a reference velocity. We
still have the same problem of the correlation of the tilt and accelerometer bias errors. We still
require maneuvering in order to estimate heading error.1
The main difference is in the accuracy of
our measurements. Our reference velocity is usually provided by another navigation system. The
accuracy of this reference system will limit the accuracy of our alignment. Another contributor to
measurement accuracy is measurement noise. The system under alignment may be connected to the
reference system by a nonrigid structure. The velocity measurement will thus be corrupted by
vibration and flexure. These effects, which are not measurable or predictable, appear as noise on
the measurement.
A navigation solution for our three-dimensional model is shown in block diagram form in
Figure 5. The inputs, fB
and ωB
, are the accelerometer and gyro outputs where
fB
= specific force in body coordinates
ωB
= angular rate in body coordinates.
These inputs will, in general, contain error terms. The angular rate is used to compute the NED to
body transformation, CN
B
. Pitch, roll, and heading can be found from the elements of this matrix.
The inverse of CN
B
is used to transform the specific force to the NED frame. The acceleration
due to gravity is added to this quantity to obtain acceleration in the NED frame (Equation (4)). This
acceleration is integrated twice to produce velocity, V, and position, R. Both V and R will be with
respect to the NED frame. The navigation outputs may then be compared to some reference
quantity to compute measurements to be processed by a Kalman filter (or equivalent). The filter
may estimate corrections to both the navigation solution and some of the sensor errors, depending
on the filter model.
1
Our treatment of heading error as different from pitch and roll error is somewhat arbitrary for the three-dimensional
case. Resolution of the gravity vector will depend on the azimuth angle error for large pitch angles, thus making it
observable. The reason these errors are treated differently is that, in practice, we almost always align at a nominally
level attitude.
14. NAWCWPNS TM 8128
10
→ →
→
→
→
V
R
RV
C
B
N
(o)
C
B
N
→
f
B
→
f
N
X
o o
CB
N
+
+
ω B
→
g
N
FIGURE 5. Flat Earth Navigator.
Numerically, the only part of this process that presents any degree of difficulty is the
integration of ωB
to obtain CN
B
. This is because to body may rotate at high rates. Since integration
errors tend to accumulate, this algorithm must be chosen carefully to ensure numerical accuracy.
We mention in passing two effects that affect sensor accuracy, coning in gyros and sculling in
accelerometers. These are due to rotational effects in strapdown systems. We will not go into the
details here. Coning and sculling are usually compensated at high frequencies in the sensor
electronics.
NONROTATING SPHERICAL EARTH
We now come to our spherical earth model. We will assume, for the moment, that the earth is
not rotating. We can see that our NED coordinate system is no longer appropriate for indicating
position with respect to the earth’s surface. We now describe our platform’s position in terms of
latitude, longitude, and altitude (Φ,λ,h) above the earth’s surface (Figure 6). We still use the NED
N
E
LATITUDE
EAST
NORTH POLE
GREENWICH
MERIDIAN
WEST
EQUATOR
LONGITUDE
D
FIGURE 6. Spherical Earth.
15. NAWCWPNS TM 8128
11
frame, but now it represents a frame that is tangent to the surface of the earth at the platform’s
present position. This frame is referred to as a locally level frame. Note that the NED frame now
moves about the surface of the earth along with the platform. We also define a coordinate frame,
which is fixed with its origin at the center of the earth, as the z axis through the North Pole, the x
axis through the Greenwich Meridian, and the y axis to complete a right-handed coordinate frame.
This is the earth-centered, earth-fixed (ECEF) frame. The transformation from the ECEF frame to
the NED frame is defined by the latitude and longitude. The attitude of the platform is described
with respect to the NED frame exactly as before.
Gravitational acceleration can no longer be considered as a constant. Gravitational acceleration
decreases inversely as the distance from the center of the earth, hence g is a function of altitude.
This leads to an instability in the navigation solution in the vertical direction. A positive altitude
error causes us to compute a weaker value for the gravitational acceleration than is actually present.
Hence we will not correctly compensate for the gravitational field in our navigation equations.
Since this residual component of gravity appears as an upward acceleration (Equation (4)), it will
cause a velocity error in the upward direction that will, in turn, increase the altitude error. This
increase is positive feedback, the vertical solution will diverge at an exponential rate. Near-earth
navigators require some external means of stabilizing the vertical channel if they are navigating for
more than a few minutes. This stabilization can be any external measure of altitude such as
barometric pressure or the output of an altimeter. We will not include stabilization in the rest of our
analysis.
Figure 7 shows a block diagram for a navigation solution for a nonrotating spherical earth
model. This model is only slightly more complicated than the flat earth model. Our inertial frame is
no longer the NED frame, it is the ECEF frame. The gyros and accelerometers now measure
angular rate and acceleration of the body with respect to the ECEF frame. First, we integrate the
angular rates to obtain the ECEF to body transformation, CE
B
. The inverse of this transformation is
then used to rotate the specific force to the ECEF frame. As before, we add gravity to this quantity
to obtain the acceleration of the body in the ECEF frame. Note the feedback path for the gravity
computation. The acceleration is integrated twice to obtain velocity and position in the ECEF
frame. These quantities are useful when the inertial navigator is being used in conjunction with a
GPS set, as GPS uses ECEF coordinates. However, to provide a useful interface for a human
CB
E
→
f
B
→
f
E
X
CB
N
+
+
→
ω B
→
g
E
X
GRAVITY
ROUTINE
→
a
E →
V
E
X
CONVERSION
ROUTINE
C N
E
→
V
N
Φ,λ,h
→
RE
FIGURE 7. Nonrotating Spherical Earth Navigator.
16. NAWCWPNS TM 8128
12
operator, we convert position to latitude, longitude, and altitude (Φ,λ,h), and rotate the velocity to
the NED frame. The ECEF-to-NED transformation matrix, CE
N
, is defined by the position. We
use this matrix along with CE
B
to compute the attitude matrix, CN
B
. We have omitted showing
initialization for the integrators in Figure 7 to keep it from becoming too busy.
The only new feature we have encountered when going from a flat earth to a nonrotating
spherical earth, other than dealing with spherical coordinates, is the positive feedback in the gravity
computation. We still align the platform to the NED frame as before. Previous remarks on
alignment still apply. Minor complications will occur because the NED frame is no longer an
inertial frame, but, in principle, everything works as before.
One difference between a flat earth model and a spherical earth model should be noted, the
phenomenon of Schuler oscillation. If we initialize our system with some initial system error, the
position, velocity, and attitude errors will be bounded (assuming perfect sensors). These errors
will oscillate with a period of approximately 84.4 minutes. In a flat earth model, these errors would
be unbounded. Any sensor errors will behave the same in both cases, they will produce system
errors that are powers of time. A derivation of Schuler oscillation is given in Appendix D.
ROTATING SPHERICAL EARTH
The next step in our analysis is to allow our spherical earth to rotate. We define a new
coordinate frame called the inertial frame, which is fixed at the center of the earth. Ignoring the
earth’s orbital motion, we regard the orientation of this frame as fixed with respect to the distant
stars. The inertial frame is defined to be coincident with the ECEF frame at zero time. Figure 8
shows the relationship between the inertial and ECEF frames. The ECEF frame rotates with respect
to the inertial frame with an angular velocity (Ω) of approximately 15.04 degrees per hour.
FIGURE 8. Rotating Spherical Earth.
17. NAWCWPNS TM 8128
13
The resulting navigation system is shown in Figure 9. The gyros and accelerometers now
measure angular rate and linear acceleration of the body with respect to the inertial frame. We
integrate the angular rate to obtain the inertial-to-body transformation, CI
B
. The inverse of this
transformation is used to rotate the specific force to inertial coordinates. We again add gravity to
obtain the acceleration of the body with respect to inertial space. The integral of this term is the time
rate of change of the position vector with respect to inertial space. (This is not a navigation
velocity. We define our navigation velocities with respect to a coordinate frame fixed in the earth
for near-earth navigators). We correct this term for rotational effects to obtain ECEF velocity, as
shown in Figure 9. The second integral of the inertial acceleration gives us the position vector in
inertial coordinates. We rotate this position vector with the inertial-to-ECEF transformation, CI
E
,
which we obtained from integrating the earth rate. The rest of the navigation solution goes much as
before, the ECEF position vector is converted to latitude, longitude, and altitude. These are used to
compute the direction cosine matrix (DCM) from the ECEF frame to the NED frame, CE
N
. Then
CI
B
, CI
E
, and CE
N
are combined to form the NED-to-body transformation, CN
B
.
X +
+
X CONVERSION
ROUTINE
f
→B
f
→I
R
→E Φ,λ,hR
→I
R
→Ia
→I
XR
→I
R
→I
Ω
→I– x X
X
GRAVITY
MODEL
DCM
XX
ω
→ I C
B
I C
B
E
C
B
N
C
N
E
g
→E
Ω
→I
C
E
I
V
→E
g
→I
.
. V
→ N
FIGURE 9. Rotating Spherical Earth Navigator.
One new consequence of the earth’s rotation other than increased mathematical complexity
exists. If we now perform a stationary alignment, we can estimate heading as well as tilts because
the earth’s rotation rate sensed by the gyros, ωB
, will be resolved in the wrong axes due to heading
error. Imagine a gimbaled platform that is leveled but has some heading error. Since, in a gimbaled
platform mechanization, the platform is torqued to keep it locally level, the platform must be
torqued to compensate for the earth’s rotation. This torque is applied about the north and vertical
axes. The north component is ΩcosΦ, where Ω is the angular rotation rate of the earth and Φ is
latitude. A small heading error of δψ will cause a tilt about the north-axis of (δψΩcosΦ) multiplied
by time. This tilt will result in an east velocity error, which can be used to estimate heading error.
Since the error signal depends on cosΦ, we see that the nearer we get to the poles, the less
effective this technique is. This technique is known as gyrocompassing, which can produce very
accurate alignments at latitudes below 70 degrees.
Neither of the navigation mechanizations shown in Figures 7 and 9 perform any of their
navigation functions in spherical coordinates. ECEF and inertial coordinates are well defined
everywhere. Quantities, such as longitude, north, and east, are meaningless at the poles, but these
are only outputs. Both systems will continue to navigate across the poles. Some of the outputs will
18. NAWCWPNS TM 8128
14
become undefined, but once the pole is crossed, the outputs will recover. Transfer alignments that
depend on locally level velocities for their measurements also become increasingly inaccurate as the
poles are approached. For these reasons, and the fact that the ECEF frame is the natural coordinate
frame for GPS, most navigation systems today are mechanized in the ECEF frame. Quantities,
such as latitude, longitude, and north and east velocities, are provided only as a human interface.
WGS84
The earth is actually approximated by an oblate spheroid rather than a sphere. We will not go
into the details here, as going to the more accurate model merely increases the mathematical
complexity of our solution, it does not provide any new insights. The current standard model for
the reference ellipsoid is the WGS-84 system defined by the Defense Mapping Agency (see
bibliography).
This completes our introduction to inertial navigation. Some special topics are addressed in
more detail in the appendixes. In the bibliography we list additional references that cover some of
this material in more depth. We have only touched on the alignment process. Alignment and
calibration are really part of the measurement process, and will be examined in more detail in the
appendix on Kalman filtering (Appendix A).
CONCLUSION
We hope the uninitiated reader has learned the basic simplicity of the navigation process
without being overwhelmed with the mathematical details, and has developed an appreciation of the
basic principles of alignment and navigation of an inertial system.
19. NAWCWPNS TM 8128
15
Appendix A
KALMAN FILTER
The Kalman filter is a recursive algorithm designed to compute corrections to a system based
on external measurements. The corrections are weighted according to the filter’s current estimate of
the system error statistics. The derivation of the filter equations requires some knowledge of linear
algebra and stochastic processes. The filter equations can be cumbersome from an algebraic point
of view. Fortunately, the operation of the filter can be understood in fairly simple terms. All that is
required is an understanding of various common statistical measures.
We begin with the concept of a state vector, which is central to the formulation of the filter
algorithms. A state vector is a set of quantities we have chosen to describe the “state” of a system.
For a navigation system, the state is naturally described by position, velocity, attitude, sensor
errors, and related quantities. Since we are trying to correct the errors in a navigation system,
working in terms of error states is convenient. Any quantity can be described in terms of its true
value plus some unknown error. For some arbitrary state x, our estimate of will be
ˆx x x= + δ
where x is the true value and δx is the error in our estimate. The purpose of the filter is to estimate
the error states (δx) and use them to correct our state estimates, ˆx.
We need to assume that all system equations and measurement equations can be expanded in a
Taylor series in the error states. Since these error states are assumed to be small, we keep only first
order terms. Therefore, our equations are linear. This formulation, where we are filtering the error
states instead of the states themselves, is referred to as an extended Kalman filter.
To illustrate the basic principles of the filter, we will begin by making the simplest assumptions
we can. We will assume that the mean of all error states is zero. (If they were not, we could make
them so by redefining the states.) We also will assume a one-state system, our error state will be
δx. We will have some a priori estimate of the initial standard deviation (σ) of this error state. To
keep the notation consistent with the literature, we will call the variance (σ2
) of the error state P.
The dynamics of our error state will, in general, be described by some differential equation. To
begin, we assume that this equation takes the simple form
δ δ˙x t F x t( ) = ( ) (A1)
where F is some constant. This equation has the formal solution
δx(t+∆t) = (expF∆t)δx(t) = Φ(∆t) δx(t), (A2)
where we have defined the state transition matrix (STM) as Φ(∆t) = (expF∆t). The STM describes
the evolution of the system error states in time, when no measurements are being processed.
20. NAWCWPNS TM 8128
16
By definition, the variance of our error state is P = E[δx2
] ,where the operation E[ ] denotes an
ensemble average. By assumption, E[δx] = 0. Applying this operator to Equation (A2), we obtain
P(t+∆t) = Φ(∆t) P(t) Φ(∆t). (A3)
We have written the equation in this form to correspond to the form it will take when we consider
state vectors of more than dimension.
We now consider the effect of a measurement on the system. Assume that we have an external
measurement of x that is corrupted by noise.
˜ ,x x= + η (A4)
where ˜x is the measured value of x and η is a zero-mean Gaussian white sequence of variance R.
We define a measurement residual as
z x x x= − = +ˆ ˜ δ η. (A5)
If it were not for the noise term, we would correct our state as
ˆ ˆx x z= − . (A6)
Since this measurement is corrupted by noise, we need to know how to weight the
measurement in correcting our state variable. This problem of computing optimal gains is the heart
of the Kalman filter. We wish to compute the Kalman gains, K, in the equation
ˆ ˆx x Kz= − (A7)
where Kz is our best estimate of the error state.
Since R is the variance of the measurement error and P is the variance of our current estimate of
δx, we logically expect the gains to be a function of these two variables. We state the result for one
dimension,
K = P/(P + R). (A8)
This result makes sense. If our measurement is more accurate than our system error, then R<<P
and K ≅ 1. In this case, Equation (A7) is approximately the same as (A6), as it should be. If we
know our system error more accurately than our measurement, we should make very little
correction due to the measurement. In this case R>>P and K ≅ P/R so the correction is very small.
We can see that the Kalman gains make a great deal of sense. Of course, this is not a derivation of
the gain equations, but we are not concerned with mathematical rigor here.
Once we make a measurement, we must change the variance of P to reflect this new
information. The result for one dimension is
P = PR/(R + P). (A9)
In the case where R<<P, this reduces to P ≅ R(1-R/P). The error state variance thus becomes
approximately equal to the measurement error variance, as we would expect. In the case where
R>>P, Equation (A9) reduces to P ≅ P(1-P/R). The error state variance changes very little in the
case, again as we would expect.
21. NAWCWPNS TM 8128
17
Since x is in general an n-dimensional vector, P will be an nxn matrix. The above equations
will become correspondingly more complicated, but the principle will remain the same. Before we
consider the complications introduced by higher dimensional spaces, we summarize the process
described. The sequence of events is as follows:
1. The variance of δx is initialized as P0.
2. The variance is propagated forward in time to the first measurement as
P1 = Φ(∆t) P0 Φ(∆t).
3. The gain is computed based on this value of P and the measurement error variance,
K1 = P1/(P1 + R).
4. The measurement residual is computed as z x x= −ˆ ˜ .
5. The state vector is corrected according to ˆ ˆx x K z= − 1 .
6. The variance of δx is updated to reflect the measurement as
P′1 = P1R/(R + P1).
7. This variance is propagated forward to the next measurement as
P2 = Φ(∆t) P′1Φ(∆t)
and the process begins all over.
Note that we have used P′k to denote the value of P immediately after the kth measurement and Pk
to denote its value immediately before.
In this description, we have neglected the effect of process noise (or plant noise). This effect
takes the form of a random forcing function in Equation (A1),
δ ˙x(t) = Fδx(t) + ν (A1)′
where ν is a zero-mean Gaussian white noise of power spectral density N. This will add a term to
Equation (A2) of
δx(t+∆t) = (expF∆t)δx(t) + expF t
t
t t
τ −( )[ ]
+
∫
∆
νdτ
= Φ(∆t) δx(t) + Φ
∆
τ −( )[ ]
+
∫ t
t
t t
νdτ (A2)′
and in Equation (A3) of
P(t+∆t) = Φ(∆t) P(t) Φ(∆t) + Q (A3)′
where
Q = N exp2F t
t
t t
τ −( )[ ]
+
∫
∆
dτ ≅ N∆t for 2F∆t <<1.
All other equations remain the same.
22. NAWCWPNS TM 8128
18
As we mentioned, δx is, in general, a vector defined as
δx =
δ
δ
δ
x
x
xn
1
2
M
(A10)
P now becomes an nxn matrix defined by
P = E[δxδxT
] (A11)
where δxT
is the transpose of Equation (A10). This is referred to as the covariance matrix. The
diagonal elements will be the variances of the individual error states. The off-diagonal elements
will be a measure of the correlations between the corresponding diagonal elements. Correlations
are important as they permit the indirect estimate of a state by measuring a correlated state.
Correlations may arise through the STM or through measurements.
The measurement Equation (A5) now appears as
z = H δx + η. (A5)′
H is the measurement (or observation) matrix. The vector form of the measurement assumes that
more that one measurement is taken. This equation also assumes that the measurements consist of
quantities that can be expanded in terms of the components of the error state vector. We made this
linearization assumption at the beginning. Equation (A3) generalizes to
P(t+∆t) = Φ(∆t) P(t) Φ(∆t)T
+ Q (A3)′′
where Q is a matrix generalization of the process noise term. The STM will not, in general, be a
simple exponential (the F matrix in Equation (A1) will not be a constant). Computation of this
matrix will usually require some form of numerical integration.
We will merely state the form Equations (A8) and (A9) take. For optimal gains, these equations
become
K = PHT
(HPHT
+ R)-1
(A8)′
and
P′ = [I - KH]P. (A9)′
In the above, R is now a matrix of the noise associated with each element of the measurement
vector. For more detail, the reader is referred to one of the many texts on Kalman filtering.
Kalman filters are useful when processing measurements in the presence of measurement noise
and system noise.Kalman filters provide a method of incorporating the measurements in an optimal
fashion. Measurements are required in inertial navigation systems to align the system and bound
the system errors over time. These measurements may take many forms: position measurements,
Doppler velocity measurements, GPS measurements, or any external system that gives us a useful
measure of one or more of the navigation parameters.
23. NAWCWPNS TM 8128
19
Alignment may be self-alignment such as the gyrocompassing technique discussed in the text
or alignment to some external source. When the external source is another inertial system, the
process is usually referred to as transfer alignment. The measurements used in the transfer
alignment process can be based on the accelerometer output or the gyro output. Examples of
accelerometer-based alignments are velocity matching and velocity integral matching. Examples of
gyro-based alignments are angular rate matching and attitude matching (a common attitude-
matching application is a SINS alignment where the platform is aligned to a ship’s inertial
navigation system). GPS position and velocity measurements are used to align a system and to
bound the system errors.
Alignment often includes calibration of some of the sensor errors. This calibration is especially
useful for sensor errors that can be modeled as bias terms. All that is required is to include these
sensor error terms as part of the error state vector. For a more detailed treatment of the
measurement process, the reader is referred to the bibliography.
25. NAWCWPNS TM 8128
21
Appendix B
NOISE AND SENSOR ERRORS
Most of the error sources that corrupt our navigation solution are sensor errors or random
disturbances. In this appendix, we describe some of the more common gyro and accelerometer
errors. The error model for any sensor will, to some extent, depend on its construction. A ring
laser gyro will behave somewhat differently than a tuned rotor gyro. The errors discussed are
common to all types of gyros and accelerometers. We also discuss some of the basic types of noise
models.
Bias Errors—The simplest type of error. A bias is, by definition, a constant signal on the output
of a sensor, independent of the input. A bias will not change during a run, but may vary from turn-
on to turn-on. A bias is modeled as a random constant. For navigation grade sensors, bias is
usually specified in deg/hr or milligees.
Scale Factor Errors—A linear error that is proportional to the input signal. Scale factor is
usually specified in parts per million.
Misalignment—Refers to mechanical misalignment. Ideally, the gyros and accelerometers define
an orthogonal triad. This idealized coordinate frame is the platform frame. Since achieving perfect
mechanical alignment in a practical system is impossible, we describe the alignment error of each
sensor to the platform axes as a random constant. This requires six numbers for the gyros and six
for the accelerometers. For example, one number describes the misalignment of the x gyro from
the x platform axis in the y direction and another the misalignment in the z direction. This error is
sometimes referred to as nonorthogonality error and is given in terms of microradians.
These errors are idealizations. For example, bias errors are only constant for short terms, they
typically exhibit drift that might be modeled as a Markov process (see below) superimposed on a
constant bias. Scale factor errors usually exhibit some degree of non-linearity. Also some error
terms are dependent on the stresses on the sensor. These error terms are due to mechanical
deformations of the sensor.
Another important error source is due to temperature. Navigation grade sensors model the
effects of temperature and compensate for this in their internal electronics. This compensation is
never perfect. The most important residual errors over temperature are a temperature dependence of
bias and scale factor errors, making calibration of the sensor difficult.
Another error source is quantization error. The output of a gyro or accelerometer is given in
terms of some smallest unit, we do not have infinite precision. This produces a white noise on the
output proportional to the magnitude of the quantization.
Many error sources are random and can only be described in terms of stochastic processes. A
stochastic process is a random time process. Many of these processes may be described by
differential equations with white noise forcing functions. A white noise signal is a mathematical
idealization that contains infinite energy and bandwidth. This signal is described in terms of its
power spectral density (PSD). Mathematically, if η is a zero-mean white process with PSD N, then
26. NAWCWPNS TM 8128
22
E[η] = 0
and
E[η(t)η(t+τ)] = Nδ(τ)
where δ(τ) is a delta function.
The most important noise associated with a gyro is random walk. Random walk results from
the integration of white noise. Random walk can be described by the differential equation
˙x= η
where η is a white noise term with PSD N. The variance of x will be
E[x2
] = Nt.
This time-dependent error restricts the accuracy of an alignment and cannot be estimated or
compensated for. Accelerometers also exhibit random walk, but the effect on a navigation system
is usually minor.
Some noise sources (system disturbances) are correlated in time. Their current value depends
to some extent on one or more past values. The most common processes of this type are Markov
processes. First and second order Markov processes are described by the differential equations
˙x + βx = η (first order)
and
˙˙x+ 2αβ ˙x +β2
x = η. (second order)
Wing flexure in a aircraft, for example, is commonly modeled as a second order Markov process.
This effect is important effect in transfer alignment. Disturbances of this type may be modeled in
the filter by state vector augmentation. For a more detailed discussion of stochastic processes, the
reader is referred to the bibliography.
The degree of accuracy with which we must model all these error sources is dependent on the
application as well as the quality of the sensor.
27. NAWCWPNS TM 8128
23
Appendix C
COORDINATE TRANSFORMATIONS
In the most general sense, a coordinate transformation over a real three-dimensional space is a
set of nine real numbers that transforms the components of a vector in one frame to another.
Suppose we have two coordinate frames that we call A and B. We write an arbitrary vector,
r
X, in
the A frame as
r
X
X
X
X
A
=
1
2
3
where the subscripts 1, 2, and 3 refer to the coordinate axes (e.g., xyz, NED). We denote the same
vector with coordinates in the B frame as
r
X
X
X
X
B
=
′
′
′
1
2
3
where we have used the primes to distinguish the components of the vector in the B frame from the
components in the A frame.
A coordinate transformation relates the components in one frame to the components in
another. Thus, if CA
B
is the transformation from A to B,
r r
X C XB
A
B A
= (C1)
similarly
r r
X C XA
B
A B
= (C2)
where CB
A
is the inverse of CA
B
.
Equation (C1) can be written in vector matrix notation as
′
′
′
=
X
X
X
C C C
C C C
C C C
X
X
X
1
2
3
11 12 13
21 22 23
31 32 33
1
2
3
(C3)
28. NAWCWPNS TM 8128
24
with a similar expression for Equation (C2). This equation can be written in a compact form as
′ = ⋅ =( )X C X jk kj j 1 2 3, , (C4)
In the matrix elements, Ckj, k is the row index and j is the column index. According to the
summation convention, repeated indices are summed from 1 to 3 unless otherwise indicated.
Orthogonality conditions on the transformation reduce the number of independent elements in
the transformation matrix from nine to three. This means that we may write the matrix elements as
functions of as few as three parameters. There are many ways to do this.
One of the most common ways of parameterizing the transformation matrix is by the use of
Euler angles. There are many conventions for specifying Euler angles, but for aircraft navigation
systems, it is almost always done as follows:
CN
B
=
−
−
−
1 0 0
0
0
0
0 1 0
0
0
0
0 0 1
cos sin
sin cos
cos sin
sin cos
cos sin
sin cosφ φ
φ φ
θ θ
θ θ
ψ ψ
ψ ψ (C5)
This is the transformation from NED to body axes, a product of three rotations in the following
sequence:
1. A positive rotation about the down (z) axis by an angle ψ
2. A positive rotation about the new y axis by an angle θ
3. A positive rotation about the resulting x axis by an angle φ
These angles define leading ( ψ ), pitch (θ), and roll (φ). Note that order is important, rotations
about different axes do not commute. For example,
cos sin
sin cos
cos sin
sin cos
cos sin
sin cos
cos sin
sin cos
θ θ
θ θ
ψ ψ
ψ ψ
ψ ψ
ψ ψ
θ θ
θ θ
0
0 1 0
0
0
0
0 0 1
0
0
0 0 1
0
0 1 0
0
−
−
≠ −
−
Since rotations are orthogonal, the inverse of Equation (C5) is just its transpose or
CB
N
=
−
−
−
cos sin
sin cos
cos sin
sin cos
cos sin
sin cos
ψ ψ
ψ ψ
φ φ
φ φ
φ φ
φ φ
0
0
0 0 1
0
0 1 0
0
1 0 0
0
0
(C6)
This is easily verified by direct multiplication of Equations (C5) and (C6) (in either order). In this
discussion, we have illustrated Euler angles with a particular example, but any rotation can be
paramaterized by three rotations in any sequence.
The rotation from one frame to another can be accomplished by a sequence of three rotations
in an infinite number of ways. One particularly useful way of describing a transformation by a
single rotation is given by
29. NAWCWPNS TM 8128
25
C I x xA
B
= ( ) + ( ) + ( ) −( )r r
θ
θ
θ
θ
θ
θ
sin cos2
2
1
(C7)
In this equation (I) is the 3 x 3 unit matrix
I( ) =
1 0 0
0 1 0
0 0 1
and, by definition
r
θ
θ θ
θ θ
θ θ
x
z y
z x
y x
( )=
−
−
−
0
0
0
The angle θ is the magnitude
θ θ θ θ= + +[ ]x y z
2 2 2 1 2/
Note that this form of the transformation still requires three parameters; θx, θy, and θz. This
represents a rotation by the angle θ about an axis defined by the unit vector
r
n
x
y
z
=
1
θ
θ
θ
θ
The components of
r
θ are the same in both frames.
Transformations are not always parameterized with three parameters. A common method of
describing rotations is by the use of quaternions. A quaternion has four components so there is
some redundancy. A quaternion is a set of four quantities, q0, q1, q2, and q3. They may be
thought of as a combination of a vector and a scalar that obey the following multiplication rules:
For two quaternions (q q0 +
r
) and ( p p0 +
r
), their product is
p p q q p q p q p q q p p q0 0 0 0 0 0+( ) +( ) = − ⋅ + + + ×
r r r r r r r r
(C8)
The details of quaternion algebra are beyond the scope of this document. We mention them because
they are commonly encountered in navigation algorithms. The quaternion may be used to construct
a 3 x 3 rotation matrix as
R
q q q q q q q q q q q q
q q q q q q q q q q q q
q q q q q q q q q q q
( ) =
+ − − +( ) −( )
−( ) − + − +( )
+( ) −( ) − −
0
2
1
2
2
2
3
2
1 2 0 3 1 3 0 2
1 2 0 3 0
2
1
2
2
2
3
2
2 3 0 1
1 3 0 2 2 3 0 1 0
2
1
2
2
2
2 2
2 2
2 2 ++
q3
2
(C9)
30. NAWCWPNS TM 8128
26
The reason quaternions are widely used is because of their integration properties. The derivative of
a quaternion is a linear differential equation. For example:
let
ωi = components of the angular velocity of the body frame with respect to inertial space, in
body coordinates
Ωi = components of the angular velocity of the NED frame with respect to inertial space, in
NED coordinates
The quaternion that represents the rotation from NED to body obeys the differential equation
˙
˙
˙
˙
/
q
q
q
q
0
1
2
3
1 1 2 2 3 3
1 1 3 3 2 2
2 2 3 3 1 1
3 3 2 2 1
1 2
0
0
0
=
− −( ) − −( ) − −( )
−( ) +( ) − +( )
−( ) − +( ) +( )
−( ) +( ) − +
ω ω ω
ω ω ω
ω ω ω
ω ω ω
Ω Ω Ω
Ω Ω Ω
Ω Ω Ω
Ω Ω ΩΩ1
0
1
2
30( )
q
q
q
q
(C10)
This equation is linear in the q’s. The corresponding equation for Euler angles, for example, is
highly nonlinear.
Another common way of keeping track of a rotation matrix is to use the elements of the
direction cosine directly. Since the rows and columns of this matrix are orthogonal, any two rows
(or columns) determine the other row (or column). The differential equations for these matrix
elements (the Ckj) are again linear. The usual Euler angles can be recovered from the matrix
elements by use of inverse trigonometric relations.
We give a very useful relation involving the derivative of a transformation matrix,
˙C C CA
B
A
B
AB
B
AB
B
A
B
= =Ω Ω (C11)
In this equation, CA
B
is a 3 x 3 orthogonal tranformation and
ΩAB
A
≡
−
−
−
0
0
0
3 2
3 1
2 1
ω ω
ω ω
ω ω
(C12)
In this antisymmetric form, the ωk are the components of the angular velocity of the B frame with
respect to the A frame, with coordinates in the A frame. The matrix ΩAB
B
is the same, except that
the coordinates of the angular velocity are in the B frame. This relation is, in fact, the basis for the
direction cosine method of integrating the transformation matrix discussed in the previous
paragraph.
As an example of how Equation (C11) is used, let us take the derivative of Equation (C1) with
respect to time. The result is
r r r˙ ˙ ˙x C x C xB
A
B A
A
B A
= + (C13)
31. NAWCWPNS TM 8128
27
In this equation,
r˙xB
is the time derivative of the vector with respect to the B frame, while
r˙xA
is the
derivative with respect to the A frame. Using Equation (C11), this equation becomes
r r r˙ ˙x C x xB
A
B A
AB
A A
= +( )Ω
By matrix multiplication, we verify that
ΩAB
A A
AB
A A
x x
r r r
= − ×ω
Substituting into the previous equation, we have
r r r r˙ ˙x C x xB
A
B A
AB
A A
= − ×( )ω , (C14)
which we recognize as the Coriolis law that relates derivatives with respect to different coordinate
frames.
Finally, we write down the relations between the derivatives of the Euler angles and the
angular velocity of one frame with respect to another. To be definite, we refer to Equation (C5).
Let
r
ω be the angular rate of the body with respect to the NED frame, expressed in the NED frame.
Using Equation (C11) we find
ω φ θ ψ θ ψ
ω θ ψ φ θ ψ
ω ψ φ θ
N
E
D
= −
= +
= −
˙ cos cos ˙ sin
˙ cos ˙ cos sin
˙ ˙ sin
(C15)
Similarly, we can relate the components of
r
ω in the body frame to the Euler angles as
ω φ ψ θ
ω θ φ ψ φ θ
ω ψ φ θ θ φ
x
y
z
= −
= +
= −
˙ ˙ sin
˙ cos ˙ sin cos
˙ cos cos ˙ sin
(C16)
As we can see from the material presented in this appendix, the subject of coordinate
transformation is far from trivial. We have only scratched the surface here. However, we have
given the most important results that apply to inertial navigation.
33. NAWCWPNS TM 8128
29
Appendix D
SCHULER OSCILLATION
In this appendix we present a simple derivation of Schuler oscillation. For simplicity, we
assume a platform located at the equator of a spherical nonrotating earth. Also assume no initial
position error and the x axis of the platform is pointing north. A tilt about the north (x) axis results
in an acceleration error in the east (y) direction of
δay = -gθx
where g is the gravitational acceleration. The velocity error due to this tilt is then given by
δ δ θ τV a dt g dy y x
tt
= = − ∫∫ 00
.
This results in an east position error of
δ δ τy v dy
t
= ∫0
.
The navigation algorithm will rotate its notion of local level according to its sensed velocity. This
corresponds to a tilt error rate of
˙˙ / /θ δ θ τx y x
t
v R g R d= = − ∫0
where R is the radius of the earth. Differentiating this equation with respect to time, we have
˙˙θx + (g/R) θx = 0.
We recognize this as the differential equation for a simple harmonic oscillator. The frequency of
oscillation is
ω = g R/ .
The period of oscillation is
T = 2π/ω .
For g = 32.17 ft/s and R=20925646 ft, we have
T = 84.46 min,
34. NAWCWPNS TM 8128
30
which is the period of Schuler oscillation. The equation for the frequency of oscillation is formally
identical to that for a simple pendulum of length R, which is why we sometimes see the statement
that an inertial navigation system acts like a simple pendulum with a length equal to the earth’s
radius. By taking the appropriate derivatives, it is easy to see that both position and velocity in the
y direction obey the same differential equation and, therefore, oscillate with the same frequency.
Since, in a practical system, we have sensor errors, we normally see the Schuler oscillation
superimposed on an error curve due to sensor errors. In general, we observe Schuler oscillations
about both level axes.
35. NAWCWPNS TM 8128
31
Appendix E
NED ROTATION RATES
The orientation of the NED frame is constantly changing with respect to the inertial frame
because of the rotation of the earth and the motion of the navigator over the earth’s surface. In this
appendix we give the expressions for the angular velocity of the earth fixed frame with respect to
inertial space and the angular velocity of the NED frame with respect to the earth fixed frame.
Differentiating the ECEF to NED transformation
CE
N
=
−
−
−
0 0 1
0 1 0
1 0 0
0
0 1 0
0
0
0
0 0 1
cos sin
sin cos
cos sin
sin cos
φ φ
φ φ
λ λ
λ λ (E1)
we obtain
˙
˙ ˙ cos
˙ ˙ sin
˙ cos ˙ sin
˙ sin ˙
sin ˙ cos
˙ ˙ cos
C C
C
E
N
E
N
E
N
= −
− −
=
−
− −
0
0
0
0
0
0
λ φ λ
λ φ λ
φ λ φ λ
λ φ φ
λ φ λ φ
φ λ φ
(E2)
where
φ = latitude
λ = longitude
From equation (C11) we find the angular rate of the NED frame with respect to the ECEF
frame in both NED and ECEF coordinates,
r
ω
φ λ
φ λ
λ
EN
E
= −
˙ sin
˙ cos
˙
(E3)
and
36. NAWCWPNS TM 8128
32
r
ω
λ φ
φ
λ φ
EN
N
= −
−
˙ cos
˙
˙ sin
These rates are often referred to as transport rates.
The longitude and latitude rates are
˙
cos
λ
φ
=
+( )
V
R h
E
p
(E4)
and
˙φ =
+
V
R h
N
m
where
VE = east velocity
VN = north velocity
h = altitude above reference ellipsoid
We have defined the radii of curvature of the reference ellipsoid as
R
a
p =
−( )1 2 2 1 2
ε φsin
/ (E5)
and
R
a
m =
−( )
−( )
1
1
2
2 2 3 2
ε
ε φsin
/
where
a = earth’s semimajor axis = 20,925,646 ft
ε2 = square of the earth’s eccentricity = 0.00669438
Combining Equations (E4) and (E3), we have the alternate form of the transport rates in NED
as
r
ω
φ
EN
N
E p
N m
E
p
V R h
V R h
V
R h
=
+
− +
−
+
/
/
tan
(E6)
37. NAWCWPNS TM 8128
33
The earth rate term is found from
r
ω
φ
φ
IE
N
E
N
E
E
E
C=
=
−
0
0 0
Ω
Ω
Ω
cos
sin
(E7)
where
ΩE = angular rotation rate of the earth with respect to inertial space
= 15.041067 deg/hr
= 7.292115 x 10-5 rad/s
These expressions have been included for reference only. The reader is referred to the
bibliography for a detailed description of the earth’s reference ellipsoid.
38. NAWCWPNS TM 8128
34
BIBLIOGRAPHY
Navigation and Alignment
Britting, Kenneth R. Inertial Navigation Systems Analysis. Wiley-Interscience, 1971.
Farrell, James L. Farrell. Integrated Aircraft Navigation. Academic Press, 1976.
Lawrence, Anthony. Modern Inertial Technology. Springer-Verlag, 1993.
Naval Air Warfare Center Weapons Division. Transfer Alignment, by Sherryl H. Stovall. China
Lake, Calif., NAWCWPNS, 1996. (NAWCWPNS TM 8069.)
Siouris, George M. Siouris. Aerospace Avionics Systems. Academic Press, 1993.
Geodesy
Defense Mapping Agency. Department of Defense World Geodetic System. DMA, 1984. (DMA
Technical Report 8350.2.)
Defense Mapping Agency. Geodesy For The Layman. DMA, 1983. (DMA Technical Report 80-
003.)
Kalman Filtering
Bierman, Gerald J. Factorization Methods for Discrete Sequential Estimation. Academic Press,
1977.
Brown, Robert Grover and Patrick Y.C. Hwang. Introduction to Random Signals and Applied
Kalman Filtering. John Wiley & Sons, 1992.
Technical Staff, The Analytic Sciences Corporation, Arthur Gelb, ed. Applied Optimal Estimation.
The M.I.T. Press, 1974.
Stochastic Processes
Papoulis, Athanasios. Probability, Random Variables, and Stochastic Processes. McGraw-Hill,
1984.