2. INTRODUCTION
• Imagine yourself walking down a street. The
street is filled with obstacles, like houses, trees,
and other people. You want to go to the
supermarket. With your eyes closed.
• Robots that move freely in the world have the
same problems as humans when walking
through the world.
• Humans, using eyes or other sensors, robots can
detect these imperfections and get a better idea
of where they are.
2
3. AUTONOMOUS MOBILE ROBOTS
• In 1920 Karel Capek introduced the word “Robot”
in the English language in his play R.U.R.,
Rossum's Universal Robots.
▫ Mobility
The mobility of robots is the degree to which robots are
able to freely move through the world.
▫ Autonomy
It depends on to what extent a robot relies on prior
knowledge or information from the environment to
achieve its tasks.
▫ Application Areas
3
4. CLASSES OF AUTONOMY
• Non-autonomous robots are completely remotely
steered by humans. The intelligence involved in these
robots consists of interpreting the commands received
from human controllers.
• Semi-autonomous robots can either navigate by
themselves, or be steered by humans. In dangerous
situations the robot takes full control; in less dangerous
situations humans can control the robot.
• Fully-autonomous robot vehicles are steered solely
by the robots themselves. Fully autonomous robot
vehicles are capable of intelligent motion and action,
without requiring any external guidance to follow or
control them
4
5. APPLICATION AREAS
• Autonomous robots can be used to explore
environments that are difficult for humans to
explore.
▫ Missions to planets, other places in space
▫ Or investigations of dangerous sites
Radioactive environments
▫ Perform repairs and maintenance remotely
▫ Entertainment sector
▫ Service sector
Intelligent wheel chairs and vacuum cleaners
Medicine or food delivery.
5
6. ROBOT NAVIGATION
• Robot navigation is the task of an autonomous robot
to move safely from one location to another.
• Three Question
▫ Where am I?
Robotic localization.
▫ Where am I going?
Goal recognition.
▫ How do I get there?
Path planning.
6
7. ISSUES OF NAVIGATION
• Computational power
▫ Real time image processing
▫ Computer vision
▫ Learning is high.
• Object and Landmark Recognition
▫ Robots need to recognize structures to perform their tasks. If these
structures are not known in advance, image processing may need a lot
more computational power than if the structures are known.
• Obstacle Avoidance
▫ At all costs robots should avoid colliding with the obstacles in their
environments.
• Multi-Modal Sensor Fusion
▫ The information that robots obtain using their sensors needs to be
combined to determine what is going on in the environment.
Uni-modal sensor fusion
Multi-modal sensor systems
7
8. KALMAN FILTERS
• The Kalman Filter (KF) is the best possible, optimal,
estimator for a large class of systems with uncertainty
and a very effective estimator for an even larger class.
▫ The KF is optimal
With respect to virtually any criterion that makes sense
for example the mean squared error.
It uses all available information that it gets.
It does not matter how accurate or precise the
information is.
▫ The KF is recursive
Not all data needs to be kept in storage
Re-processed every time
▫ The KF is a data processing algorithm or filter.
Only knowledge about system inputs and outputs is available
for estimation purposes.
Variables of interest can not be measured directly
A filter tries to obtain an optimal estimate of variables
8
10. PROBLEM
• The problem of robot localization is sometimes
referred to as the most fundamental problem in
making truly autonomous robots.
• The problem of robot localization consists of
answering the question Where am I? from a
robot's point of view.
▫ This means the robot has to find out its location
relative to the environment.
10
11. PROBLEM INSTANCES
• Position tracking problem the robot knows its initial
location.
▫ Tracking or local techniques is used to solve this problem.
• Wake-up robot or global positioning problem since
the robot does not know its initial position.
▫ Global techniques is used to solve this problem.
• Kidnapped robot problem, the robot does exactly know
where it is localized, but all of a sudden it is transferred, or
„kidnapped‟ , to another location without the robot being
aware of this.
▫ Global techniques is used to solve this problem.
• The dynamics of the environment complicate the robot is
driving around in. Dynamic environments contain other
moving objects and in these environments localization is
significantly more difficult
11
12. AVAILABLE INFORMATION
• First, it has a-priori information gathered by the
robot itself or supplied by an external source in an
initialization phase.
▫ Maps
▫ Cause-effect relationships
• Second, the robot gets information about the
environment through every observation and action
it makes during navigation.
▫ Driving
▫ Sensing
12
13. RELATIVE POSITION MEASUREMENTS
• Acquiring relative measurements is also referred
to as dead reckoning, which has been used for a
long time, ever since people started traveling
around.
• In robotic applications, relative position
measurements are either acquired by
▫ Odometry
▫ Inertial navigation.
13
14. ODOMETRY
• Greek words for road and measure
• Most used technique for making a robot find out its position.
• Odometry works by integrating incremental information over
time.
• By using wheel encoders to count the number of revolutions
of each wheel, the robot measures the distance it traveled and
its heading direction.
• It gives good short-term accuracy, inexpensive, and allows for
very high sampling rates
• Although Odometry causes increasing error in the location
estimate,
• It is the most easy to access form of position information and
therefore it is an important source of information for
localization.
14
15. INERTIAL NAVIGATION
• Estimates from inertial navigation are acquired by
integrating the obtained information from the sensors.
• It use gyroscopes and accelerometers to measure the rate of
rotation and acceleration of the robot.
▫ Gyroscopes, or rate gyros, or simply gyros, detect small
accelerations in orientation.
▫ Accelerometers measure small accelerations along the x or y
axis of a robot vehicle.
They suffer from extensive drift and are sensitive to bumpy
ground.
This problem can partially be solved by including a tilt sensor
that can cancel the gravity component.
• Problems
▫ The position estimates drift over time
▫ Thus the errors increase without bound.
15
16. ABSOLUTE POSITION MEASUREMENTS
• Supply information about the location of the
robot independent of previous location
estimates.
• The location is not derived from integrating a
sequence of measurements, but directly from
one measurement.
• Advantage:
▫ The error in the position does not grow
unbounded.
▫ Either supply the full location, or just a part of it.
16
17. • Methods
▫ Landmark Based
Active Landmarks, also called beacons
Passive Landmarks
Artificial Landmarks.
Natural Landmarks.
▫ Map Based
17
ABSOLUTE POSITION MEASUREMENTS
18. MULTI-SENSOR FUSION
• It can rely on a probabilistic approach, where notions
of uncertainty and confidence are common
terminology.
• Algorithms that solve the localization problem
combine initial information and relative and absolute
position measurements to form estimates of the
location of the robot at a certain time.
• Fusion of information from multiple sensors is
important, since combined information from
multiple sensors can be more accurate.
• It can reduce the effects of errors in measurements.
18
19. CONCLUSION
• Using absolute position measurements alone can be
done, but the disadvantages suggest that a
combination of relative and absolute position
measurements is better.
• The relative position measurements provide precise
positioning information constantly, and at certain
times absolute measurements are made to correct
the error in the relative measurements.
• Multi-sensor fusion provides techniques to combine
information from different sensors in providing a
best estimate of the robot's position.
19
