22nd International Conference on Integrated Navigation Systems
1. Indoor Navigation Using Foot-Mounted IMU Aided
with Heterogeneous Additional Information
A.S. Smirnov, A.A. Panyov, V.V. Kosyanchuk
2. GNSS doesn’t work
inside buildings
Firefighters and rescue
workers need to navigate
indoors
There is no any prior
information about the
building. Setting up
infrastructure is not possible
High accuracy is
required
The problem
3. Solution
1. Calculation of integrated navigational
solution on the basis of IMU
measurements.
2. Using ZUPT technique
The main drawback is accumulation of error over time.
IMU is just 22.5x20mm. It
can be foot-mounted or
embedded into the hill
IMU has built-in accelerometer,
gyroscope, magnetometer and
barometer
4. Solution
1. Calculation of integrated navigational
solution on the basis of IMU
measurements.
2. Using ZUPT technique
3. Magnetic heading correction.
IMU is just 22.5x20mm. It
can be foot-mounted or
embedded into the hill
IMU has built-in accelerometer,
gyroscope, magnetometer and
barometer
5. Solution
1. Calculation of integrated navigational
solution on the basis of IMU
measurements.
2. Using ZUPT technique
5. Particle filter
• corrects position by comparing current RSSI
measurements with the radio-map.
• Using digital plan of the building (in a case of
availability)
3. Magnetic heading correction.
4. Using SLAM technique, specifically building
radio-map of the building on the basis of RSSI
measurements
Predicted radio-map of the building for a Wi-Fi
transmitter. Color bar denotes the value of
RSSI
6. Results
Comparison of SINS (blue line) and SINS with SLAM
(purple line) performance on one of closed-loop
trajectory with 149 m length. Start/end position is
marked over with red circle.
A prototype of navigation system was created. It allows to perform instant navigation in unknown
environments with acceptable level of accuracy (<1% of travelled distance).
Track
length, m
SINS error,
m
SINS +
SLAM
error, m
112 1.18 1.33
224 2.55 1.82
560 4.86 1.52
896 7.21 1.19
1121 8.93 1.03
Comparison of algorithms performance: SINS and SINS with
SLAM method and magnet course correction