Cooperative navigation (part II) An example using foot-mounted INS and UWB-transceivers Jouni Rantakokko
Aim Increased accuracy during long-term operations in GNSS-challenged environments for - First responders - Dismounted soldiers
Outline Short recapitulation - user needs, preliminary requirements and enabling technologies Foot-mounted INS IR-UWB Cooperative navigation Summary
Outline Short recapitulation - user needs, preliminary requirements and enabling technologies Foot-mounted INS IR-UWB Cooperative navigation Summary
First responder requirements What is desired of an indoor positioning system? - Robust, light-weight, small, power efficient, low-cost - Reliable estimation of room/floor (or equivalent) - Error estimates (integrity monitoring) - No dependence on pre-installed infrastructure - Seamless outdoor-indoor coverage, with consistent performance in all relevant scenarios
First responder requirements Requirement estimates ~ 1-3 m accuracy ~ near 100% availability in all environments << 1 kg, incl. processing unit and visualization ~ 8-24 h battery ~ 1k >> 30 min s indoors but they also depend heavily on user group, type of operation, and usage (e.g. avoid friendly fire or efficient C2),
Summary enabling technologies Today s technology insufficient crucial user requirements cannot be fulfilled (simultaneously) with a single positioning sensor/sub-system - 3D accuracy and availability indoors - Integrity monitoring, estimate of position errors - Price, size, weight, battery - Tactical behavior Sensor fusion approach needed - What sensors should be used? - How should the sensor data be combined?
Summary enabling technologies Multi-sensor system is required - GNSS-receivers GPS + Galileo + GLONASS(?) Different services and signals from all SatNav systems are available - Inertial sensors, magnetometers and barometer Foot-mounted, back-mounted and/or co-located with camera - Imaging sensors -Doppler radar - Ranging devices (RF, acoustic, ) - Cooperative navigation HW - focus on inexpensive and lightweight sensors - e.g. MEMS-based SW - quality of multi-sensor fusion algorithms and integrity monitoring of positioning sub-systems dictates performance
Vision for the future Consumer market - Smartphone as multi-sensor platform accelerometers, gyros, magnetometers, barometer, camera Radio-based ranging Acoustic Map WLAN Cooperation - Positioning app s available
Vision for the future First responder market - Individual sensors accelerometers, gyros, magnetometer, barometer camera, acoustic, Doppler-radar - Radio-based ranging and cooperative navigation Impulse radio (UWB) Ranging waveform in future FR radios (SDR) FR-to-FR and/or FR-to-vehicle -Maps LiDAR (3D) and cooperative mapping
Possible future concept for robust first responder positioning systems in urban operations IR-Camera IR-Camera IMU IMU Cooperative navigation for high accuracy during long-term operations GPS GPS - Impulse radio (UWB) for ranging and exchange of information? - What sensor data should be exchanged to enhance efficiency of cooperation? Accelerometers Accelerometers Gyro Gyro Magnetometers Magnetometers Barometric Barometric altimeter altimeter First responder equipped with individual sensors for position estimation J. Rantakokko et al, Accurate and reliable soldier and first responder indoor positioning: Multi-sensor systems and cooperative localization, IEEE Wireless Communications Magazine, April 2011
Possible future concept for robust soldier positioning systems in urban operations Camera Camera IMU IMU Cooperative navigation for high accuracy during long-term operations GNSS GNSS GUI GUI Processor Processor Accelerometrar Accelerometrar Gyro Gyro Magnetometrar Magnetometrar Barometer Barometer - Impulse radio (UWB) for ranging and exchange of information? - What sensor data should be exchanged to enhance efficiency of cooperation? Soldier equipped with individual sensors for position estimation - What sensors should be used? - How should they be combined (sensor fusion)? J. Rantakokko et al, Accurate and reliable soldier and first responder indoor positioning: Multi-sensor systems and cooperative localization, IEEE Wireless Communications Magazine, April 2011
Backpack contains laptop, battery, converter and UWB - UWB-ranging measurements and analysis - Foot-mounted inertial navigation experiments - Collaborative sensor fusion experiments Work consists mainly of three parts Scope of work
Outline Short recapitulation - user needs, preliminary requirements and enabling technologies IR-UWB Foot-mounted INS Cooperative navigation Summary
Cooperative navigation approach UWB-ranging Transfer of pos-info P. Strömbäck et al, Foot-mounted Inertial Navigation and Cooperative Sensor Fusion for Indoor Positioning, Proc. of ION ITM, San Diego, CA, USA, January 2010.
UWB-ranging TimeDomain PulsOn 220 EVK - 3.2 GHz 10-dB BW (approx. 3.1-5.3 GHz) - Round-trip time estimation - Sample waveforms for ranging, comm s and radar Error models - Differentiate between LOS NLOS with direct path NLOS w/o direct path
UWB-ranging (LOS) LOS experiments - Distance 2-50 meter, multipath environment - Small ranging errors and low standard deviation - Scale-factor increase in ranging error with decreasing distance (calibration issue)
UWB-ranging (NLOS) 4th floor measurements - Plaster-board walls, whiteboards, safety-cabinets -a=6 m -b=9 m - c=12 m - d=15 m
UWB-ranging (NLOS) Approx. ranging distance < 15 m (<4 walls) Fairly static multipath Probability of undetected direct path ~6% at 12m Not corrected for distance scaling factor
UWB-ranging (NLOS) 3rd floor measurements - Concrete walls, different thickness, whiteboards
UWB-ranging (NLOS) Approximate ranging distance <9 m (<3 walls) Fairly static multipath Probability of undetected direct path ~7% at 6m
UWB-ranging through whiteboard Whiteboard has low signal penetration, generates multipath
UWB-ranging between floors
Outline Short recapitulation - user needs, preliminary requirements and enabling technologies IR-UWB Foot-mounted INS Cooperative navigation Summary
Foot-mounted INS Navigation algorithm based on a 15-state Extended Kalman Filter - 3D position, velocity & orientation (9 states) - Accelerometer and Gyro bias (6 states) Measurement updates to KF consists of a zero velocity (ZUPT) during the stance phase Detection of ZUPT based on gyro signal amplitude averaged over a few samples - Accurate during regular walking, errors may increase during other movement regimes
Foot-mounted INS Accelerometers Gyros ZUPT detection Corrections Measurement update (ZUPT) INS Mechanization Position Velocity Orientation Velocity EKF
Outline Short recapitulation - user needs, preliminary requirements and enabling technologies Foot-mounted INS IR-UWB Cooperative navigation Summary
Cooperative navigation Estimate position, velocity, and orientation (and uncertainties) based on foot-mounted sensors Measure range to cooperative unit Exchange position estimates and uncertainties between cooperative units Update units positions, velocities, and orientations RF-ranging Transfer of pos-info
Cooperative navigation -overview Accelerometers Gyros ZUPT detection Corrections Measurement update (ZUPT) INS Mechanization Position Velocity Orientation Velocity EKF
Range, Position & Covariance Cooperative navigation -overview UWB Accelerometers Gyros ZUPT detection Measurement update (Range) Corrections Measurement update (ZUPT) INS Position Mechanization Position Velocity Orientation Velocity EKF
Experiments Scenario 1 - single floor (horizontal plane) Scenario 2 - two floors (horizontal and vertical motion) Multiple laps for each test, at two different occasions Closed loop tests - Start point and stop point are the same within a few centimeters Initial orientation, position and velocities - Known to a few deg & cm & mm/s Initial gyro bias - Chosen as the mean value of gyro signal during a minute of standing still (this is only a rough estimate)
Results - Cooperative positioning No cooperation Cooperation Scenario 1 One floor Blue: 9.0 m (3D pos error) Red: 8.2 m (3D pos error) Blue: 1.4 m (3D pos error) Red: 3.5 m (3D pos error)
Results - Cooperative positioning No cooperation Cooperation Scenario 2 Two floors Blue: 3.8 m (3D pos error) Red: 2.8 m (3D pos error) Blue: 1.3 m (3D pos error) Red: 2.7 m (3D pos error)
Results One floor Distance between start and end position (3D-position error at end of lap) Red Unit Blue Unit Lap Name Foot-mounted Cooperation Foot-mounted Cooperation 2009 1A 2.3m 2009 1B 2.5m N/A N/A 6.8m 7.6m N/A N/A 2010 1A 79m N/A 4.9m N/A 2010 1B 8.2m 3.5m 9.0m 1.4m 2010 1C 12.7m N/A 6.8m N/A
Results Two floors Distance between start and end position (3D-position error at end of lap) Red Unit Blue Unit Lap Name Foot-mounted Cooperation Foot-mounted Cooperation 2009 2A 0.94m 2009 2B 1.24m N/A N/A 5.0m 6.0m N/A N/A 2010 2A 3.19m N/A 2.2m N/A 2010 2B 3.01m N/A 8.3m N/A 2010 2C 2.8m 2.7m 3.8m 1.3m
Summary Cooperation has the potential to improve performance significantly - The improvement in these two tests is likely larger than what will be achieved in general Smaller errors now with foot-mounted system Heading drift experienced particularly in Scenario 1 ver favorable towards cooperative approaches - Multiple cooperative units will lead to further improvements
Q&A