Ubiquitous Navigation James Pinchin
Overview Introduction Ubiquitous Navigation Inertial Navigation Initialisation & Drift Correction for Inertial Navigation
Positioning Technologies 0.1 1 5 10 50 100 10k Remote Rural Suburban Urban Indoor GPS AGPS SBAS (GPS) LAD GPS RTK GPS PPP HS GPS SA GPS Cell Phone Network / DAB / DVB Infrastructure UWB RFID Bluetooth WLAN Accuracy (m)
Inertial Navigation System (INS) Stand alone No infrastructure needed “Cheap” $10 – $100k Sensors exist and are widely available
Inertial Navigation System (INS) Measure Acceleration Integrate to Obtain Velocity Integrate to Obtain Change in Position Measure Rotation Rate Integrate to Obtain Change in Attitude
INS - Problems Initialisation Position Orientation Sensor Biases Error Propagation Initialisation Biases Sensor Biases / Noise
INS - Problems
INS - Augmentation Velocity Aiding Eg. ZUPTs Constrains drift in INS position estimate & allows IMU sensor biases to be estimated ZUPTs allow low cost IMUs to be used for high accuracy navigation
Pedestrian INS Foot Mounted IMU ZUPT every ~0.4 second Lasts around half of a step
Final Position Error (~75m) Pedestrian INS Heading Error in Shop Final Position Error (~75m)
CHAIN
GPS Integration - RTK Problems Remain Initialisation Position Drift Correction
The High Accuracy Testbed
GPS Integration - RTK
GPS Integration - HSGPS
GPS Integration - HSGPS
GPS Integration – HSGPS - Tight
GPS Integration – HSGPS - Tight
Inertial Navigation - Initialisation A position can be refined interactively by using using a phone’s location capabilities (GPS, Wi-Fi, compass, etc), allowing the user to describe their position to a higher level of precision.
Inertial Navigation - Initialisation A structured data representation of the area, allows meaningful landmarks or features to be identified and a dialog to be generated to interactively refine the user’s precise location.
Inertial Navigation - Initialisation Additional detailed sources can be iteratively added as the location becomes more precise. For instance the dialog may determine that the user is indoors and supplement the data with floor plans.
Future Work Further Development & Testing of ‘Tight’ GPS Integration Android Implementation of Inertial / CHAIN solution Map / Local Area Model Aiding
www.horizon.ac.uk Questions? James.Pinchin@nottingham.ac.uk