IMPROVED LOCALISATION FOR TRAFFIC FLOW CONTROL

Abstract

Localisation of vehicles plays an important role in future traffic control concepts. To solve this task several sensors may be used. Each of these different sensors has its own, specific disturbances. Incremental measurement of wheel rotation to get information about the covered distance is distorted by slip and rugged roads, measuring the orientation of the vehicle by means of a magnetometer suffers from internal and external disturbing magnetic fields, and vehicle vibrations disturb yaw rate measurement. A method to improve autonomous localisation based on Kalman filtering is presented. By estimating the variance of the different sensor data the Kalman Filter parameters can be varied to achieve improved system behaviour. Results are presented for an in-town drive. The system is just about to be implemented in real-time in a test vehicle at the University of Karlsruhe.