Feature-level map building and object recognition for intersection safety applications

Abstract

Accidents at intersections happen when drivers perform inappropriate manoeuvres. Advanced sensor systems will enable the development of Advanced Driver Assistance Systems (ADAS) which can assess the potential for a collision at a junction. Accurate localisation of the driver's vehicle and path prediction of other road users can be fused with traffic signal status and other information. The ADAS system can use this fused data to assess the risks to the driver and other road users of potentially hazardous situations and warn the driver appropriately. The accurate localisation of the host vehicle is achieved by utilising individual sensors' feature-level maps of the intersection. The INTERSAFE project will independently use video and Laserscanner sensing technologies for localisation and then fuse the individual outputs to improve the overall accuracy. The Laserscanner system will also be used to track and classify other road users and obstacles, providing additional data for the path prediction and risk assessment part of the application.