Road junction recognition and turn-offs for autonomous road vehicle navigation

This paper describes a computer vision system for an autonomous road vehicle (ARV) that is capable of negotiating complex road junctions and intersections in real time. In this system a model-based object recognition method has been used for detecting road junctions. Two main steps are performed to detect road junctions in images. The first step is to find a boundary between the candidate road junction surface and the current road surface. The second step is to locate the road junction surface. In this step, results of a region growing segmentation process and an edge detection process using edge segments are fused using the model voting strategy to locate the road junction surface. A vision-based re-orientation mode has been used to guide the ARV during the turn-off manoeuvre process. Results are presented for real road intersection images and performed on the hardware system in real time