A Machine Vision Based System for Guiding Lane-change Maneuvers

We propose a new approach for vision based longitudinal and lateral ve-hicle control which makes extensive use of binocular stereopsis. Longitudi-nal control - i.e. maintaining a safe, constant distance from the vehicle infront - is supported by detecting and measuring the distances to leadingvehicles using binocular stereo.A known camera geometry with respect tothe locally planar road is used to map the images of the road plane in thetwo camera views into alignment.Any significant residual image disparitythen indicates an object not lying in the road plane and hence a potentialobstacle. This approach allows us to separate image features into those lyingin the road plane, e.g.lane markers, and those due to other objects. Thefeatures which lie on the road are stationary in the scene and appear to moveonly because of the egomotion of the vehicle. Measurements on these featuresare used for dynamic update of (a) the camera parameters in the presenceof camera vibration and changes in road slope (b) the lateral position of thevehicle with respect to the lane markers.In the absence of this separation,image features due to vehicles which happen to lie in the search zone for lanemarkers would corrupt the estimation of the road boundary contours. Thisproblem has not yet been addressed by any lane marker based vehicle guid-ance approach, but has to be taken very seriously, since usually one has tocope with crowded traffic scenes where lane markers are often obstructed byvehicles. Lane markers are detected and used for lateral control, i.e. followingthe road while maintaining a constant lateral distance to the road boundary.For that purpose we model the road and hence the shape of the lane markersas clothoidal curves, the curvatures of which we estimate recursively along theimage sequence. These curvature estimates also provides desirable look-aheadinformation for a smooth ride in the car.