Obstacle detection and avoidance for automated vehicle: a review

In any vehicle motion, obstacle need to be read very carefully for detection; if detection is reliable and faithful, then and then only optimized solution to avoid it can be precisely decided. There are various approaches to read the obstacles. One of the initial approaches was by using ultrasonic sensors and laser range scanner and was adopted with acceptable accuracy. But as the camera applications geared up with their advances in types and sophistications, a pair of camera or stereo vision has geared up for this application. Previously, data received from the sensors were used after conditioning or pre-processing, consequently computational overhead to decide (perception variation) disparity. This disparity needs to be overcome in real time or with acceptable latency, so as to avoid collision with obstacle along the decided path. This paper gives one such approach and then discusses the methodologies used by the earlier researchers for detection using stereo vision. This will help them to deduce better approach further in this area.