Stereovision-based 3D obstacle detection for automotive safety driving assistance

This paper describes the implementation of a real-time architecture dedicated to obstacle detection in the automotive domain, and more particularly to pre-crash situations. The method, based on stereovision, is of high complexity and can not run in real-time on standard processors. Therefore, the application is accelerated with the use of special purpose hardware; in particular, a highly parallelized disparity engine is presented. A prototype board was built, which achieves a performance of 460 GOPS and computes the application at the rate of 22 frames per second, thus reaching road safety constraints.