The invention discloses an artificial neural network-based system for pre-judging behaviors of surrounding vehicles. The system comprises a video analysis system, a radar system and an analytical processing system. The video analysis system acquires an all-directional video image of a current vehicle in real time and analyzes obstacle targets and road conditions in the image; the radar system acquires distances and contours of vehicles all around the current vehicle, in real time; the analytical processing system comprises an experience database; the experience database records collision data of the vehicles under different road conditions, different collision targets and different speeds; the experience database analyzes in real time the obstacle target and road conditions around the current vehicle and the distances and contours of the vehicles so as to pre-generate a target function of collision; historical data in the experience database is called to train the target function of collision to obtain a mature target function; the obstacle targets and road conditions acquired in real time and the distances and contours of the vehicles also acquired in real time are plugged into the mature target function so as to pre-judge the behaviors of the surrounding vehicles.
Artificial neural network-based system for pre-judging behaviors of surrounding vehicles
2015-06-10
Patent
Elektronische Ressource
Englisch