ON THE REINFORCED RELIABILITY OF FORWARD COLLISION WARNING SYSTEM WITH MACHINE LEARNING

Researches on Advanced Driving Assistance System (ADAS) have been increased and ADAS has become a major field in automotive industry over the past few decades. One of the essential means comprising ADAS is Forward Collision Warning System (FCWS), which, from the safety point of view, should be designed to have high reliability and robustness against disturbance in noisy environment. FCWS in the market, however, still have vulnerability sensitive to the driving conditions and mounted sensors. We thus introduce an integrity monitoring approach on FCWS using the vehicle braking distance predictor which employs the neural network to predict the braking distance and aims to improve the reliability of the FCWS by preventing the accident by judging the abnormality of the system. The performance of the proposed vehicle braking distance predictor demonstrated the superiority in term of the prediction capability and robustness against the external noise compared with the existing mathematical braking distance model.