A New Approach to Forward Collision Avoidance

Forward collision occurs most frequently and involves the largest number of fatalities among various types of collisions in traffic accidents. Active research has been carried out to develop systems to avoid or mitigate forward collision. These systems are effective in reducing the number and the severity of occupant injuries, however, there is still margin for improvement, especially for reducing jerk felt by the occupants while improving system performance. We propose a new approach to forward collision avoidance by integrating distance control and emergency braking according to likelihood of collision. We also use a low-cost camera vision sensor for detecting lanes and vehicles, and estimating collision risks. Evaluation of the proposed approach using our driving simulator showed the improvement in collision avoidance in terms of earlier system intervention, increased relative distance and reduced relative speed to a preceding car. Meanwhile, the reduced deceleration rate implied that vehicle occupants would feel jerk less during system intervention.