Abstract
The research aims to develop the system called “autonomous driving intelligence” as a driver assistance system considering expert drivers’ anticipatory behaviour to avoid potential collisions. The motivation of the research is to compensate the degraded driving ability in recognition, decision and operation process of low-performance drivers, resulting in further reduction of accidents. The current active safety systems are based on warning and autonomous braking in emergency situations which the effectiveness of the system is limited. In order to further reduce accidents, it is important to consider how we can enhance the collision avoidance performance by utilizing the knowledge from expert drivers. The proposed driver assistance system as the autonomous braking/steering system is implemented in the experimental vehicle. Here, the driving task is to overtake a parked vehicle where there might be a pedestrian darting out behind the parked vehicle. The LIDARs attached on the vehicle are used to detect the parked vehicle and the pedestrian. Based on the risk potential algorithm, the experimental result shows that the vehicle gradually decelerated when getting closer to the parked vehicle and then the steering system was controlled to move the vehicle to a free space according to the path planning by risk potential. The automatic emergency brake was activated when the pedestrian darted out to the vehicle driving corridor. Test drives show that the system can work in practice without collision with the pedestrian in safe manner similar to the expert driver.
KEYWORDS – Driver Assistance System, Autonomous Driving, Collision Avoidance, Risk Potential, Hazard Anticipatory