Abstract
Most collision avoidance systems warn the driver or act on the brake pedal in order to reduce speed and stop the vehicle when an obstacle is detected on the vehicle trajectory. In some cases, braking is not effective enough and other kind of manoeuvre is required. This paper presents a system that looks for free areas around the vehicle and moves it to those areas when a dangerous situation is detected. The system acts automatically on the brake and accelerator pedals and on the steering system.
The assistance system consists of 3 modules: 1) Obstacle detection module whose mission is to detect and track obstacles and determine their speed and direction, and positions the vehicle and the obstacles in accurate and detailed digital maps providing extra information for free areas identification; 2) Decision module that decides the best possible action to take to avoid an accident or reduce its consequences based on the information from the surroundings taking into account the road characteristics, the own vehicle movement, the obstacles and should generate manoeuvres that are feasible in practice according to vehicle dynamics and should not be surprising for the drivers; 3) Autonomous manoeuvring module that carry out the manoeuvre selected by the previous module by acting on the vehicle’s controls, without removing the capability of the driver of acting over those controls if he wants.
The system has been installed in a passenger car and has been tested on the University Institute for Automobile Research test track. Some controlled scenarios have been defined and the system has been tested. The detection system, the decision module and the autonomous manoeuvres are assessed. Even considering that the detection module is not the core of what is here presented; false alarms could be removed thanks to novel algorithms. On the other hand, the decision module have shown its capacity for selecting the most convenient manoeuvre in each case, considering vehicle dynamics, free areas in the surroundings of the vehicle and the simplest manoeuvre to avoid the accident. Finally, the automatic action on the vehicle controls is correct and quick enough; so the response of the vehicle follows accurately what the system predicted.
The system takes advantage of a complete automation of the vehicle controls in order to decide the most convenient manoeuvre to avoid accidents. The system includes general decision algorithms that can provide a better solution when positioning and detection systems are improved.
KEYWORDS – Collision avoidance; automation; brake; steering; pre-collision system