Sensor Perception Reliability in Automated Vehicles

With the increasing number of on-road vehicles, safety is one of the important concerns. Advanced Driver Assistance Systems (ADAS) provide many features that help in reducing human errors while driving. These features assist the driver either by giving warning signals or performing some tasks to avoid collision. Sensors being the eyes of ADAS have to provide accurate information about the surroundings to enable proper decisions in critical situations. To overcome the limitations of individual sensors and to improve the efficiency of the sensor performance, input from two different sensors is fused which will help to achieve accurate decisions. In this paper, we propose estimation of uncertainty parameters that affect the perception reliability for sensor fusion technique by varying noise. The vision camera and radar sensors are considered for sensor fusion and tested for synthetically generated various driving scenarios.