Comparison of Sensor Data Fusion Algorithms for Automotive Perception System

Reliable environment perception has an increased role in the spread of advanced driver assistance and autonomous systems. However, at present, neither automotive sensor can reliably meet the requirements for environmental sensing. The goal of sensor fusion is to construct a more accurate and reliable environment model using the available sensor data than either sensor is applied alone. The architecture of multi-object tracking algorithms is suitable for tracking and fusing object detections from multiple sensors. However, the parameters of widely used object-tracking algorithms such as PDA and JPDA can significantly affect their computational capacity demand and the reliability of the fused environment model. The present study aims to present the parameter sensitivity of these algorithms and their effects.