The Environment Perception and Path Planning Algorithm for Driverless Cars based on Computer Processing and Multi-sensor Fusion

Abstract

This paper selects suitable sensors and equipment based on the characteristics of urban road traffic. Then, a multi-sensor information fusion system based on millimeter-wave radar/camera is established. In this way, the road traffic safety can be effectively controlled. Then, by improving D-S evidential reasoning, a "goal-decision" two-level information fusion method is established. Then, the multi-layer fusion experiment of multi-source sensing data under a tunnel environment is carried out. Experiments show that the ROI region correlation using camera and millimeter wave radar can improve the detection accuracy by 9.66%, effectively solving the limitations of single-sensor detection for automatic driving under complex conditions. The D-S evidential reasoning method processes the automatic driving sensor data. This reduces the false report rate of the sensor by 2.75%.