Multi-target Detection based on Multi-sensor Redundancy and Dynamic Weight Distribution for Driverless Cars

Abstract

The perception system of unmanned vehicles has developed rapidly in recent years, but in the process of multisensor fusion, a certain sensor's perception error often occurs which causes great errors in the perception results of the entire multi-sensor system and is harmful to human life and property. Therefore, it requires redundant fusion of the objects perceived by the sensors, and sets weights during the fusion process to eliminate sensors with poor detection results. This paper establishes a multi-sensor perception and a dynamic weight distribution and eliminate system. Multi-sensor weight distribution is performed by Kalman tracking variance. Such a system will greatly reduce the errors of the perception system.