基于1R1V的传感器融合算法开发

The Development of 1R1V Sensor Fusion Algorithm

高级驾驶辅助系统中使用摄像头和毫米波雷达融合的方案可以进一步提升感知结果的鲁棒性,扩展系统的使用场景。本文以满足在微控制器中使用的1R1V(1 Radar 1 Vision)融合算法为设计目标,先后完成了卡尔曼滤波模块、Munkres匹配算法模块及其它融合辅助模块的开发。之后在仿真环境中搭建了测试场景,对1R1V融合算法的效果进行了测试。结果表明,本文开发的融合算法能够对目标进行稳定的跟踪,最大的速度跟踪误差小于0.02 m/s,最大的位置跟踪误差小于0.11 m。

The fusion solution using camera and radar in the advanced driving assistance system can further improve the robustness of perception results and expand the system's usage scenarios.To meet the design goal of the 1R1V(1 Radar 1 Vision)fusion algorithm used in the microcontroller,this paper successively completed the development of the Kalman filter module,Munkres matching algorithm module and other fusion auxiliary modules.Then a test scene was built in the simulation environment to test the effect of the 1R1V fusion algorithm.The results show that the fusion algorithm developed in this paper can track the target stably,with the maximum velocity tracking error less than 0.02 m/s and the maximum position tracking error less than 0.11 m.