人工势场法在无人车避障系统中的改进

Improvement of the Artificial Potential Field Method Implementation

传统的人工势场法易出现陷阱区域,在障碍物前面产生震荡等缺点.通过建立避障的动力学模型,引入对障碍物的模糊分类和BP网络对传感器数据的融合优化,使得无人驾驶车在通过有障碍物的道路中能够实现自主避障,做到提前预判,有效消除了震荡或死区引起的系统不稳定性.相比于传统人工势场算法,采用改进算法使得系统在稳定性和效率上有进一步提高.

The traditional artificial potential field method has shortcomings of often falling in trap area and oscillating in front of obstacles. A dynamic model of obstacle avoidance, which uses fuzzy classification to identify obstacles and BP network to fuse the sensor data optimally, is established for unmanned vehicles to anticipate obstacles in the path in advance and avoid obstacles autonomously to achieve eliminating oscillating or dead zone causing system instability. Simulation indicates that, compared with traditional artificial potential field algorithm, the proposed algorithm further improves the system stability and efficiency.