基于改进型人工势场的无人车局部避障

Local Obstacle Avoidance Based on Improved Artificial Potential Field

针对经典人工势场的局限性,引入速度势场函数并采用高斯函数作为速度势场与障碍物势场函数主体,设定不同的横纵向危险影响范围,设计了改进型人工势场,获得灵敏度不同的横向与纵向反应空间。采用模型预测控制算法,将改进型势场函数作为模型预测算法的成本函数参数,并将局部路径规划算法和轨迹跟踪控制转化为基于模型预测求解势场、控制增量最小的优化问题。通过Simulink/CarSim及实车实验验证了该算法的可行性:在保证安全避障的同时,车辆自主避障过程中侧向加速度、横摆角等参数均在稳定性要求范围内。

Aiming at the limitations of the classic artificial potential field,the velocity potential field function was introduced,and the Gaussian function was used as the main body of velocity potential field and obstacle potential field function,different lateral and longitudinal hazard influence ranges were set.The improved artificial potential field was designed to obtain lateral and longitudinal reaction spaces with different sensitivities.The model predictive control algorithm was used to model the vehicle dynam⁃ics,and the improved potential field function was used as a cost function parameter of the model predic⁃tion algorithm.By this method,the local path planning algorithm and trajectory tracking control were transformed into the optimization problem based on model prediction to solve the potential field and minimize the control increment.Simulink/CarSim and real vehicle experiments show that the algorithm is feasibility.While ensuring safe obstacle avoidance,the parameters such as lateral acceleration and yaw angle are within the range of stability requirements during vehicle autonomous obstacle avoidance.