Multi-shooting非线性MPC无人驾驶汽车轨迹跟踪控制

Multi-shooting nonlinear MPC trajectory tracking control for autonomous vehicles

中文关键词  无人驾驶汽车  轨迹跟踪  模型预测控制  Multi-shooting算法

英文关键词  autonomous vehicles  track tracking  model predictive control  multi-shooting method

中文摘要

      针对复杂路径下无人驾驶汽车轨迹跟踪精度不足和稳定性差的问题,本文研究了非线性MPC轨迹跟踪控 制策略.首先基于brush轮胎模型建立了考虑轮胎滑移的车辆动力学模型.然后利用Multi-shooting算法将状态方程 转化为函数连续性约束,独立计算每个预测子区间的状态微分方程组,采用序列二次规划方法对非线性最优控制问 题进行求解,得到最优控制输入.最后通过CarSim和MATLAB进行联合仿真,分别在36km/h,54km/h和72km/h3种 车速下跟踪双移线轨迹,与传统的线性MPC控制器对比,3种车速下本文的控制策略横向跟踪累积误差降低了 41.6%, 46.6%和36.5%, 控制效果得到提高,对不同车速有较好的鲁棒性. 与基于Single-shooting的非线性MPC控制 器对比,计算效率提高了61.9%,60%和52.8%,算法实时性得到了提高.

英文摘要

      Aiming at the problem of insufficient tracking accuracy and poor stability of driverless vehicle trajectory under complex path conditions, this paper studies the nonlinear MPC trajectory tracking control strategy. Firstly, a vehicle dynamics model considering tire slip is established based on the brush tire model. Then, the multi-shooting algorithm is used to transform the state equation into a function continuity constraint, and the state differential equations of each prediction sub-interval are calculated independently. The sequential quadratic programming method is used to solve the nonlinear optimal control problem, and the optimal control input is obtained. Finally, through the joint simulation of CarSim and MATLAB, the dual lane shifting trajectory is tracked at three speeds of 36 km/h, 54 km/h and 72 km/h, respectively. Compared with the traditional linear MPC controller, the lateral tracking cumulative error of the control strategy in this paper is reduced by 41.6%, 46.6% and 36.5%, and the control effect is improved, which has good robustness to different speeds. Compared with the nonlinear MPC controller based on single-shooting, the computational efficiency is improved by 61.9%, 60% and 52.8%, and the real-time performance of the algorithm is improved.