为了在车辆颠簸、光照阴影、路面其他特征干扰等环境下为无人驾驶车辆规划控制系统提供有效范围内准确可靠的导向信息,提出一种基于视觉的车道线检测算法。该算法通过预处理消除车辆颠簸以及光照阴影对视觉图像的影响;基于马尔可夫模型,根据单帧检测结果更新图像俯视图中车道线位置的概率分布图,在存在路面其他特征干扰时可靠地完成车道线检测;根据检测结果,自动选择曲线模型对直道弯道进行拟合,并将拟合结果以期望路径点序列的形式输出给规划控制系统。该算法应用于无人驾驶车辆平台,在2013年"中国智能车未来挑战赛"中,单纯依靠视觉完成17 km的城郊环湖路面行驶,时速达到40 km/h.
A multi-constrained model predictive control(MPC)algorithm for trajectory tracking of an autonomous ground vehicle is proposed and tested in this paper.First,to simplify the computation,an active steering linear error model is applied in the MPC controller.Then,a control increment constraint and a relaxing factor are taken into account in the objective function to ensure the smoothness of the trajectory,using a softening constraints technique.In addition,the controller can obtain optimal control sequences which satisfy both the actual kinematic constraints and the actuator constraints.The circular trajectory tracking performance of the proposed method is compared with that of another MPC controller.To verify the trajectory tracking capabilities of the designed controller at different desired speed,the simulation experiments are carried out at the speed of 3m/s,5m/s and 10m/s.The results demonstrate the MPC controller has a good speed adaptability.
