【作者】 李玮； 段建民； 龚建伟；

【Author】 LI Wei~1,DUAN Jian-min~1,GONG Jian-wei~2 (1.College of Electronic Information and Control Engineering,Beijing University of Technology,Beijing 100124,China; 2.Intelligent Vehicle Research Center,Beijing Institute of Technology,Beijing 100081,China)

【机构】 北京工业大学 电子信息与控制工程学院； 北京理工大学 智能车辆研究所；

【摘要】 以智能车辆换道过程为研究对象,结合多项式理论和动态RBF神经网络,提出1种车辆换道路径规划方法,得到在一定边界条件下智能车辆换道路径的近优解。该方法首先利用矩形包裹换道车辆及障碍车辆并对其进行碰撞检测,然后利用动态RBF神经网络生成合理的车辆换道边界条件,最后在边界条件以及性能指标函数的约束下,根据多项式理论得到以时间为参数的换道路径近优解。其中动态RBF神经网络具备在线学习能力,能够利用具有优良性能指标的边界条件实现自更新。计算机仿真验证了该方法的正确性及有效性,尤其是在复杂路面情况下体现了该换道路径规划算法的优势。更多还原

【Abstract】 Based on polynomial theory and radial basis function(RBF) neural network,a path planning method for the intelligent vehicles lane changing process was proposed.The near-optimal solutions of the lane changing path in the fixed boundary conditions can be obtained by this method.In this method the lane changing vehicle and obstacle vehicles were presented by rectangle,and then in the constraints of collision detect conditions,boundary conditions and comfort performance index that the near-optimal solutions of the lane changing path were calculated.In addition,the dynamic RBF neural network was used to solve the problem that how to select a reasonable boundary conditions.By this dynamic RBF neural network the reasonable boundary conditions were calculated and the neural network has the function of online learning,which was optimized by itself.Simulation results prove the correctness and feasibility of this algorithm, and illustrative examples show the advantage of this new method in the case of lane changing with multiple obstacles.更多还原