为实现结构化道路下智能车最优避障轨迹的规划，设计了基于层次分析法的轨迹择优体系。首先，以三次 B 样条曲线为路径规划器，生成满足曲率连续且曲率最值受控的路径，并以此构造路径簇; 其次，为实现主客观指标的量化表达，以平滑性和经济性为准则，以路径长度、曲率和、曲率变化率和以及偏离目标点的距离为子准则，构造 AHP 路径择优体系，筛选出最优的路径; 然后，采用三次多项式表征速度相对时间的变化，以满足速度、加速度、加速度导数的连续; 最后，参考国家标准设计测试场景，验证方法的稳定性与算法的实时性。经 5000 次循环仿真测试表明: 算法具有较高的实时性，0. 1s 内能规划出可行轨迹的概率在 94%，0. 16s 内则能 100%规划出可行轨迹; 规划的轨迹便于车辆跟踪，方法具有较高的稳定性，实车峰值横向误差小于0. 21m，峰值车速误差小于0. 42m/s，平均车速误差小于0. 11m/s，总体呈收敛的趋势。

The optimal trajectory selection system based on analytic hierarchy process (AHP) was designed，in order to plan optimal obstacle avoidance trajectory for intelligent vehicle in structured road． Firstly，cubic B-spline curve was employed as the planner to generate path with continuous curvature and controlled maximum curvature，based on which path sets were constructed．Secondly，in order to achieve quantitative expression of subjective and objective indicators，the AHP system for optimal path selection was constructed and the optimal path was selected，by considering smoothness and economic criteria as the criteria，path length，curvature sum，curvature change rate sum and distance from target point as sub-criteria． Then，cubic polynomial was used to represent the change of velocity relative to time，so as to satisfy the continuity of velocity，acceleration and acceleration derivative． Fi-nally，test scenarios were designed according to national standards to verify the stability of the method and the real- time performance of the algorithm． After 5 000 cycles of simulation test，the result shows that the algorithm has a good real-time performance． The probability of feasible trajectory planning within 0. 1 s is 94%，while within 0. 16 s it is 100%． The planned trajectory is convenient for vehicle tracking and the method has a high stability． The peak lateral error of real vehicle test is less than 0. 21 m，the peak speed error is less than 0. 42 m/s，and the av-erage speed error is less than 0. 11 m/s，which show an overall trend of convergence．