Abstract: In the robot navigation scene with dynamic obstacles, the timed elastic bands (TEB) algorithm cannot distinguish the types of obstacles, and it is easy to treat dynamic obstacles as static types, resulting in collisions in the navigation process and unable to complete the navigation task. Aiming at the real dynamic environment of the robot, two mean filters were used to filter the laser point cloud to achieve dynamic obstacle detection, and then the dynamic obstacle cost map layer was added to cluster the detected obstacles. Kalman filtering was used to perform the trajectory tracking and state prediction on the motion state of dynamic obstacles. Based on the current motion state of the robot, a dynamic obstacle avoidance trajectory plan was carried out. The Ackerman steering structure robot was used for simulation and the repeated experiments in real dynamic environments. The research results show that the dynamic obstacle avoidance trajectory planning system with improved TEB algorithm can perform real-time trajectory planning in complex dynamic environments and generate safe and smooth local trajectories, which realizes the dynamic obstacle avoidance to complete the navigation task and can meet the dynamic obstacle avoidance requirements of mobile robots.

Key words: vehicle and mechatronics engineering; TEB algorithm; dynamic obstacle avoidance; Kalman filter; trajectory planning

摘要： 在动态障碍物存在的机器人导航场景下，时间弹性带算法（timed elastic bands, TEB）无法区分障碍物类型，易将动态障碍物视为静态类型去处理，致使导航过程出现碰撞而无法完成导航任务。针对机器人运行真实动态环境，采用两个均值滤波器对激光点云进行滤波处理，实现动态障碍物的检测，而后增加动态障碍物代价地图层对检测到的障碍物进行聚类，利用卡尔曼滤波对动态障碍物的运动状态进行轨迹跟踪及状态预测，结合机器人当前运动状态作出动态避障的轨迹规划，采用阿克曼转向结构机器人进行仿真及真实动态环境下的重复实验。研究结果表明：改进TEB算法的动态避障轨迹规划系统能够在复杂动态环境中进行实时轨迹规划，生成安全平滑的局部轨迹，实现动态避障完成导航任务，能够满足移动机器人的动态避障要求。

关键词: 车辆与机电工程；时间弹性带算法；动态避障；卡尔曼滤波；轨迹规划