基于图优化DWA算法的智能分拣机器局部运动轨迹最优规划

Optimal planning of local motion trajectory for intelligent sorting machines based on graph optimized DWA algorithm

摘要:

智能分拣机器人最优运动轨迹规划对于分拣效率和自动化程度息息相关。研究将以智能分拣机器人为例,创新性对图优化动态窗口方法的局部运动轨迹规划算法进行了分析。该方法首先利用动态窗口方法获取多条轨迹,然后引入避障和增加全局路径、点间距、非完整动力学、加速度、速度等约束到每条运动轨迹,进而创建超图。最后,采用C++软件开源的一般图优化采样生成的运动轨迹,并完成运动轨迹评价,找到最优运动路径。图优化前后DWA的局部运动轨迹规划算法在竖向方向位置的估计误差值较大,最小差值和最大差值分别为0.02m和3.25m,对应的时间为345s和697s。图优化前后DWA的局部运动轨迹规划算法的估计误差稍微偏大,差值约为0.02m/s。改进人工势场法的局部路径规划算法、改进时间弹性带的局部路径规划算法的目标运动轨迹重合度依次为72.68%和68.25%。研究设计的图优化DWA的局部运动轨迹规划算法能够更好地实现对障碍物的合理避让,与目标运动轨迹重合度为89.25%。研究成果有效解决了智能分拣机器人最优运动轨迹规划存在的规划效率低等问题,为实际移动机器人的移动控制技术的开发提供新的可能。

Abstract:

The optimal motion trajectory planning of intelligent sorting robots is closely related to sorting efficiency and automation level. The research will be based on the IoT mobile data collected by intelligent sorting robots, and innovatively analyze the local motion trajectory planning algorithm of the graph optimization dynamic window method. This method first uses the dynamic window method to obtain multiple trajectories, and then introduces obstacle avoidance and increases global path, point spacing, non holonomic dynamics, acceleration, velocity, and other constraints to each motion trajectory, thereby creating a hypergraph. Finally, using open-source C++software for general graph optimization, the motion trajectory generated by sampling is optimized, and the evaluation of the motion trajectory is completed to find the optimal motion path. The local motion trajectory planning algorithm of DWA before and after graph optimization has a relatively large estimation error value in the vertical position, with a minimum and maximum difference of 0.02m and 3.25m, respectively, and corresponding time of 345s and 697s. The estimation error of the local motion trajectory planning algorithm for DWA before and after graph optimization is slightly larger, with a difference of about 0.02m/s. The local path planning algorithm for improving the artificial potential field method and the local path planning algorithm for improving the time elastic band have a target motion trajectory overlap of 72.68% and 68.25%, respectively. The local motion trajectory planning algorithm of the graph optimized DWA designed for research can better achieve reasonable avoidance of obstacles, with a coincidence degree of 89.25% with the target motion trajectory. The research results have effectively solved the problems of low planning efficiency in the optimal motion trajectory planning of intelligent sorting robots, providing new possibilities for the development of actual mobile robot motion control technology.