基于在线预测和规划的机器人动态跟踪抓取方法

RobotDynamic Tracking and Grasping Method Based on Online Prediction and Planning

基于机器人操作系统(Robot operating system, ROS)框架,以6自由度ABB机器人为研究对象,设计了一种基于在线预测和规划的机器人动态跟踪抓取方法。系统通过预先检测目标物的运动状态在线预测其运行轨迹,并对预估轨迹进行运动规划得到机械臂的跟踪轨迹,以实现机器人对移动目标物体进行动态跟踪抓取。针对ROS系统与机器人通信反馈机制造成机器人跟踪过程断续、跟踪轨迹不连续的问题,设计了一种多轨迹融合方法,用于运动规划时融合多段轨迹,实现了机器人持续地跟踪移动物体。针对机器人对不同移动速度的目标物体跟踪不及时、跟踪落后的情况,设计了一种运动规划模型,以物体预估轨迹的时间长度规划机械臂跟踪轨迹,解决了机器人动态跟踪滞后的问题。系统通过传送带获取目标物的运动状态,以动态跟踪抓取实验来验证多融合轨迹方法的有效性与动态跟踪模型的鲁棒性。

Based on the Robot operating system(ROS) framework and taking the 6-DOF ABB robot as the research object, a robot dynamic tracking and grasping method based on online prediction and planning is designed. The system predicts its running trajectory online by detecting the motion state of the target in advance, and performs motion planning on the estimated trajectory to obtain the tracking trajectory of the manipulator, so as to realize the dynamic tracking and grabbing of the moving target by the robot. Aiming at the problem of intermittent tracking process and discontinuous tracking trajectory caused by the communication feedback mechanism between the ROS system and the robot, a multi-track fusion method is designed to fuse multiple trajectories during motion planning, so that the robot can continuously track moving objects. Aiming at the situation that the robot does not track the target objects with different moving speeds in time and lags behind, a trajectory planning model is designed. The robot arm tracking trajectory is planned based on the estimated trajectory time length of the object, which solves the problem of robot dynamic tracking lag. The system obtains the motion state of the target through the conveyor belt, and uses dynamic tracking and grabbing experiments to verify the effectiveness of the multiple fusion trajectory method and the robustness of the dynamic tracking model.