Abstract: To build a robot tracking and quality inspection system on roller conveyor,a novel tracking approach based on visual object tracking and real-time guidance was proposed,that was tracking object's 6D pose with camera in real time and processing robot online.The objects were captured by the static camera in real time and a Multi-Object Tracking(MOT)algorithm was proposed to track the locations and rotations of these moving objects.The object's 6D pose and the trajectory of robot were solved in real time.The robot was processed with the real-time trajectory to keep the end-effector coincident with the work point.The experiments showed that the proposed approach could be used on roller conveyor.Compared to the tracking approach based on detection,the novel approach did not accumulate the estimation error of objects during tracking.

Key words: conveyor, robot, visual guidance, multi-object tracking

摘要： 针对滚筒输送线场景下的机器人随线质检问题,提出一种基于视觉跟踪实时引导的传送带跟随方法,即通过定位相机跟踪待测目标的运动,实时引导机器人调整动作跟随运动目标并检测。首先,通过顶部固定相机获取包含待测目标的实时视频,构建多目标跟踪(MOT)算法跟踪待测目标的位置与旋转角;然后,通过图像信息解算目标的6D位姿,进一步解算对应的机器人工作点坐标;最后,根据实时工作点坐标在线调整机器人动作,保持机器人末端与工作点重合,实现跟随检测。实验结果表明,所提方法可实现滚筒输送线场景下的机器人随线检测;相较于现有的基于目标检测的跟随方法,不存在跟随偏差累积问题。

关键词: 传送带, 机器人, 视觉引导, 多目标跟踪