Key points are not available for this paper at this time.
Abstract The stoppage of a mobile platform is generally scheduled to feed parts for machines on production lines, such as fenced industrial robotic manipulators. A non-stop mobile robotic part feeding system can contribute to production efficiency and flexibility but contains several challenging tasks. For example, the industrial robotic manipulator needs to perceive the positions of the mobile robot accurately and robustly before grasping the supplies when the mobile robot moves around. Thus, based on the relative distance between the two robots, an interaction mode of the integrated robotic system consisting of a fixed robotic manipulator and a mobile robot is developed for robotic interaction. In order to accurately and robustly perceive the positions of a mobile robot, two different positioning approaches for the robotic manipulator positioning mobile robot in an indoor environment are utilised. One approach is ultrasonic sensors fused with inertia measurement units (IMU) by extended Kalman filter (EKF). Furthermore, an outlier rejection mechanism is implemented to escape outliers from ultrasonic measurement. Another positioning approach is achieved by detecting an ArUco marker with visual sensor. Lastly, a positioning switching strategy according to the visual sensor state allows the robotic manipulator to reposition the mobile robot seamlessly. According to the static experiments, EKF-based positioning approach fusing IMU with ultrasonic sensor can export high-accuracy (the root mean square error is 0.04 m) and high-precision (the standard deviation is 0.0033 m) in positioning while keeping a high update frequency of 181.9 HZ in static positioning. Evaluations through dynamic experiments demonstrate that the proposed positioning system can suppress the positioning drifts over time in comparison with wheel encoder-based positioning method. The two-stage repositioning strategy can support the robotic manipulator to identify the positions of the mobile robot robustly, even in the case when the visual sensor is occluded.
Yang et al. (Fri,) studied this question.