A Flexible Grasping Policy Based on Simple Robot-Camera Calibration and Pose Repeatability of Arm

Manipulation has been a rising focus of robotics research, from industrial automation to personal service robots. Most existing control method are based on accurate kinematic modelling of robot manipulators and robot calibration that needs trained personnel, laboratory environment and data collection is time-consuming. However, robots, especially service robots, similar to other mechanical devices can be affected by slight changes or drifts caused by wear of parts, dimensional drifts, and tolerances, and all of them need a new calibration. Most methods lack robustness due to these slight changes. In this work, for mobile robots, we propose a convenient robot-camera calibration approach and a flexible control approach based on simple robot sensor calibration and pose repeatability of arm. In prototype grasping experiments on the robot Kejia, the approach is validated to be effective and robust, achieving a high success rate of grasping, and using the same control policy, we achieve 1st place in Manipulation and Object Recognition of RoboCup@Home League 2016, here Manipulation and Object Recognition is a sub challenge of Home League.