Development and self-calibration of a robotic visual inspecting system

The paper proposes a high-accuracy, large-scale robotic visual inspecting system which consists of an industrial robot and an optical scanning sensor fixed on the robot hand. Traditionally, calibration of a robotic visual inspecting system is separated into three parts: Hand-eye calibration, robot calibration and robot exterior calibration. Compared with the traditional calibration method, this paper presents a new self-calibration method to calibrate and compensate for the robotic visual inspecting system's kinematic errors. The proposed calibration approach has two unique features: First, it can be calibrated without external measurement devices and human intervention. Second, it simultaneously calibrates the kinematic parameters of the whole inspecting system in one mathematic model to avoid error propagation. These features not only realize the automatic calibration of the inspecting system but also avoid error propagation. Experiments are conducted on a 6 DOF serial robot to validate the good performance of the proposed method.