Simultaneous kinematic calibration, localization, and mapping (SKCLAM) for industrial robot manipulators

Abstract

Recently, the demand for more accurate, productive, and economical robot manipulators is increasing in the robotics industry. However, a manipulator will produce kinematic errors during production. Thus low-cost kinematic calibration is demanded. Moreover, environmental mapping is also demanded to plan the motions of the manipulator. In this paper, we proposed a simultaneous kinematic calibration, localization, and mapping (SKCLAM) method, which can simultaneously calibrate the kinematic parameters of an industrial robot manipulator using a commercial RGB-D camera attached to its end effector to reconstruct its surroundings. In our method, the kinematic calibration is achieved with feature detection and epipolor geometry. Synthetic and real data experiments were conducted to verify the SKCLAM method. We succeeded in reducing the kinematic errors of the manipulator and reconstructing dense 3D maps of the workspace in the experiments.