Eye-In-Hand Pose Estimation of Industrial Robots

This paper proposes a pose estimation approach of an industrial six-degree-of-freedom robot without the need of externally placed sensors. An RGB-D camera that is placed at the robots end-effector is combined with a SLAM algorithm to act as a sensor system. The presented system based on the novel integration approach is evaluated by performing test trajectories and is capable of estimating the tool-center point with a standard deviation of 2.6 mm and performing a joint state estimation with a standard deviation of 13.6 mrad without any external sensor.