Parameter identification for physical robot-environment interaction using internal and external sensors
Keywords:
industrial robots, human-robot interaction, identification of physical interaction pa-rameters, force-moment sensing, robot-environment interactionAbstract
Introduction: The development of a robotic system in which a robot and a human jointly solve a given task in a safe and efficient way is one of the key challenges in the field of industrial robotics. In a collaborative robotic workcell direct physical interactions between robots, humans and the environment are expected. Purpose: To develop a method for the identification of physical interaction parameters including the force and the point of its application. Results: We propose an analytical algorithm for the interaction parameter estimation using the measurement data obtained from the internal torque sensors in robot joints and the external depth sensor. The algorithm is based on the extension of the static equilibrium equations, which makes it possible to find the desired interaction force and the line of its action. This general solution is then combined with geometric constraints describing the manipulator surfaces and the corresponding friction cones. Particular attention is paid to singular cases that arise when the interaction force action line intersects one or several sensor axes, which causes multiple solutions. Practical relevance: The key feature of the proposed algorithm is its analytical nature, which makes it possible to significantly reduce the computation time as compared to conventional approaches based on direct optimization methods. Thus, the proposed algorithm is well suited for real industrial applications where response time is critical for safety reasons. In addition, the algorithm is able to estimate the required parameters even in some complex cases with many possible solutions while the existing methods usually ignore such cases. The results obtained show that it is extremely difficult to identify the parameters of robot-environment interaction using external sensors only, but their use can significantly reduce the uncertainty of the results obtained using the internal sensors of the robot in singular cases.