Control of an anthropomorphic manipulator involved in physical human-robot interaction

Abstract

The objective of the dissertation is to flexibly control the end effector velocity of a redundant 7-DOF manipulator by using a differential kinematics approach, while ensuring the safety of the robotic arm from exceeding the physical limits of joints in terms of position, velocity and acceleration. The thesis also contributes with a real-time obstacle avoidance strategy for controlling anthropomorphic robotic arms in dynamic obstacle environments, taking account of sudden appearances or disappearances of mobile obstacles. A method for compensating force errors due to changes in the orientation of end effectors, independent from structures of force sensors, is developed to achieve high accuracy in force control applications. A novel method, the Virtual Elastic System, is proposed to control mobile manipulators for physical Human-Robot Interaction (pHRI) tasks in dynamic environments, which enables the combination of an Inverse Differential Kinematics for redundant robotic arms and a Dynamical Systems approach for nonholonomic mobile platforms. Experiments with a 7-DOF robotic arm, side-mounted on a nonholonomic mobile platform, are presented with the whole robot obstacle avoidance, proving the efficiency of the developed method in pHRI scenarios, more specifically, cooperative human-robot object transportation tasks in dynamic environments. Extensions of the method for other mobile manipulators with holonomic mobile platforms or higher degrees of freedom manipulators are also demonstrated through simulations.