Combined Torque and Velocity Control of a Redundant Robot System

Abstract

One of the most important features of the next generation of robots is the mobility and the ability to work in an unstructured environment. Conventional manipulator arms have bounded workspace and are thus appropriate only for some limited tasks. Hence, they need additional devices such as conveyor belts and handling devices in order to perform a task requiring large workspace. A new promising concept is to place a manipulator arm on a mobile platform. Such system is usually called mobile manipulator. This adds extra degrees of freedom to the system, which makes the mobile manipulator more dexterous and it may become redundant. Redundancy is an important feature of the new generation of robots making them more versatile. For example, a redundant robot can avoid obstacles in the workspace while executing given task or it can optimize joint torques without modifying primary task, which is usually position and/or force tracking of the robot tool mounted on the top of the robot arm. Mobile manipulators typically consist of a robot manipulator mounted on a mobile platform. Typically, the workspace of the fixed base manipulators is limited but they have good accuracy and fast dynamics. On the contrary, mobile platforms have an “infinite” workspace but they are slow and inaccurate, they can not perform any task by itself. Integration of a fast and accurate manipulator and a platform results in a mobile manipulator that should integrate good properties of both subsystems. Hence, the mobile manipulator has large workspace and high dynamics. Its accuracy is comparable to the fixed manipulator accuracy using appropriate sensors and appropriate control algorithm. Due to the remarkable properties mobile manipulators are often used as service robots that help humans in everyday jobs in everyday environments. When working in unknown environment robot autonomy is crucial. The key feature to assure autonomy is the obstacle avoidance. Therefore, appropriate sensors are needed. Compliant behavior is essential when a robot is in contact with the environment (Asada and Slotine [1986]), e.g. in robot assembling, grinding, driving a screw etc. The usual approaches to compliant motion control are the impedance control (Hogan [1985], Salisbury [1980]), the dynamic hybrid control (Raibert and Craig [1981]) and the (resolved) acceleration control (Luh et al. [1980]). The impedance control does not control forces or positions directly, but it controls the desired dynamics of the end-effector or its stiffness. When the dynamic hybrid control is used, the position is controlled along selected directions, while forces are