A Novel Arm-Wrestling Robot Using Motion Dependant Force Control

Abstract

In recent decades, aging population of most advanced countries has increased sharply, and elderly health care has become economically a big social issue. One way to cope with this problem is to introduce various service robots such as an arm-wrestling robot proposed in the article into our daily lives, which are able to promote elderly health and to save elderly welfare cost. For successful execution of many tasks of service robots (Engelberger, 1989) such as robotic arm-wrestling, control of interacting force between a robot and the environment is crucial. Several interacting force control strategies have been developed during the past two decades (Whitney, 1987; Gorinevsky et. al., 1996; Siciliano & Villani, 1999; Natale, 2003), which can be classified into two groups; indirect force control without explicit closure of a force feedback loop, and direct force control with explicit closure of a force feedback loop (Natale, 2003). Indirect force control includes compliance (or stiffness) control (Salisbury & Craig, 1980; Mason, 1981) and impedance control (Hogan, 1985; Anderson & Spong, 1988), in which the position error is related to the contact force through a mechanical compliance or mechanical impedance between the robot and the environment. On the other hand, direct force control includes hybrid position/force control (Raibert & Craig, 1981) and inner/outer motion/force control (Caccavale et al., 2005), in which actual force is measured using a force sensor and then is controlled via direct force feedback. If a detailed model of the environment is available, the hybrid position/force control can control position along a constrained direction and force along an unconstrained direction using a selection matrix. If a detailed model of the environment is not available, inner/outer motion/force control can control force and motion, in which outer force control loop is closed around the inner motion control loop. These force control strategies generally assume that the environment is fixed in space. However, there are applications that the environment being contacted by the robot is no longer fixed in space, but is moving. In this case, contacting force control is much more challenging problem compared to the fixed environment case. For example, in robotic armwrestling considered in this article, the environment should be modeled as a moving object with compliance, viscous friction and inertia.