Utilizing compliance to manipulate doors with unmodeled constraints

Increasingly, robots are being applied to challenges in human environments such as soldier and disability assistance, household chores, and bomb disposal. To maximize a robot's capabilities within these dynamic and uncertain environments, robots must be able to manipulate objects with unknown constraints, including opening and closing doors, cabinets, and drawers. Practicality suggests that these tasks be done at or near human speed. A simple and low cost method is proposed to achieve these ends - utilizing joint compliance to resolve forces non-tangent to the path of travel. In this paper, joint compliance is achieved by means of a clutch mechanism located in line with the manipulator joint motors. When an object is to be moved, the motors are disengaged from the joints using the clutch, thus allowing the joints to move freely with the object while force is applied by the mobility platform. This enables the robot to move an object within its constraints without the need for a precise forcing vector, minimizing sensing needs as well as computation time. Other implementations of the technique are also possible, including use of inverse dynamics, back-drivable motors, and/or actively controlled slip clutches for gravity and friction compensation. The effectiveness and robustness of this approach are demonstrated through kinematic analysis, dynamic simulation, and physical experimentation on three differently sized doors and a drawer.