Human-Directed Coordinated Control of Assistive Mobile Manipulator

This paper considers the control of a mobile manipulator with the end-effector commanded by a human user. The motivation is to empower mobility impaired individuals to tackle certain activities of daily living. The prototype robotic system consists of a redundant articulated arm mounted on a mobile base. The human user specifies the desired motion of certain degrees of freedom, such as the translation or rotation of the arm end-effector. An optimization-based control algorithm determines the full arm and base motion to minimize deviation from the human intent while satisfying certain equality and inequality constraints. These constraints may involve physical requirements of the system, e.g., collision avoidance and joint limits, and human user consideration, such as intuitiveness of the combined arm and base motion. This constraint optimization problem is strictly convex and may be efficiently solved as a quadratic problem. We demonstrate the effectiveness of the algorithm in simulation and experimentation, using a dual-arm Baxter robot mounted on a wheelchair as a prototype testbed. Human user commands the end-effector of one of the arms using a 3-degree-of-freedom interface, either a sip-puff or a joystick type of device. We demonstrate effective object retrieval and placement in an environment with obstacles.