This project establishes a sound engineering framework for designing next-generation robots and intelligent robotic devices that can be used to perform difficult manipulation tasks quickly and reliably. The framework is useful both for traditional robots already installed in manufacturing plants and future "soft robots" that will make human-robot collaboration safe. The focus is on constrained manipulation problems such as assembly and repair tasks, many of which are dull, difficult, and dangerous. The results will allow robots to be more effectively used in a wide range of domains such as service, healthcare, space exploration, and manufacturing.
The project yields a means to simultaneously control both a robot's position and its elastic behavior using kinematically redundant arms driven by actuators with controllable stiffness. It will be demonstrated that proper design of the robot path plan and admittance plan will ensure that constraints imposed by manipulation tasks are accommodated and high-level task objectives are reliably satisfied despite uncertainies (e.g., the door is opened, the assembly is completed properly). The four planned major outcomes of this research program are: 1) general procedures for selecting the optimal manipulation plan (identifying the optimal motion and admittance plans) using numerical procedures to ensure that the specified manipulation task progresses quickly, appropriately, and reliably despite uncertainties, 2) procedures for realizing a specified admittance plan involving the use of a redundant serial manipulator with variable impedance actuation (a series elastic actuator with controllable stiffness) at each joint, 3) a demonstration of improved robot dexterity in constrained manipulation tasks using redundant serial manipulators with variable impedance actuation to verify the manipulation plan selection methods and validate the overall manipulation approach, and 4) a library of constrained generic task categories with associated optimal manipulation plans.
