Task-based resource allocation for improving the reusability of redundant manipulators

Because of its increased range of resources, a redundant/reconfigurable robot has the potential to be deployed for a variety of applications saving time and money. This work provides decision tools which increase the utility of deployed manipulators by operating them intuitively and economically in terms of a broader set of tasks while maximizing performance. System utility is expanded by developing a general decision making strategy that decreases the burden on operators who have little or no robotics expertise yet must operate the system. This utility function is a Redundancy Resolution Strategy (RRS).; The RRS uses a generalized, parametric system manipulator model and a large set of criteria to compare the capabilities of the infinite number of joint configurations available. The RRS then selects and prioritizes a subset of criteria that are appropriate in a given situation by evaluating critical boundaries associated with manipulator constraints and/or task description. The RRS is paired with a Redundancy Resolution Technique (RRT) which determines, in real-time, the best configuration in terms of the selected criteria. Together, the criteria and these components are the Decision Making System (DMS).; Most DMS components have received attention at the University of Texas and elsewhere, but the RRS and component integration has been largely dismissed as implementation detail. This ‘detail’ is largely responsible for the lack of success in deploying redundant/reconfigurable systems. If the advantages of these systems are to be realized, we must apply physically meaningful criteria to determine the best allocation of redundant resources by creating a procedure to select a subset of appropriate criteria. The selected criteria must improve performance so that a larger number of tasks are possible. The burden on the operator, task completion time, and time required to prepare for a new task all must be reduced. The life of the manipulator is also increased by using resources efficiently. By managing resource allocation correctly, the RRS can also remove the need for intervention by an expensive ‘robotics expert.’ The selected RRT, reviewed performance criteria, and proposed RRS are implemented using existing operational software and used to complete several complex simulated tasks.