This research uses new developments in redundancy resolution and real-time capability analysis to improve the ability of an articulated arm to satisfy task constraints. Task constraints are specified using numerical values of position, velocity, force, and accuracy. Inherent in the definition of task constraints is the number of output constraints that the system needs to satisfy. The relationship of this with the input space (degrees of freedom) defines the ability to optimize manipulator performance. This is done through a Task-Based Redundancy Resolution (TBRR) scheme that uses the extra resources to find a solution that avoids system constraints (joint limits, singularities, etc.) and satisfies task constraints. To avoid system constraints, we use well-understood criteria associated with the constraints. For task requirements, the robot capabilities are estimated based on kinematic and dynamic manipulability analyses. We then compare the robot capabilities with the user-specified requirement values. This eliminates a confusing chore of selecting a proper set of performance criteria for a task at hand. The breakthrough of this approach lies in the fact that it continuously evaluates the relationship between task constraints and system resources, and when possible, improves system performance. This makes it equally applicable to redundant and non-redundant systems. The scheme is implemented using an object-oriented operational software framework and its effectiveness is demonstrated in computer simulations of a 10-DOF manipulator.

This content is only available via PDF.
You do not currently have access to this content.