Multiple Model-Based Robot Control: Development and Initial Evaluation
Abstract
A new form of adaptive model-based robot control has been developed and experimentally evaluated. The Multiple Model Based Control (MMBC) technique utilizes knowledge of nominal manipulator dynamics and principles of Bayesian estimation to provide payload-independent trajectory tracking accuracy. The MMBC is formed by augmenting a model-based controller, which employs feedforward dynamic compensation and constant gain PD feedback, with a payload estimate provided by a Multiple Model Adaptive Estimator. Extensive simulation studies demonstrated the MMBC's ability to adapt to variations in manipulator payload quickly and accurately. Initial experimental evaluations on the first three links of a PUMA-560 validated the algorithm's potential. Keywords: Algorithms, Case studies, Theses, Closed loop parameter estimation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1988
- Accession Number
- ADA205725
Entities
People
- Larry D. Tellman
Organizations
- Air Force Institute of Technology