Pneumatic Muscle Actuator Control
Abstract
This research is concerned with investigating methods for the control of McKibben pneumatic actuators, or pneumatic muscles (PMs). PMs are a novel type of actuator that closely mimic human skeletal muscles in size and power capabilities. PMs are considered by the Air Force for use in exoskeletons to be worn by humans for strength augmentation and for use as actuators in robotic systems. In this research, we investigate adaptive, sliding mode, and soft computing approaches to control of PMs and robotic systems actuated by PMs. The soft computing approaches include neuro-fuzzy modeling of an actual PM in the Human Effectiveness Lab at Wright Patterson Air Force Base, and evolutionary design of a fuzzy PID controller based on this model. We also investigate a type of MIMO fuzzy model predictive control for a planar arm actuated by four PMs. Some of the controllers are tested on the actual PM at WPAFB while others are proven in simulations. A byproduct of this research is an evolutionary fuzzy training algorithm useful for identification of dynamical systems as well as classification problems.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 2004
- Accession Number
- ADA420339
Entities
People
- John H. Lilly
Organizations
- University of Louisville