Adaptive Control, Learning and Cost Effective Sensor Systems for Robotics or Advanced Automation Systems. First Annual Report, March 31, 1983-March 31, 1984

摘要

The objectives of the research are to: (1) explore the use of adaptive-learning control systems with complex tasks, usually done by humans, which involve modeling and control in the presence of noisy data and incomplete and time-varying models; (2) determine the relevance of the approach to research on robotic learning systems; and (3) determine a focus for testing these theories both analytically and experimentally. The report discusses the intellectual argument for the adaptive-learning control approach to this class of problems, the detail work on the grinding or snagging problem, and plans for future research.