Command generation by tree search in artificial intelligence
Investigation is made into treating the noncausal open-loop command generator for a totally automatic controller of the TAFCOS-type (which divides control effort into open-loop and closed-loop parts) as a heuristic tree-search problem in artificial intelligence. Discussion is given of earlier analytic dynamic-optimization and optimal-control approaches. It is the purpose of this work to create a software bed for achieving, in a command generator, heuristic open-loop control actions similar to humans, and thus to relieve downstream ignorant closed-loop controllers of responsibility for dealing with large errors. Emphasis is on an Edisonian - in contrast to a Newtonian - approach. Appendices contain a simple Riccati-equation solver, two different search prototypes in BASIC, and a more general tree=based command generator written in C, with a Users' Guide.
- Research Organization:
- Texas Tech Univ., Lubbock, TX (USA)
- OSTI ID:
- 6093933
- Country of Publication:
- United States
- Language:
- English
Similar Records
United States Army Training and Doctrine Command (TRADOC) - artificial intelligence and robotics symposium
Development of a concurrent tree search program