Mark C. Torrance

Learn More
The construction of a program that generates crossword puzzles is discussed. As in a recent paper by Dechter and Meiri, we make an experimental comparison of various search techniques. The conclusions to which we come differ from theirs in some areas-although we agree that directional arc consistency is better than path-consistency or other forms of(More)
We describe an implemented system which mediates between an un-modiied reactive mobile robot architecture and domain-restricted natural language. We introduce reactive-odometric plans and demonstrate their use in plan recognition, fault-tolerant plan execution, and learning to associate human terms with locations in the environment which are perceptually(More)
Simulators used in research on autonomous mobile robots have been criticized for their tendency to change the nature of the problems the robot control architecture has to solve. In this paper we address those arguments, and find that under certain conditions simulators can be a valuable tool to supplement research with physical robots. We conclude with(More)
The construction of a program that generates crossword puzzles is discussed. The aim of this research is to draw conclusions that apply to conjunctive search generally, and the experimental results obtained in the crossword-puzzle domain lead us to the following: (1) Lookahead is extremely important when solving conjunctive queries, (2) Compile-time control(More)
Systems in which the processing of language happens as part of the action deliberation cycle support interactive and responsive discourse with relative ease. They usefully treat language understanding and production as special kinds of action , and thus are able to interleave these processes with real-time responsiveness to other activity demands. This(More)