Sensing and Representing Dynamic Environments: Judicious Use of Representation Measurably Improves PerformanceReport
A model for integrating perception and action systems operating in dynamic environments is presented, in which the information gathered by a sensor is maintained over time to expand the sensor's effective field of view. The information in the effective fields of view of the sensors is used to determine appropriate actions to achieve the agent's goals. An implementation of the approach in dynamic virtual and physical environments achieves improved performance over purely reactive strategies. The virtual environment is instrumented to quantify the performance improvement We show improved performance for agents using very limited memory structures, and further, that use of additional memory may not improve performance, especially in dynamic environments.
Note: Abstract extracted from PDF file via OCR
All rights reserved (no additional license for public reuse)
Brill, Frank, Glenn Wasson, and Worthy Martin. "Sensing and Representing Dynamic Environments: Judicious Use of Representation Measurably Improves Performance." University of Virginia Dept. of Computer Science Tech Report (1996).
University of Virginia, Department of Computer Science