Title :
Fast suboptimal planning with nexus states
Author :
Briggs, Will ; Dawson, Brad
Author_Institution :
Dept. of Comput. Sci., Lynchburg Coll., VA, USA
Abstract :
Despite recent advances, general-purpose planning still lags behind domain-specific planning in efficiency. We propose a method for automatic incorporation of domain knowledge into general-purpose planning by the identification of nexus, or well-connected, states in the domain. We present an analysis of under what circumstances this method will be applicable, and show that the method significantly improves planning time in two common benchmark domains, albeit with a measurable loss of plan optimality
Keywords :
planning (artificial intelligence); benchmark domains; domain knowledge; general-purpose planning; nexus states; plan optimality; planning; suboptimal planning; Automatic programming; Computer science; Design automation; Educational institutions; Motion planning; Polynomials; Process planning; Robot programming; Robotics and automation; Time measurement;
Conference_Titel :
Tools with Artificial Intelligence, Proceedings of the 13th International Conference on
Conference_Location :
Dallas, TX
Print_ISBN :
0-7695-1417-0
DOI :
10.1109/ICTAI.2001.974459