Title :
Evolutionary agent planning
Author :
Baczewski, Joachim ; Tani, Jun
Author_Institution :
Dept. of Adm. Eng., Keio Univ., Yokohama, Japan
Abstract :
Optimal planning in an environment of multiple agents is mostly a highly dynamical process which can be characterized as a tradeoff between time and costs. This tradeoff can reach critical values and lead to a dilemma. We investigate this situation by using an ecosystem consisting of multiple agents which have to execute plans in order to fulfil their goals. Because resources are limited competition emerges. We propose a potential field method which biases the planning characteristics of agents towards deliberation or reactiveness by a process of evolution. Finally we show preliminary simulation results and discuss the future direction of this work
Keywords :
cellular automata; cooperative systems; knowledge based systems; optimisation; planning (artificial intelligence); search problems; simulation; software agents; competition; costs; deliberation; dynamical process; ecosystem; evolution; evolutionary agent planning; multiple agents; optimal planning; plan execution; planning characteristic bias; potential field method; reactiveness; simulation; time; Computer science; Cost function; Delay effects; Ecosystems; Intelligent agent; Process planning; Real time systems; Robots; Stability; Stock markets;
Conference_Titel :
Evolutionary Computation, 1996., Proceedings of IEEE International Conference on
Conference_Location :
Nagoya
Print_ISBN :
0-7803-2902-3
DOI :
10.1109/ICEC.1996.542695
