Planning in an Artificial Ecosystem
スポンサーリンク
概要
- 論文の詳細を見る
Doing optimal planning in a competitive environment of multiple agents is mostly a highly dynamical process which can be characterized as a tradeoff between time and costs. We refer to this as a tradeoff between deliberation and reactiveness. Deliberative planning corresponds to cost minimization for the plan execution and reactive planning to a guarantee of plan execution. Goal-driven selfish agents compete for resources (e.g. food) in an ever changing dynamic environment. In order to reach those resources agents have to do planning. Agents have different planning characteristics between extremly deliberative and extremly reactive. Very deliborative agents might be confronted with a changed environment when executing their plan while very reactive agents might have too high plan execution costs which can inhibit plan execution. Executing a plan is connected with costs. Fulfilling a goal gives a payoff to an agent. Even taking no actions is costly for the agents although on a much lower scale.
- 一般社団法人情報処理学会の論文
- 1997-09-24
著者
-
Tani Jun
Sony Computer Science Laboratory Inc.
-
Baczewski Joachim
Department Of Administration Engineering Faculty Of Science And Technology Keio University
関連論文
- Evolving Planning Behaviors in an Artificial Ecosystem
- An Evolutionary Approach to Balancing Deliberation and Reactiveness in a Multi-Agent Scenario
- Planning in an Artificial Ecosystem
- Self-organizing Internal Representation in Learning of Navigation: A Physical Experiment by the Mobile Robot YAMABICO
- Embedding Symbolic Process into Deterministic Chaos