Pattern-based planning system for complex environments

• Authors: Mahshid S. Ehsaei;Abolfazl T. Haghighat
• Venue: Proceedings of the 2nd International Conference on Interaction Sciences: Information Technology, Culture and Human
• Year: 2009

Coordination among agents in a dynamic uncertain environment to achieve a specified goal is a complicated task. Common approaches such as centralized techniques need high speed computations and are brittle to single points of failure. Distributed approaches have improved these problems, but they can be highly sub-optimal. Market-Based techniques are one of the best approaches and are more responsive than two other techniques to change, but they need high message passing ability. In this paper, taking advantages of Market-Based architecture, we introduce a novel approach called Pattern-Based Planning System (PBPS) that can be used in environments with forcible limitations on message passing. On the other hand, PBPS reduces search space using some patterns as input to generate a plan that determines each agent's tasks and the sequence of execution of those tasks. It also monitors the plan during its execution and updates it whenever is needed. As a test-bed, we chose Robocup 3D soccer simulation environment and implemented our planner in MRL 3D soccer simulation team's agents. The simulation results show that our approach increase the probability of achieving the goal.