部分的初期化によるParticle Swarm Optimizationの拡張（第2報，マルチロボットシステム問題における検証）

概要

論文の詳細を見る
This paper investigates a particle swarm optimization (PSO) approach to a multi-robot system. PSO is known to be relatively implementable with ease, and efficient for solving various optimization problems. However, the inefficiency for large scaled multimodal problems has also been reported. In order to overcome this unwanted characteristics of PSO, we have been developing a noble method that performs the partial initialization in a small probability and demonstrating its basic effectiveness to benchmark functions. Further discussion on a real world problem is conducted in this paper. We tackle the control of autonomous mobile robots by PSO. Similar to typical methods of evolutionary robotics, artificial neural networks (ANN) is employed as a controller of each robot. PSO is utilized as a methodology for designing ANNs. To examine our approach, computer simulations are conducted using a cooperative box-pushing task. The results obtained are compared with a standard PSO results. It is found that our proposed approach outperforms a conventional one.