Learning of multiple robots in quasi-ecosystem

Abstract

This paper deals with multiple robots in quasi-ecosystem. An ecosystem model composed of insects and plants, which are in a relationship of parasitism, is simulated on the discrete cell space. In this ecosystem, the plants become easy to be eliminated as the population size of the insects increases. Consequently, it is required to maintain the species of plants in the quasi-ecosystem. Therefore, multiple robots are introduced to remove some insects from the quasi-ecosystem. However, if the robots eliminate all of insects, the viruses will eliminate the plants owing to diseases. In this ecosystem with complicated relationship, the robots should acquire strategies to maintain plants. In this paper, we apply a neural network for learning a strategy for removing insects. Furthermore, we show several simulation results of behavior learning of robots. Finally, we demonstrate a simple experiment of robots.