An Adaptive Controller System Using mnSOM (2nd Report: Implementation into an Autonomous Underwater Robot)

Abstract

Autonomous underwater vehicles (AUVs) have great advantages for activities in deep oceans [1], and are expected as the attractive tool for underwater development or investigation near future. However, AUVs have various problems which should be solved such as motion control, acquisition of sensors’ information, behavioral decision, navigation without collision, self-localization and so on. Therefore, the AUVs should be autonomous and adaptive to their environment.

In this paper, a new self-organizing decision making system for AUVs using modular network Self-Organizing Map (mnSOM) [4] proposed by Tokunaga et al., is described. The proposed decision making system is developed using recurrent Neural Networks type mnSOM and the efficiency of the system is investigated through the simulations. And, we report that the decision making system is implemented into an autonomous underwater robot “Twin-Burger” (Fig.1)[5].