Title :
Dynamics classification of underwater robot and introduction to controller adaptation
Author :
Takemura, Yasunori ; Ishii, Kazuo
Author_Institution :
Dept. of Mech. & Electron. Eng., Nippon Bunri Univ., Oita, Japan
Abstract :
Underwater robots and underwater machines are counted on helping the salvage procedure, escaping prevention of pollution, lifesaving at sea, scientific exploration in the ocean and so on. In the underwater robot, therefore, Autonomous Underwater Vehicles (AUVs) have been developed actively during recent year. However, AUVs have various problems which should be solved such as motion control, acquisition of sensors´ information, behavioral decision selflocalization and so on. Regarding to consider about these problems, robot should be learning on selforganizing about relation ship to own status, environment and behaviors. In this paper, a new self-organizing controllers system for AUVs using modular network SOM proposed by Tokunaga et al., is described. The proposed control system is developed using Recurrent Neural Network (RNN) type mnSOM. And, we report that the control system is implemented into the AUV “Twin-Burger”.
Keywords :
mobile robots; neurocontrollers; recurrent neural nets; remotely operated vehicles; robot dynamics; self-adjusting systems; self-organising feature maps; underwater vehicles; autonomous underwater vehicles; controller adaptation; dynamics classification; modular network SOM; recurrent neural network type mnSOM; selforganizing controller system; underwater machines; underwater robot; Artificial intelligence; Underwater vehicles; Adaptive Controller; Underwater robot; modulear network SOM;
Conference_Titel :
World Automation Congress (WAC), 2010
Conference_Location :
Kobe
Print_ISBN :
978-1-4244-9673-0
Electronic_ISBN :
2154-4824