概要

Towards continuous planetary surface exploration by a rover (i.e., space prove), this paper proposed the adaptive learning mechanism based on Genetic Algorithm (GA) which promotes a rover to get out of stuck and immobilized situations through the learning of its appropriate motions. For this propose, we developed the rover with two movable joints stabilizer like an arm, which learns how to use them by the proposed mechanism. In detail, the proposed mechanism explores an appropriate sequence of the movements (i.e., rotations) of the two joints stabilizer to get out of stuck and immobilized situations. To evaluate the effectiveness of the proposed mechanism, we conducted the field experiment of our rover in Black Rock Desert, USA as the planetary exploration. The experimental results found the following implications: (1) our rover can get out of stuck and immobilized situations by an appropriate sequence of the motions of two movable joints, which are found through the learning by the proposed mechanism; and (2) the success rate of getting out of stuck and immobilized situations is 0.95.

論文誌情報

題目: Adaptive Learning Based on Genetic Algorithm for The Rover in Planetary Exploration
著者: Fumito Uwano, Akinori Murata and Keiki Takadama
誌名: Proceedings of the International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS2016)
詳細: Beijing, China, June 2016

Bibtex or Download

@inproceedings{fumito uwano 2016adaptive,
  title={Adaptive Learning Based on Genetic Algorithm for The Rover in Planetary Exploration},
  author={Fumito Uwano and Akinori Murata and Keiki Takadama},
  booktitle={Proceedings of the International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS2016)},
  year={2016},
  month={June}
}