Map Learning of an Office Conversant Mobile Robot, Jijo-2, by Dialogue-Guided Navigation
We are building a talking mobile robot, Jijo-2, which learns the building map through conversational dialogues. Its application is to aid office works including route guidance, finding vacant meeting room, finding people, etc. This paper proposes the two-layered multiagent architecture to realize both the reactive behaviour and the dialogue-guided route seek. In the upper layer, the talker and navigator behaviours cooperate to build a map which represents actions to reach goal in a probabilistic network.
Unable to display preview. Download preview PDF.
- [Matsui, et al., 95a]T. Matsui, et al., The Office Conversant Learning Mobile-Robot Project, Jijo-2, 13th Robotics Society of Japan Conference, 1995 [in Japanese]. Google Scholar
- [Matsui, et al., 97b]T. Matsui, et al., An Event-Driven Architecture for Controlling Behaviours of the Office Conversant Mibile Robot, Jijo-2, Proc. of IEEE ICRA’97, Albuquerque 1997.Google Scholar
- [Matsui, 97c]T. Matsui, http://ci.etl.go.jp/~matsui/euslisp/
- [Asoh, et al., 96]H. Asoh, et al., Acquiring probabilistic map with dialog-based learning, Proc. of the ROBOLEARN-96,11–18, 1996.Google Scholar
- [Asoh et al., 96]H. Asoh, et al., Combining Probabilistic man and dialog for robust life-long office navigation, Proc. of the IROS’96, 807–812, 1996.Google Scholar
- [Motomura, et al., 96]Y. Motomura, et al., Neural-Baye-sian network learning in the Jijo-2 Robot, Artificial Intelligence Society of Japan, First meeting on Information Integration, 1996.Google Scholar
- [Itou et al., 1993]K. Itou, S. Hayamizu, K. Tanaka, and H. Tanaka. System design, data collection and evaluation of a speech dialogue system. IEICE Transaction on Information and Systems, Vol. E76-D, 121–127, 1993.Google Scholar
- [Torrance, 1994]M.C. Torrance. Natural Communication with Robots, Master Thesis, Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 1994.Google Scholar