Abstract:A control method that trains robot to learn motion skills by natural language is explored in this paper, where fuzzy neural networks are used as a general structure of controller for different motion primitives. For any specified motion primitives, the controller acquires knowledge and is trained by instructions of natural language from a supervisor. This might be a more natural way to construct a motion controller. Based on the trained motion primitives, more complex motion tasks can be implemented. Experiments on natural language training for motion primitives have been done on a wheeled robot.
[1] Schaal S. Is imitation learning the route to humanoid robots?. Trends in Cognitive Sciences, 1999,3(6):233-242 [2] Sugeno M, Park G. Learning Based on Linguistic Instructions Using Fuzzy Theory. the J of Japan Society for Fuzzy Theory and Systems, 1992,4(6):1164-1181 [3] Sugeno M, Park G. An Approach to Linguistic Instruction Based Learning and its Application to Helicopter Flight Control. Proc of fifth Int'l Fuzzy Systems Assoc. World Congress'93, 1993,(IFSA'93):1082-1085 [4] Miyamoto A, Goto K, Itoh O, Migita H, Sugeno M. Method for Adjustment Control Rules by Lingusitic Instructions. Japanese Journal of Fuzzy Theory and Systems, Allerton Press, Inc, 1996,8(5):847-861 [5] Mokato FUJII, Takeshi FURUHASHI. A Proposal of Human-Machine Interactive System through Linguistic Instructions Using Fuzzy Classifier System. Nagoya University, Japan [6] Nakaoka K, Furuhashi T, Uchikawa Y. A Study on Apportionment of Credits of Fuzzy Classifier System for Knowledge Acquistion of Large Scale Systems, Proc of third IEEE Int'l Conf on Fuzzy Systems, 1994,(FUZZ-IEEE'94):1797-1800 [7] Furuhashi T, Nakaoka K, Uchikawa Y. A Study on Fuzzy Classifier System for Finding Control Knowledge of Multi-Input System. Journal of Studies in Fuzziness and Soft Computing, 1996,8:489-502 [8] Hasegawa T, Kamei D, Furuhashi T, Uchikawa Y. A Study on Autonomous Understanding of Linguistic Instructions and Behavior Improvement Using Fuzzy Classifier System. 12 th Fuzzy System Symposium in Japan, 1996. 459-462 [9] Fujii M, Hasegawa T, Furuhashi T, Uchikawa Y. A Study on Understanding of Linguistic Instructions and Behavior Improvement. 6th Intelligent System Symposium in Japan, 1996. 185-188 [10] 周志坚, 毛宗源.智能控制中的模糊控制、神经网络控制计遗传算法. 广东自动化与信息工程,1998,19(1) [11] Rober Fuller. Neural Fuzzy Systems. Abo Akademi University [12] Watkins C. Learning from delayed rewards, PhD Thesis, University of Cambridge, England [13] Richard S Sutton, Andrew G Barto. Reinforcement Learning:An Introduction. MIT Press, Cambridge, MA,1998 [14] Sutton R S, Barto A, Williams R. Reinforcement learning is direct adaptive optimal control. Proc of ACC, Bostan, june 1991 [15] 印炅,Java. 与面向对象程序设计教程. 高等教育出版社,1999 [16] Cay S. Horstmann and Gary Cornell. Java 2 核心技术京京工作室译,机械工业出版社,2000 [17] Nuance Communications, Inc. Introduction to Nuance System 1999 [18] Nuance Communications, Inc. SpeechObjects Developer's Guide,1999