Abstract
It was confirmed that a real mobile robot with a simple visual sensor could learn appropriate motions to reach a target object by direct-vision-based reinforcement learning (RL). In direct-vision-based RL, raw visual sensory signals are put directly into a layered neural network, and then the neural network is trained using back propagation, with the training signal being generated by reinforcement learning. Because of the time-delay in transmitting the visual sensory signals, the actor outputs are trained by the critic output at two time-steps ahead. It was shown that a robot with a simple monochrome visual sensor can learn to reach a target object from scratch without any advance knowledge of this task by direct-vision-based RL.
Similar content being viewed by others
References
Anderson CW (1989) Learning to control an inverted pendulum using neural networks. IEEE Control Syst Mag 9:31–37
Morimoto J, Doya K (2001) Aquisition of stand-up behavior by a real robot using hierarchical reinforcement learning. Robotics Autono Syst 36:37–51
Tesauro GJ (1992) Practical issues in temporal difference learning. Mach Learn 8:257–277
Asada M, Noda S, Tawaratsumida S, et al. (1996) Purposive behavior acquisition for a real robot by vision-based reinforcement learning. Mach Learn 24:279–303
Shibata K, Okabe Y, Ito K (2001) Direct-vision-based reinforcement learning using a layered neural network. For the process from sensors to motors (in Japanese). Transaction of SICE 37(2):168–177
Shibata K, Okabe Y (1997) Reinforcement learning when the visual signals are directly given as inputs. Proceedings of ICNN Houston 3:1716–1720
Shibata K, Sugisaka M, Ito K (2001) Fast and stable learning in direct-vision-based reinforcement learning. Proceedings of AROB 6th'01, Tokyo, pp 200–203
Shibata K, Ito K, Okabe Y (1998) Direct-vision-based reinforcement learning in “Going-to-a-target task” with an obstacle and with a variety of target sizes. Proceedings of NEURAP'98, Marseille, pp 95–102
Barto AG, Sutton RS, Anderson CW (1983) Neuron-like adaptive elements that can solve difficult learning control problems. IEEE Trans SMC 5:835–846
Author information
Authors and Affiliations
Corresponding author
About this article
Cite this article
Iida, M., Sugisaka, M. & Shibata, K. Direct-vision-based reinforcement learning in a real mobile robot. Artif Life Robotics 7, 102–106 (2003). https://doi.org/10.1007/BF02481156
Received:
Accepted:
Issue Date:
DOI: https://doi.org/10.1007/BF02481156