Integration model of learning mechanism and dialogue strategy based on stochastic experience representation using Bayesian network

We propose a method for personal robots to acquire autonomous behaviors gradually based on interaction between human and robots. In this method, behavior decision models and dialogue control models are integrated using Bayesian networks. This model can treat interaction experiences using statistical processes, and sureness of decision making is represented by certainty factors using stochastic reasoning. The robots not only decide behavior, but also make suggestions to and ask questions of the user using the certainty factors. Consequently, the certainty factors enable the behavior acquisition to be more effective. We investigate the feasibility of this method with obstacle avoidance tasks for mobile robots. Through experiments on a real mobile robot, we have confirmed that the mobile robot acquires robust behavior decision models against changes of environment and uncertainties of sensors, with only a few teaching and learning sessions.