Answers are grounded in this paper and cite their sources.
In this research the authors evaluate a new method for control using several prediction models and recognition of movement series. In MOSAIC (MOdule Selection And Identification for Control), which uses a prediction model with several modules as proposed by Wolpert and Kawato (1998), a module that pairs a prediction model which predicts the future state to be controlled and a controller are switched and assembled based on the size of the prediction error in the prediction model. The authors propose a method using MOSAIC to divide continuous time patterns for human or robot movement into their constituent parts as several series of movement elements. Moreover, the authors evaluate a method to recognize movement patterns of another person using one's own module and imitation learning based on this method. From the results of simulations of acrobot control, the authors show that symbolization of movement patterns and imitation learning based on that are possible.
No takes yet. Share an insight, caveat, or question.
Samejima et al. (2006) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: