摘要：Human intelligence is deeply dependent on its physical embodiment, and its development requires interaction between its own body and surrounding environment. However, it is still an open problem that how we can integrate the lower level motor control and a higher level symbol manipulation system. One of our research goals is to make a computational model of human intelligence from the motor control to the higher level symbol manipulation. To this end, we propose a robot motor control learning as the first step in this paper. The method is based on HMMs (Hidden Markov Models). The robot moves its arm randomly by changing torques of joint angles and obtains the pose of its arm. The HMM uses state space for representing the relationship between joint torques and pose of the arm by segmenting the obtained sensory-motor information autonomously. The robot can gradually learn to move its arm to a specific position by planning the torque sequence using the learned model. Moreover, we also discuss a future plan for the ultimate goal. We are planning to probabilistically integrate the proposed motor control HMM and the language acquisition model, which has already been proposed by the authors. In this paper, we describe an overview of the integrated model with some important building blocks for our future plan.
