Simple perception-action strategy based on hierarchical temporal memory

This paper presents a simple strategy for perception-action of robots in indoor environments using Hierarchical Temporal Memory which is the theory of modeling the rationale of the neocortex. The main idea of the present study is that the input of the HTM network is images of objects that robot perceives in environment, and the output of HTM network is action, such as moving along the wall, moving away, opening, and moving forward, etc. Experiments results show that the proposed method can be applied for robot learning and navigation because it imitates humans´ thinking mode to process the information it receives.