Action understanding using an adaptive Liquid State Machine based on environmental ambiguity

Recently, humans-robots interaction steps aside the traditional master/slave relationship to evolve in a new paradigm of cognitive robotics. Their conception requires the comprehension of human cognitive functions and how they develop. In this paper, we present how an adaptive Liquid State Machine using environment ambiguity may lead to a better emergence of action prediction abilities in a simple robot. The simulation results indicate the efficiency of the proposed method by which a simple robot develop its own prediction capability. These results are promising towards building robots able to develop more complicated capability such as understanding others' intention and further, cooperation with other agents.