Learning anticipatory motor control

Recent psychological studies raise the issue of understanding anticipatory motor control and the place of social intention as a possible component of motor intentions [1], [2]. Motor anticipation should play an important role for a robot involved in interactive behaviors. It is a prerequisite to coherent motor primitives and gives access to physical compliance, a necessary condition for sharing safely the workspace of other robots or humans. Moreover, biologically-inspired robotic models studying how social intention can be traced from motor intention are interesting in the debate about the motor theory of social cognition. In this paper, we present an architecture for learning anticipatory motor control. The model is based on functions of brain structures known to be involved in motor control : the motor cortex, the hippocampus, the cerebellum and the striatum (Fig. 1).