A computational model of cognitive development for the motor skill learning from curiosity

Curiosity is a key enabler of autonomous learning for cognitive development of humans. It allows them to learn motor skills through autonomous exploration of the environment. This paper proposes a new neurobiologically-inspired cognitive computational model: C-DCCM (Curiosity-Driven Cognitive Computing Model). The objective of this model is to implement the effect of curiosity that can be applied to the balance learning problem of the two-wheeled self-balancing robot. As basis of the modeling we use and extend the Actor-Critic model. Verification is conducted via simulation on a computing system of sustained dopamine modulation mechanism and its effects on the cerebral cortex. In order to illustrate the effect of the curiosity on the autonomous exploration, some comparison experiments about the balance learning problem have been done.