A portable stand-alone bi-hemispherical neuronal network model of the cerebellum for adaptive robot control

Development of computational models of the brain is relevant not only for deepening our understanding of the biological system but also for potential applications to various engineering problems. In this paper the implementation of a bi-hemispherical neuronal network model of the cerebellum (biCNN) in a stand-alone, portable real time (RT) device is presented. The biCNN is tested during a control engineering application, namely, control of a highly unstable two-wheel balancing robot. The RT device considered is the National Instruments myRIO-1900, which provides flexibility and portability to the biCNN. Execution times obtained with the RT device are compared with a personal computer implementation as reference. The results demonstrate the suitability of the RT implementation of the biCNN for robot control, and provide a successful bridge between the cerebellar research and engineering.