An adaptive, self-organizing dynamical system for hierarchical control of bio-inspired locomotion
Central pattern generators (CPGs) are widely used in robotics locomotion control. However, currently most of them are realized by algorithms on digital processor, which may consume high power and occupies a large area of chipṪhus analog circuit realized CPGs have been attracting more and more interests recently. They may reduce power consumption, have smaller sizes, and enable low-cost production. In this paper several such CPGs: 1) Cellular Neural Network (CNN) proposed by Chua and Yang; 2) Nervous (Nv) and Neural (Nu) Neurons proposed by Mark Tilden; 3) an adaptive aVLSI CPG chip by Lewis et al.; and 4) an analog CMOS CPG proposed by Nakada et al. are introduced. Their advantages and disadvantages are compared. Future research direction is proposed at the end of the paper.