CPG-based locomotion generation in a Drosophila inspired legged robot

In this paper a CPG approach to locomotion pattern generation and control in a Drosophila-like hexapodal structure was implemented. The CPG consisted of a two-layer network of nonlinear oscillators: the first layer is devoted to guarantee a stable locomotion pattern, whereas the second was designed according to the kinematic constraints of each leg. An accurate leg motion required the synchronization control in parallel neural networks where different oscillation frequencies have to co-exist. A strategy based on the use of an intermittent laplacian coupling was used to solve this issue. Numerical results are reported, which deserve further analytical inspection.