Speed Control on a Hexapodal Robot Driven by a CNN-CPG Structure

Locomotion control in legged robots is an interesting research field that can take inspiration from biology to design innovative bio-inspired control systems. Central Pattern Generators (CPGs) are well known neural structures devoted to generate activation signals to allow a coordinated movement in living beings. Looking in particular in the insect world, and taking as a source of inspiration the Drosophila melanogaster, a hierarchical architecturemainly developedwithin the paradigmof a Cellular non-linear Network (CNN) has been developed and applied to control locomotion in a fruit fly-inspired simulated hexapod robot. The modeled neural structure is able to show different locomotion gaits depending on the phase locking among the neurons responsible for the motor activities at the level of the leg joints and theoretical consideration about the generated pattern stability are discussed. Moreover the phase synchronization, altering the locomotion, can be used to modify the speed of the robot that can be controlled using a reference signal. To find the suitable transitions among patterns of coordinated movements, a reward-based learning process has been considered. Simulation results obtained in a dynamical environment are reported analyzing the performance of the system.