Design of Spiking Central Pattern Generators for Multiple Locomotion Gaits in Hexapod Robots by Christiansen Grammar Evolution

• Main Authors: Andrés Espinal, Horacio Rostro Gonzalez, Martin Carpio, Erick Israel Guerra-Hernandez, Manuel Ornelas-Rodriguez, Marco Sotelo-Figueroa
• Format: Article
• Language: English
• Published: Frontiers Media S.A. 2016-07-01
• Series: Frontiers in Neurorobotics
• Subjects: central pattern generator; FPGA; Spiking Neural network; SPIKE-Distance; Evolution strategy; Legged robot locomotion
• Online Access: http://journal.frontiersin.org/Journal/10.3389/fnbot.2016.00006/full

This paper deals with the design of Spiking Central Pattern Generators (SCPGs) to achieve locomotion at different frequencies in legged robots and their hardware implementation in a FPGA and validation on a real hexapod robot. Herein, the SCPGs are automatically designed by a Christiansen Grammar Evolution (CGE)-based methodology. It is, the CGE performs a solution for the configurations (synaptic weights and connections) of each neuron in the SCPG. This is carried out through the indirect representation of a candidate solution that evolves to replicate a specific spike train according to a locomotion pattern (gait) by measuring the similarity between the spike train with the SPIKE-Distance to drive the search to a correct configuration. By using this evolutionary approach, several SCPG design specifications can be explicitly added into the fitness function to achieve the SPIKE-distance criteria, such as: looking for SNNs with minimal connectivity or a CPG able to generate different locomotion gaits only by changing the initial input stimuli. The SCPG designs have been successfully implemented on a Spartan 6 FPGA board and a real time validation on a 12 DOFs hexapod robot is presented.