Evolutionary Synthesis of Structure and Control for Locomotion Systems

This paper presents a global design approach for self-reconfigurable locomotion systems based on evaluation by dynamic simulation and optimization by artificial evolution. The main objective of this approach is to obtain fully integrated robotic solutions in term of morphology (topology and kinematics) and control (architecture and command). To achieve this goal, a modular robotic system is designed, including modules design and topology representation. Both topology and control are to be co-evolved through an evolutionary algorithm that accounts for the technological constraints (with precise module design) and multi-objective robotic missions (with detailed realistic simulation). Several design examples of modular systems are studied (rover and snake) and the autonomous reconfiguration ability is shown through simulation from rover to snake. The snake is simulated using the ODE tool showing how the robot actually walks (or crawls). The robot is feedback controlled in velocity to perform a snake-like crawling. Results of behavior and desired and simulated velocities validate the design method for locomotion systems.