Toward biomorphic robotics: A review on swimming central pattern generators

Neuro- and biomorphic approaches in the design of intelligent robotic systems and, more specifically, various technical applications have attracted much attention from researchers and engineers. Biomorphic robotics implies that a machine should be able to reproduce movement and control it the same way animals do in a real-world environment. Fish-like swimming robots seem to be the simplest candidates to reproduce biological mechanics of movement in aquatic medium adhering to the principles of its control and navigation. At the heart of the fish movement control system is its central pattern generator (CPG) located in the spinal cord. This CPG creates a robust rhythmic signal that activates muscles inducing movement in space, i.e. locomotion. The fish actuator system involves body muscles and fins and looks quite simple in comparison with land-walking animals. Hence, it has become the center of attention for many modeling and engineering studies that we review in this article. Many fish-like robots have been developed since rather simple CPG controllers can induce robot swimming. However, existing robotic solutions are still far from natural prototypes in terms of speed performance, power efficiency, and maneuverability. Something seems to be missing in understanding the actuator control principles and hence appropriate CPG design. A tuna fish’s cruising speed of more than a hundred kilometers per hour, and acceleration of dozens of g in pike attacking its prey remain unreachable digits for existing robotic solutions. Along with the development of bionic muscle-like actuators, state-of-art research in this field focuses on finding possible ways of CPG integration with sensorial systems and higher-level brain controllers. Needless to say, a close study of biological fish swimming in terms of its biomechanics and control still raises fundamental questions about how fishes are capable of moving so efficiently. Inertial and dense aquatic medium requires CPG to be highly integrated with sensorial receptor systems. Fish swimming is finely optimized relative to energy loss into fluid turbulence. How this control is organized remains a question. We also review some concepts on how a higher-level of movement control can be incorporated into the intelligent CPG design.