CPG-Based oscillator networks are increasingly being used to drive multi-limbed robots. To produce effective gaits with these networks, the relationship between the CPG parameters and the characteristics of the gait must be determined. However, due to the nonlinear nature of the oscillators, this relationship is challenging to ascertain. In this work a reinforcement learning algorithm is used to determine the CPG parameters that produce propulsively beneficial kinematics in a multi-fin underwater robot. Due to the high computational cost in creating high fidelity simulations of underwater systems, an alternate method using a low fidelity simulation is explored. To better simulate the dynamics of a two-finned swimming robot a thorough force sweep is conducted on the subject robot in a controlled environment. The resulting force data is used as the dynamic information in a simple simulation. This method allows for the learning of CPG weight settings that produce desired kinematic operating conditions and their resulting forces in simulation. Using this method, when the learned CPG parameters were applied directly to the physical robot, the robot executed the same desired kinematics and forces as expected from simulation with no additional learning needed.

The remarkable swimming abilities of bony fish are the result of multiple interacting subsystems, each tuned to perform certain roles. These subsystems, which include the statically unstable body, multiple highly actuated fins, oscillatory neural controllers, and distributed senses, are not often studied as mutually dependent systems. This research program is developing biorobotic models of these systems and integrating the systems into a biorobotic fish so that interdependencies can be explored during free swimming. The robot body was derived from a bluegill sunfish, and has a tunable mass distribution and a mixture of rigid and flexible sections so that dynamical characteristics of the fish body can be explored. Five highly deformable fins have structural properties scaled to those of biological fins and can create gait patterns for steady swimming and maneuvers. A first generation artificial CPG has been programmed for each fin on a network of five low power microcontrollers. Finally, a dedicated biorobotic pectoral fin has been developed and instrumented with distributed sensory systems so relevant physical (e.g., fin curvature) and hydrodynamic (e.g., pressure) data can be identified and used to predict fin force for closed loop control.