An evolutionary approach to feline rover gait planning

In this paper, a Genetic Algorithm is used to solve for the optimization problem of a quadruped rover, which is developed based on a two-dimensional model of a domestic cat. The rover´s gait, pose and control parameters are optimized for two separate objectives. First, the energy consumption through applied joint torque is minimized for fixed thrust speeds. Secondly, the rover´s thrust speed is maximized. The elimination of impact on the leg at touchdown is considered by using smooth, periodic trajectories for the motion of the toe. The optimum gaits resemble those in nature, and the design variables follow similar trends with increasing speed as found in nature and other works, suggesting that energy minimization may be one of the many objectives of natural evolution.