AquaClimber: a limbed swimming and climbing robot based on reduced order models.

Many legged robots have taken insight from animals to run, jump, and climb. Very few, however, have extended the flexibility of limbs to the task of swimming. In this paper, we address the study of multi-modal limbed locomotion by extending our lateral plane reduced order dynamic model of climbing to swimming. Following this, we develop a robot, AquaClimber, which utilizes the model's locomotive style, similar to human freestyle swimming, to propel itself through fluid and to climb vertical walls, as well as transition between the two. A comparison of simulation and model results indicate that the simulation can predict how hand design, arm compliance, and driving frequency affect swimming speed and behavior. Using this reduced order model, we have successfully developed the first limbed aquatic-scansorial multi-modal robot.