ABSTRACT
The utility, efficiency, and reliability of legged robots has increased dramatically in recent years. Limbed robots are now capable of locomotion across a variety of terrains, however, achieving both rapid and efficient operation when ground conditions are complex or deformable is still challenging. Resistive terrains such as streams, snow, mud, littoral regions, and tall grass are an important class or set of complex and difficult terrain which are commonly found in the desired operating environments of legged robots. This work presents a reduced-order, dynamic model designed to capture the effect of these environments on the legs of a robot while running. This model, and an experimental platform, are used to evaluate the efficacy of a pair of strategies for adapting running to the inevitable slowing that occurs in resistive terrains. Simulation and experimental results show that intelligent retraction of the foot during flight has a more beneficial effect on the maximum achievable velocity and cost of transport of the runner than a 'punting gait' for a range of fluid depths. However, this performance gap became much smaller in deep fluids suggesting that fluid depth may drive transition from a foot retraction gait to a punting gait.
