Modeling and control of locomotion in complex environments
MetadataShow full item record
In this dissertation, we developed predictive models for legged and limbless locomotion on dry, homogeneous granular media. The vertical plane Resistive Force Theory (RFT) for frictional granular fluids accurately predicted the reaction forces on intruders (with complex geometries) translating and rotating at low speeds ( < 0.5 m/s). Using RFT and multibody simulation, we predicted the forward moving speed of legged robots. During the locomotion of lightweight robots and animals where instantaneous limb penetration speed can reach values greater than ~0.5 m/s, a Discrete Element Method (DEM) simulation was developed to capture the limb-ground interaction. We demonstrated that hydrodynamic-like forces generated by accelerated particles can balance the robot weight and inertia, and promote the rapid movement on granular media. Forces from the environment can not only determine locomotion dynamics, but also affect the locomotion strategy. We studied and simulated the limbless locomotion of snakes in a heterogeneous environment and demonstrated how touch sensing was used to adjust the movement pattern. In heterogeneous environments, the long-term locomotion dynamics is also poorly understood. We presented a theory for transport and diffusion in such settings.