Energy Efficient Quadruped Robot Trotting via Adaptation of Leg Parameters over Unknown Ground

This work presents the concept of a biologically-inspired quadruped robot with under-actuated legs and adaptive parameter updates. The adaptive system is designed to maximize the energy efficiency or equivalently, minimize the cost of transport of the robot while performing prescribed tasks robustly. The stride length of the robot which corresponds to the maximum angular range of motion between the upper leg and the main body of the robot and the stiffness of the translational springs in the lower limbs are adapted using a gradient-descent like algorithm. Simulation results presented in this thesis demonstrate the ability of the algorithm to achieve near-optimal cost of transport without having any information about ground properties. The dynamic stability analysis method of the trotting gait using the inverted pendulum model has been presented.

Major Advisor: V John Mathews
Committee: Xiao Fu
Committee: Jinsub Kim
Committee: Ravi Balasubramanian
GCR: Len Coop