M.Sc Thesis


M.Sc StudentNatan Shemer
SubjectMinimalistic Control Strategies for Damped and Undamped
SLIP-Based Dynamic Running Robots
DepartmentDepartment of Robotics and Autonomous System
Supervisor Assistant Professor Degani Amir
Full Thesis textFull thesis text - English Version


Abstract

The ability of legged locomotion to traverse rough terrain is incomparable to the limited ability of wheeled and tracked platforms.  The Spring Loaded Inverted Pendulum (SLIP) model attempts to understand legged dynamics. The model consists of a point mass and a springy leg.

A known method to increase the robot stability to terrain height variation for the un-damped SLIP model is a Swing Leg Retraction (SLR) method. The leg angle is retracted starting from the flight apex. This had been demonstrated both numerically and experimentally.

In this seminar we begin with the un-damped problem formulation, we will present an instantaneous SLIP model which is a simplified SLIP model. Using this simplified model we present an analytical formulation of the SLR method. This formulation enables us to find optimal parameters for the SLR method. We validate this controller both at simulations and experiments.  Although the SLR method is dramatically more robust then constant leg angle schemes it still did not prove to be robust enough in experiments.

Therefore a new controller is suggested that takes into account damping, using the same methodology that originated the SLR for the un-damped model.  We find that the leg angle should be constant whereas the amount of energy inserted into the system should be varied at a polynomial fashion.  We validate this controller with simulations and experiments and show the high robustness of this simple control method.