Generating and Converging to Surfaces of Gaits

Motivation

The passive Compass Gait Walker (CGW) and other passive walkers exhibit perfect efficiency in a theoretical setting. While modeling and simulating these walkers through hybrid dynamics is fascinating, the greater challenge lies in translating these theoretical concepts into real-world applications. A key question is how we can leverage machine learning models to align real-world walkers with theoretical models and implement advanced controllers, such as Nonlinear Model Predictive Control (NMPC), to stabilize real-world walkers and achieve theoretical gait cycles.