A model matching framework for the synthesis of series elastic actuator impedance control

The fundamental goal of robot impedance control is to shape a given system´s behavior to match that of a predefined desired dynamic model. A variety of techniques are used throughout the literature to achieve this goal, but in practice, most robots ultimately rely on straightforward architectures akin to PD control that have intuitive physical interpretations convenient for the control designer. This is particularly true of systems employing series elastic actuators (SEAs) in spite of the potential that more complex controllers have for improving impedance rendering in devices with higher order dynamics. The model matching framework presented here leverages H_∞ control approaches, that are yet to gain widespread use in the robotics community, to significantly simplify the impedance control design task. This framework provides a novel means by which to synthesize a dynamic feedback controller for an SEA that accommodates a wide range of desired impedances and available feedback. The ease of employing this synthesis approach and its potential benefits for SEA control are discussed in light of the limitations of other existing techniques. This discussion, and the insight gained from a series of simulations comparing impedance controllers designed using established passivity-based techniques to controllers born out of our model matching framework, lay the foundation for further adoption of H_∞ synthesis in SEA control.