Virtual Constraints Paradigm for Robotic Locomotion Control

Background. Designing highly mobile biologically inspired legged, winged, and limbless robots has advanced considerably in the past decade. Developing motion control algorithms for these robotic systems, however, is mainly carried out using ad-hoc techniques resulting in motions that are far less agile and efficient than the biological species that have inspired the robot morphology. A recent design paradigm, which has been originally introduced for biped robot agile and efficient control [1], is based on controlling robots using virtual constraints (VC). Virtual constraints are relations among the biped generalized coordinates encoding stable walking gaits that can be re-programmed on-the-fly and are enforced via feedback (see Figure 1). Under virtual constraints, bipeds are capable of achieving stable motions over a wide variety of terrain profiles [1]. MABLE, which is the world’s fastest planar biped with knees, employs virtual constraint-based algorithms in its motion control system [2]. This methodology has recently been applied to control of powered prosthetic legs for amputation patients [3], [4].