Walking Speed Control of Planar Bipedal Robot with Phase Control

Abstract

In this study, we aim to provide the dynamic walking of a five-link planar bipedal robot without the need for optimization and regardless of initial conditions. Since bipedal robots are underactuated systems, classical control methods cannot be applied. Joint trajectories must be determined to control the robot. The trajectories that will allow the biped robot to walk can be derived with a trajectory optimization. Periodic stability and dynamic balance are provided by following the trajectories with virtual constraints method. However, determining the optimization-based trajectories before walking is the weakness of the virtual constraints method. As a solution, the method we call “phase control” is newly proposed. Thanks to the phase control method, the need for trajectory optimization is eliminated. Walking speed can be adjusted adaptively depending on user input without interrupting walking. Although the walking speed changes, it is shown with the limit cycle and Poincaré return map that the periodic stability is provided. In addition to mathematical model-based simulation, a physics engine-based Gazebo simulation is also performed to test the feasibility of the method. The advantage of the proposed phase control method is that the planned trajectories can provide limit cycles even if they are not optimized. This advantage provides freedom of trajectory selection that does not require optimization.