Perturbation rejection and active fall recovery for biped robots based on the capture point dynamics

This work presents a control procedure to overcome perturbations during biped gait. It exploits the Capture Point dynamics to adapt the gait pattern. This is implemented in two ways. The first is based on modifying the pattern to resist a constant push applied to the hip of the robot, without further changes in the controller. The second procedure introduces an active fall recovery control module that is only triggered when large disturbances are detected. The active fall recovery aims to restore the robot balance by modifying the step position, the center of mass (CoM) height and the step duration. Several simulations were carried out to assess the effectiveness of the controllers to reject perturbations, including pushes and trips. The fall recovery algorithm was proposed as a solution to increase robustness while maintaining energy efficiency. The simulations show that the gait pattern modification is successful to reject a constant perturbation. Moreover, the switch to a fall recovery controller is capable of increasing significantly the robustness of a nominal, energy-efficient gait pattern.

Automated summary

This work presents a control procedure to overcome perturbations during biped gait by utilizing the Capture Point dynamics. It introduces two approaches: modifying the gait pattern to resist a constant push without changing the controller, and implementing an active fall recovery control module triggered by large disturbances. Simulations demonstrate the success of gait pattern modification in rejecting constant perturbations and the significant improvement of robustness by switching to the fall recovery controller.