Balance Control of Robots with Point Feet

IROS 2020

This project applies the balance theory development by Prof. Roy Featherstone to a quadruped robot balancing on two point feet.

The proposed strategy consists in constructing a kinematic equivalence between the quadruped robot balancing on two point-feet in 3D space, and a virtual underactuated inverted pendulum with a flywheel balancing in a 2D plane. An angular-based balance control law is designed for the virtual model, and the resulting momentum is translated to the quadruped through the aforementioned kinematic map. This is paired with a control framework known as the Reactive Controller Framework (RCF) to achieve both balance and locomotion on two point-feet. This is showcased in the challenging task resembling tightrope walking.

The following video shows the results of our balance controller (Gonzalez et al., 2020):

Physical experiments of balance controller on HyQ and simulations showing versatility to also perform line walking locomotion.

References

2020

  1. IROS
    hyq-line-walk.gif
    Line walking and balancing for legged robots with point feet
    Carlos Gonzalez, Victor Barasuol, Marco Frigerio, and 3 more authors
    In IEEE/RSJ International Conference on Intelligent Robots and Systems, 2020