QP-based Adaptive-Gains Control to Lower Damage in Humanoid Falls

We address the problem of humanoid falling with a decoupled strategy consisting of a before-impact first stage and an at-impact and post-impact second stage. In the first stage, geometrical reasoning allows the robot to choose appropriate impact points in the surrounding environment and to adopt a posture to reach these target impact points while avoiding impact-singularities. The surrounding falling environment can be unstructured and may contain cluttered obstacles. The second stage uses a quadratic program (QP) that adapt on-line the low-level joint motor proportional-derivative (PD) gains to make the robot compliant –to absorb impact and post-impact dynamics, and lower damage as much as possible. This is done by incorporating the stiffness and damping gains directly as decision variables of the QP along with the usually-considered variables of joint accelerations and contact forces. Constraints of the QP prevent the motors from reaching their torque limits event at the impact peak, therefore protecting the motors from excessive damage caused by the impact. Several experiments in a full-dynamics simulator are presented to illustrate the approach.