Robots get damaged and fall in the real world all the time. Just like humans, they can learn how to use their surroundings to avoid that. This video shows how humanoid robots can exploit a wall to avoid falling. D-Reflex is a method that learns from a neural network to choose a contact position based on “wall orientation, the wall distance, and the posture of the robot.”
As the researchers explain, contact position is used by a whole-body controller to reach a stable posture. The above video shows a TALOS robot using this approach to avoid more than 75% of avoidable walls.