The Ascento generation you see in the video has more degrees of freedom than the famous Ascento Pro. This mechanical freedom combined with our Reinforcement Learning (RL) algorithm allows Ascento to stand up from all positions, including from the front, back and side. It learns to stand up from over 80 millions simulations, where the robot is placed in millions of different random laying positions. It also learns to stand up over ramps, steps and rough terrain, to be able to cope with real world terrains.

If you want to see any specific challenging stand up, let us know and we will post videos of those.