Figure 02 Humanoid Robot Learns Natural Walking

Nobody enjoys being around robots that have awkward or unnatural movement. Many companies are working on perfecting their robots’ movements. Figure 02 is now capable of natural walking. An end-to-end neural network, trained with reinforcement learning (RL), for humanoid locomotion, was used to achieve this. These robots have learned to walk like humans with a high fidelity physics simulator.

The physics simulator used for reinforcement learning collected years worth of demonstrations in a few hours. The above video shows this robot in action.

[HT]