Solving Rubik’s Cube with a robot hand
OpenAI Blog
A robot hand trained with neural networks and reinforcement learning can solve a Rubik's Cube and adapt to unexpected interference like being poked by objects. The system was trained entirely in simulation using a technique called Automatic Domain Randomization, which allows it to handle situations absent from training data. This demonstrates that reinforcement learning can transfer from virtual environments to physical manipulation tasks requiring fine motor control.
Why it matters
We’ve trained a pair of neural networks to solve the Rubik’s Cube with a human-like robot hand. The neural networks are trained entirely in simulation, using the same reinforcement learning code as OpenAI Five paired with a new technique called Automatic Domain Randomization (ADR). The system can handle situations it never saw during training, such as being prodded by a stuffed giraffe. This shows that reinforcement learning isn’t just a tool for virtual tasks, but can solve physical-world problems requiring unprecedented dexterity.
