A reinforcement learning framework for long-horizon dexterous manipulation, enabling robots to stably execute multi-step, high-precision tasks in real-world environments.

A robust vision-language-action model designed to be generalizable and capable of executing long-horizon and dexterous tasks.

ByteDexter is a dexterous robotic hand with 21 active degrees of freedom (DoF), fitted with highly sensitive tactile sensors. It can reproduce complex human hand movements in real time with high fidelity, while enabling seamless hand-arm coordination.