ETH Zurich researchers have trained their quadruped robot ANYmal to pick up a ball and throw it across a field, demonstrating manipulation skills beyond simple walking or running.

The architecture employs two loops: an outer loop using GPT-4 for refining the reward function, and an inner loop for reinforcement learning to train the robot's control system.

A team of researchers from NVIDIA, ETH Zurich, and the University of Toronto open-sourced Orbit, a simulation-based robot learning framework. Orbit includes wrappers for four learning libraries, a sui ...

Trained entirely in simulation using advanced reinforcement learning, the Tesla Bot demonstrates the company’s cutting-edge progress in AI and robotics, capable of translating digital training ...

Figure AI has developed a new humanoid robotic natural walking capability for its humanoid robots, leveraging reinforcement learning (RL) and simulation-based training. This approach enables the ...

Through the Nvidia Humanoid Robot Developer Program, Neura Robotics is using Nvidia Isaac Sim and Isaac Lab capabilities to simulate more robots in parallel using reinforcement learning (a type of ...

As part of the relationship, Agility Robotics (“Agility”) is expanding adoption of NVIDIA Isaac Sim and NVIDIA Isaac Lab robot simulation and learning frameworks to train and test behaviors on ...