In a video posted on YouTube hours ago, Boston Dynamics shows how the Atlas robot is advancing in terms of interaction and capabilities.
The all-electric humanoid , known for its agility and dexterity, now incorporates reinforcement learning policies that allow it to perform complex maneuvers with a surprising level of precision.
Using a motion capture suit, Atlas has impressively replicated human movements, a development achieved in collaboration with the Robotics and Artificial Intelligence Institute (RAI).
The company describes it as the world’s most dynamic humanoid robot. Source: Boston Dynamics
Reinforcement learning: key to the Atlas robot’s movements
Reinforcement learning , an advanced technique in artificial intelligence , has been key to this progress. Rather than relying on direct human intervention, Atlas is trained in a physical simulator that generates vast amounts of data, allowing the robot to perfect its movements by repeating more than 150 million simulations .
This allows it to perform tasks such as lifting objects or navigating difficult terrain with dexterity, opening up new possibilities for humanoid robotics .
Collaboration with NVIDIA
In addition to this collaboration with the RAI Institute, Boston Dynamics has expanded its work with NVIDIA to further advance Atlas’s artificial intelligence. Through its partnership with NVIDIA, the Jetson Thor computing platform and Isaac Lab simulation environment provide the tools necessary for Atlas to continue developing complex capabilities.
This partnership is focused on creating more dynamic and functional robots through the use of advanced artificial intelligence, bringing Atlas one step closer to becoming the world’s most capable and adaptable humanoid robot .
Follow us on social media and don’t miss any of our posts!
YouTube LinkedIn Facebook Instagram X (Twitter) TikTok
Source and main photo: Boston Dynamics
Internal photo: Boston Dynamics website
