"Why Do You Keep Dropping and Closing the Lid?"... Hyundai Robot Unfazed by Human Interference

"Responds Autonomously to Unexpected Situations"
Boston Dynamics Releases New Video
Atlas Operates with AI Intelligence

Hyundai Motor Group's robot-specialized affiliate, Boston Dynamics, on the 20th (local time) released a video showing the humanoid robot Atlas intelligently responding and solving problems on its own even in unexpected situations.

"Why Do You Keep Dropping and Closing the Lid?"... Hyundai Robot Unfazed by Human Interference

Boston Dynamics applied the Large Behavior Model (LBM), co-developed with Toyota Research Institute (TRI), to Atlas, enabling it to make decisions and move naturally like a human.

Last year, Atlas successfully performed the task of moving engine cover parts to a mobile storage container. In the latest video, Atlas demonstrated moving the robot dog Spot's parts to a loading area or shelf. Recently, humanoid robots have been rapidly advancing by combining sophisticated artificial intelligence (AI) technology, raising expectations that they can assist with repetitive tasks in real industrial settings and improve productivity.

Hyundai Motor Group's robot-specialized affiliate Boston Dynamics released a video on the 20th (local time) showing the humanoid robot Atlas intelligently responding and solving problems on its own even in unexpected situations. Hyundai Motor

In this video, Atlas showcased full-body movements such as walking, squatting, and lifting objects in order to perform tasks like sorting and organizing parts.

What stands out in particular is Atlas's ability to solve problems on its own. While Atlas was transferring parts from one box to another, a researcher interfered by closing the lid of the parts box or dropping a part next to the box. However, Atlas calmly opened the lid or picked up the dropped part and placed it accurately in the box.

Atlas also lifted Spot's leg part, folded it, and precisely aligned it on the shelf. It demonstrated placing another part in the box on the lowest shelf. When Atlas determined that a part was caught on the shelf and could not be placed directly into the box, it smartly pulled the box forward to load the part and then returned the box to its original position.

The LBM applied to Atlas utilizes an end-to-end approach, allowing the robot to quickly learn how to handle various types of objects and make autonomous decisions and controls without having to modify the development code each time.

In particular, while early robots struggled to respond immediately to problematic situations, Atlas's strength lies in its ability to successfully complete assigned tasks by leveraging learning experiences without changing its algorithms or hardware.

LBM is an AI model being researched so that robots can autonomously make decisions and act like humans by learning from large-scale data collected through sensors, including text, images, and videos. Through this, it is known that the robot's motion prediction performance can be improved, allowing the robot to operate up to about twice as fast, even without additional training processes.

Boston Dynamics stated that Atlas is being trained to skillfully handle irregular objects, such as tying knots in ropes or spreading out messy blankets. The company explained that this demonstrates the potential for Atlas to serve as an assistant not only in production sites but also in households.

Boston Dynamics and TRI plan to continue their research on organically integrating humanoid robots that use their entire bodies, such as for advanced manipulation abilities and dynamic movements, with LBM.

Scott Kuindersma, Head of Robotics Research at Boston Dynamics, said, "This video is an example of how general-purpose robots could change our daily lives and work," adding, "Single neural network training for various manipulation tasks will not only drive the advancement of general-purpose robots, but also lay the foundation for high-performance robots like Atlas to use their entire bodies with precision and flexibility."

Meanwhile, in October last year, Boston Dynamics announced plans to accelerate the development of general-purpose humanoid robots in collaboration with TRI, and to cooperate in areas such as human-robot interaction and safety. Boston Dynamics is working to enhance its technological capabilities by utilizing NVIDIA's high-performance robotics chips to advance humanoid robots, and by collaborating with the Robotics and AI Institute (RAI) to strengthen reinforcement learning-based robot AI research.