Boston Dynamics is becoming increasingly autonomous from people

Boston Dynamics continues to implement new algorithmic precision to Humanoid robot Atlas can gain autonomy and not require human intervention.

This time, Atlas uses machine learning and advanced sensors to perform sorting tasks in a simulated factory environment.

The task of moving engine caps “between supplier containers and a mobile sequencing platform” isn’t particularly exciting, but it does demonstrate how some of the new robot’s capabilities work together. Given only “a list of container locations for moving parts,” Atlas uses various sensors and machine learning models to determine the location of the containers and how you need to manipulate your body, arms, and three-fingered hands to grab them. . and move the pieces inside them.

At one point in the video, one part is positioned too high to easily fit into one of the compartments of the sequencing cart. When the robot encounters resistance, it removes the part and then re-evaluates its orientation and position before making adjustments and trying again successfully.