IT Home reported on October 31 that Boston Dynamics released a demonstration video of its latest version of the humanoid robot Atlas, showing the latest progress of this all-electric robot. While previous videos have focused on Atlas's flexible and versatile movement capabilities, this time's focus is on the use of machine learning and upgraded sensors to perform sorting tasks in a simulated factory environment.
In the latest video, Atlas is tasked with "moving the engine cover from the supplier container to the mobile sequencing trolley", and while the task itself is not complex, it demonstrates the synergy of the new robot's multiple capabilities. With a list of bin locations that need to be moved, Atlas is able to determine the exact location of the bin with the help of a variety of sensors and machine learning models, and precisely adjust the body, arms, and three-finger claws to pick and handle the part.
IT Home noted that the new Atlas was able to accomplish assigned tasks and was able to respond in real-time to environmental feedback. For example, when placing a part, it is difficult for the part to enter the compartment of the trolley smoothly due to its high position, and Atlas will temporarily take the part out after encountering resistance, re-evaluate the angle and position, and finally successfully place it. This approach demonstrates its responsiveness to the environment and its ability to adapt flexibly.
The "fully automated" watermark that appears throughout the video shows that Boston Dynamics is trying to highlight its latest humanoid robot that can accomplish complex tasks with minimal human intervention. Just a few weeks ago, Tesla Inc. demonstrated the Optimus humanoid robot that served as a waiter to receive guests at the Cybertruck launch event, but it was later confirmed that the robots rely mainly on human remote operation.
Boston Dynamics' latest Atlas robot is more than just a tool to showcase the company's cutting-edge technology if it can be reasonably priced in the future.