What we’re watching next in humanoids

Boston Dynamics has unveiled an updated version of its humanoid robot, Atlas, now boasting an impressive forty-two degrees of freedom—double what the previous iteration had in 2022.

This increase in degrees of freedom (DOF) allows Atlas to perform complex movements with greater precision and versatility, a key factor in its ability to navigate unstructured environments and interact with objects effectively. The latest demonstrations show Atlas executing tasks such as folding laundry and picking up fragile items, like eggs, showcasing improvements in both dexterity and reliability.

The technical specifications reveal that Atlas now has a payload capacity of 30 kg, making it significantly more capable of handling various objects compared to its predecessor, which was limited to around 20 kg. This enhancement is a result of refined actuator designs and optimized control algorithms, allowing for smoother and more controlled motions. Engineering documentation shows that the new actuators have been designed to deliver improved torque specifications, which are critical for maintaining stability during dynamic movements.

Despite these advancements, the robot is not without limitations. Lab testing confirms that Atlas still struggles with tasks requiring fine motor skills under real-world conditions. For instance, while it can fold laundry, the results are not always consistent, revealing the ongoing challenge of developing robust algorithms that can adapt to unpredictable scenarios—something that many humanoid robots face.

From a Technology Readiness Level (TRL) perspective, Atlas remains in a controlled environment stage. While it has made significant strides in laboratory settings, deploying it in real-world applications presents additional challenges. The robot's runtime is approximately 90 minutes on a full charge, with charging requirements involving a dedicated power station that can handle its specific voltage and current needs.

Comparison to the previous generation highlights several key improvements beyond just DOF and payload capacity. The latest Atlas features enhanced battery technology, which contributes to a longer operational time, and has undergone software updates that improve its learning capabilities through machine learning algorithms.

The real-world performance of Atlas will ultimately depend on how well it can adapt to varied environments and tasks beyond what it has been explicitly trained for. The path to commercial viability for humanoids like Atlas is fraught with failure modes, including potential issues with sensor integration and environmental adaptability.

As we look ahead, the robotics community is keenly observing several signals that will indicate the future trajectory of humanoid robots like Atlas:

Sensor Fusion: Monitor developments in sensor technology that can improve spatial awareness and object recognition in complex environments.

Adaptive Algorithms: Watch for breakthroughs in machine learning that enable robots to learn from failure and adapt their behaviors autonomously.

Market Adoption: Keep an eye on pilot programs in industries like logistics and healthcare where humanoids might be tested in practical applications.

Cost Reduction: Track advancements in manufacturing processes that could lower the cost of components, making robots like Atlas more accessible for commercial use.

Regulatory Frameworks: Assess how evolving regulations around robotic deployments influence research and development priorities in the humanoid sector.

The next few years will be pivotal for humanoid robots as companies like Boston Dynamics push the envelope on what these machines can accomplish. While incremental progress might seem slow, each advancement is a testament to the complexities of robotics engineering and the challenges of bringing these machines from the lab to everyday life.

Sources

IEEE Spectrum Robotics

The Robot Report

Boston Dynamics