What we’re watching next in humanoids

Boston Dynamics’ latest humanoid prototype packs more joints than Atlas did in its prime—and it’s still learning to walk on straight lines.

In a world that has grown tired of vaporware, this lab demo signals a concrete step forward in the ongoing tug-of-war between more degrees of freedom and practical reliability. Engineering documentation shows the new design touts a higher joint count than its predecessors, with early field notes suggesting somewhere in the mid-40s for total degrees of freedom (DOF). By comparison, Atlas, the best-known humanoid from a few years back, has long been cited around the high-20s in DOF. The new entrant’s claimed DOF figure, if confirmed, implies a twice-as-flexible system in theory—but more joints are only useful if the control software, sensing, and actuation can keep everything coordinated.

DOF and payload: The engineering commentary surrounding the prototype points to 42 DOF being a talking point in internal documents, versus Atlas’s roughly 28 DOF. The payload capacity remains undisclosed publicly, a common stance for evolving humanoids where the manipulation stack is still under active tuning. What is clear from the demonstrations and coverage is that the new rig emphasizes more nuanced gait control and hands that can reach and grasp with greater dexterity, not merely stiffer limbs. For the Atlas baseline, payload capacity has not been officially published in modern disclosures, leaving this as a traceable gap in public specs.

Power, runtime, and charging are equally murky in official accounts. The Atlas lineage uses hydraulic actuators powered by an onboard pump and battery system, a setup that provides high-torque capability but complicates endurance and maintenance. The new prototype’s power architecture is described in broad terms in vendor notes, with no published runtime figures or charging profiles. In short: you get more joints, but you don’t yet get a reliable, field-ready endurance story.

Technology readiness remains firmly in the lab-demo lane. The demonstrations described in IEEE Spectrum coverage and The Robot Report emphasize controlled-environment testing, precise balance on varied terrains, and selective manipulation tasks. There is no public evidence yet of outdoor testing, autonomous navigation in messy real-world settings, or long-duration reliability data. Translation for deployment teams: the system shows how far the actuation and control stack has come, not how it behaves in a factory floor, disaster-site, or home setting.

What this progress actually buys you, beyond a glossy demo, are two core shifts. First, more DOF translates toward more natural human-like motion and finer manipulation, enabling tasks that require subtle hand-eye coordination. Second, the march toward higher DOF increases the demand for robust perception, predictive control, and fault-tolerant software. The biggest current limitation remains real-time coordination at higher joint counts; stumbles in balance, timing, and grip stability become the most common failure modes when you push into higher DOF with imperfect sensing or latency.

Compared to prior generations, the trajectory is clear: more joints, more capable hands, and more ambitious task profiles. But the gap to field-readiness remains substantial. The industry’s pragmatic expectation is that next-year milestones will center on extended endurance, safer interaction with humans and objects, and a clearer, publishable payload specification. Until then, we watch for independent tests that quantify stability margins, energy efficiency, and how gracefully the robot recovers from a slip.

What we’re watching next in humanoids

Transparent DOF accounting and actual payload tests from a public field trial

Real-world endurance data under varied terrain and loads

Robust grip reliability for everyday objects and safety in close human-robot interaction

Independent validation of balance control and recovery from perturbations

Sources

IEEE Spectrum Robotics

The Robot Report

Boston Dynamics