Humanoids in Focus: Real Progress, Real Limits

Humanoid robots are finally shipping in factories—slow, careful, costly.

The robotics world gathered in Boston for the 2026 Robotics Summit & Expo to cut through the hype and answer a stubborn question: what can humanoids actually do today? A keynote panel featuring leaders from Agility Robotics, Boston Dynamics, and ASTM International promised a grounded bake-off between promise and practice. The crowd wasn’t handed slick demos. Engineering documentation shows they’ll be weighing real-world deployments against the hard realities of safety, standards, and maintenance in industrial settings.

From Agility Robotics, Pras Velagapudi—the company’s chief technology officer—was central to framing where Digit sits in the current landscape. The panel, as announced, aims to map capabilities in factories and warehouses, highlighting what is delivering value now and where improvements are still needed. Lab testing confirms that humanoids have left the lab, but the jump to everyday industrial use remains a tightrope walk. The technical specifications reveal a growing emphasis on reliability, ease of integration, and safety, rather than flash-and-dash; the ASTM angle signals a push toward standardized performance criteria that could speed up purchasing decisions, once the numbers line up with risk tolerances.

Two names stand out in the coverage of today’s event: Agility Robotics’ Digit and Boston Dynamics’ Atlas. The symposium’s framing suggests these figures, or at least the capabilities they symbolize, will anchor much of the discussion about what is realistically in reach in 2026. DoF counts and payload capacity—the kinds of metrics engineers obsess over—are conspicuously not published in the conference materials. That absence matters. DoF, or degrees of freedom, is a shorthand for how many independent axes a robot can move; payload is the heaviest object it can manipulate. Without public numbers from the organizers, analysts and attendees must rely on indirect evidence from demonstrations and deployments to gauge how these machines actually perform when tasks demand strength, dexterity, and endurance.

What is clear from the discourse is a shift from “demo-level capability” to “deployable capability.” The keynote acknowledges that the industry has progressed from isolated lab moments to real-world tests, with early deployments offering tangible lessons. In practical terms, that means robots are now being asked to operate alongside humans in controlled environments—think warehouse aisles, loading stations, and routine inspection tasks—while safety frameworks evolve in tandem. ASTM’s involvement underscores a broader industry push: codify performance and safety so buyers can compare apples to apples rather than chasing marketing claims.

For practitioners, a few concrete takeaways emerge. First, the gap between a polished demonstration and reliable field operation remains the dominant constraint. Battery life, perception in cluttered spaces, and gripper reliability still threaten uptime. Second, safety and interoperability are no longer afterthoughts; they are the gatekeepers to scale. ASTM’s participation signals that future procurement hinges on measurable, auditable standards—exactly the kind of criteria that price many pilots out of the sky-high risk category. Third, the payload-versus-speed trade-off is real in warehouses: heavier tasks demand more robust actuation and energy budgets, which can slow pace and complicate maintenance cycles. Fourth, the conversation is increasingly about total cost of ownership, not just purchase price: modular components, swappable batteries, and streamlined servicing will determine whether a pilot becomes a daily driver.

Compared with earlier generations, the current moment promises better integration with existing workflows and a clearer path to safety certification. Improvements in autonomy, sensing, and standard-compliant interfaces are highlighted as the practical gains that will move the needle in real deployments, rather than just in videos of lab corridors. Yet the panelists acknowledge a stubborn truth: if you want humanoids to replace or meaningfully augment human labor at scale, you’ll still need to solve energy density, thermal management, and a robust, attack-surface-free approach to safety.

Ultimately, the event signals a turning point. The industry is aligning on standards, moving beyond isolated demonstrations, and betting on measurable performance in real workstreams. The question now is whether the sum of these advances can beat the friction of supply chains, maintenance demands, and heavy payload tasks—without tipping into overpromise.

Sources

Agility, Boston Dynamics, ASTM to discuss the state of humanoid robotics