Safety by Design: How Humanoid Robots Must Evolve to Depart the Walled Garden

Humanoid robots continue their march toward becoming a commercial reality, with analysts such as Omdia projecting that global shipments could exceed 10,000 units by 2027 and reach 38,000 units in 2030. Adoption of the technology could be nearing an inflection point, a time when rising industrial demand meets advances in onboard AI computing, servo motors, harmonic drives, and sensors.

The biggest gap to bridge may be cultural as prospective users question why conventional mobile robots and AGVs equipped with robot arms cannot offer sufficient functionality and mobility without the need to design an entire new form factor. But the unique form factor of humanoid bots is precisely what makes them purpose-built for any environment or mechanical task to which humans are suited. Humanoid bots can navigate stairs, manipulate tools, work alongside people on an assembly line, or move through a warehouse without infrastructure modifications. Their human-like adaptability is their value proposition.

The humanoid form factor, however, is also the root of the technology’s most pressing unsolved problem. A robotic arm behind a safety cage is dangerous within a predictable envelope. A 150-pound bipedal machine moving autonomously through a shared workspace carries risks that existing safety frameworks were never designed to address. Before humanoid robots can engage humans collaboratively, they must be able to engage humans safely. That not only requires new standards for humanoids but system-level engineering.

Humanoid Standards Take Shape

Safety standards are a crucial component to the use of industrial robots. Since there are multiple types of robots and robot applications, there are standards to satisfy the different classifications and their unique characteristics.

In its first major revision since 2011, the U.S. national industrial robot safety standard ANSI/A3 R15.06-2025 was published in September 2025. The newly revised standard updated functional safety requirements, integrated collaborative robot application guidance previously covered under ISO/TS 15066, and formally recognized cybersecurity vulnerabilities as potential physical safety hazards. In a similar way, ANSI/RIA R15.08-1 & 2 cover the safety requirements for industrial mobile robots (IMRs) and the system requirements for integration of IMRs.

Neither of these standards cover the safe use of humanoids. An important distinction not covered within R15.06 or R15.08 is that a humanoid is a dynamically stable industrial mobile robot (DSIMR). These are new capabilities which have become possible due to recent advancements of robotic technologies and therefore require a new standard document.

What is a DSIMR?

DSIMRs are a class of robots that includes humanoids as well as other form factors listed below. What they have in common is that they rely on dynamic management of their motions in order to maintain their stability. Put into simple terms, in the absence of power they will fall over. This means that the "usual" behavior of a machine when the E-stop (emergency stop) is pushed (remove all power to the actuators) is NOT the optimal response for a DSIMR, because that will create a new hazard due to the falling robot.

DSIMR Form Factors

• Bipedal (also known as "humanoid") robots
• Quadrupedal robots
• Legged robots with any number of legs
• Wheeled balancing robots
• And robots with other form factors

Knowing this, it’s clear to see how standards created for traditional IMRs would fall short of safeguarding the use of DSIMRs. Because of the rapidly developing capabilities of these DSIMRs, they are becoming increasingly common in industrial settings. To create a system where human safety is ensured around these advanced robot form factors, a new safety standard was announced in May 2025. ISO/AWI 25785-1 Robotics -- Part 1: Safety requirements for industrial mobile robots with actively controlled stability (legged, wheeled, or other forms of locomotion) is a new document providing technical requirements to address the recent advancements of robotic technologies.

Industrial Settings vs. Home Use

Though the notion of humanoid bots entertaining children or cleaning the garage has attracted media headlines, no standards have yet looked beyond industrial settings. The regulatory picture for home deployment is essentially a blank page. ISO 13482 covers personal care robots in general terms, and the EU AI Act classifies autonomous robots as high-risk AI systems that require human oversight — but neither was written with bipedal humanoids in mind. During a panel discussion at A3’s Humanoid Robot Forum last September, moderator and A3’s managing editor Brian Heater described home environments as the “Wild West” of humanoid edge cases. Unlike a comparatively controlled industrial environment, a human home is a hazard course of pets, toddlers, and uneven surfaces. It is without protocols.

Though home deployments are a compelling long-term vision, the factory floor provides ample engineering challenges for humanoid designers to bridge first.

Humanoid Pilot Programs Behind Walled Gardens

Today, industrial settings for humanoid pilot programs restrict their operation to so-called “walled gardens.” These zones define where robot and human workers might share a workspace — but not at the same time. This is operationally useful and legally defensible, but it concedes the core argument for the humanoid form factor: confined. Robots confined carry an inherent constraint in their ability to function and collaborate.

But so do humanoids restricted to a walled garden. The aim for humanoid designers is to transition from safety by separation to safety by design. That transition is a shorter step for some humanoids than others. Wheeled humanoids carry lower instability risk: If they lose power, they stop rather than fall.

Bipedal robots must take an extra step as they present a more complex risk profile. Their primary hazards are dynamic: unexpected falls, stumbles, and unpredictable recovery movements from a machine that could weigh as much as a person but that reacts with far less situational judgment.

Achieving genuine human-robot collaboration beyond the walled garden requires component- and system-level advances that can enable multiple, layered safeguards. Depth cameras, LiDAR, and proximity sensors must provide robots with a continuously updated model of their environment, including the erratic, unpredictable movements of nearby humans. Force-limiting actuators need to detect unexpected contact and shed torque before injury occurs. Active balance systems must manage recovery from a bump or stumble without redirecting the robot's mass toward a worker. And if all else fails, mechanical joint brakes must prevent a full collapse. No single component can provide adequate protection. Humanoid safety becomes an emergent property of hardware, software, and environment working together.

The engineering work to develop safer humanoids is making strides,  and standards are responding. What remains is the harder task of proving, through sustained real-world operation and rigorous validation, that collaborative humanoid robots are not just technically feasible but demonstrably safe — consistently, at scale, and outside the walled garden.

There is no shortage of internet videos displaying the incredibly flexible physical and performative feats that humanoids are capable of — or not. In most cases, these videos are notable for how rarely humanoids are operating in close proximity to humans, which underscores that the biggest challenge for humanoids is much more workaday. Before humanoid robots can become something more than an internet sideshow, engineers and standards authors will need to ensure these bipedal machines are as safe as they are helpful.

The A3 Humanoid Robot Forum is Happening at Automate 2026!

As humanoid robots transition from experimental labs to the factory floor, the Humanoid Robot Forum provides the blueprint for what comes next. This intensive two-day event, taking place from June 23-24 at Automate in Chicago, explores the cutting-edge insights into the performance constraints and measurable outcomes of today's most advanced bionic systems. Learn more and join us there!