Safety of Humanoid Robots: How the Industry is Addressing the Issue of "Uncontrollable Machines"
Viral videos of robots kicking children or dancing uncontrollably have highlighted a key problem for manufacturers: how to deploy a machine in a warehouse without harming people. The industry is responding to these risks with multi-layered safety systems — from Nvidia chips to completely abandoning legs in favor of wheels.
The Problem of Probabilistic Systems
Traditional industrial robots — welding machines, palletizers, forklifts — operate by rigid rules. They are deterministic: their actions are predictable and follow strict logic. Humanoid robots, on the contrary, use AI and belong to probabilistic systems. They act based on statistical probabilities rather than rigid certainty. This is precisely why they need multi-layered protection to work alongside humans.
As robotics functional safety expert Michelle Silva notes, the simplest threat is a power failure: "If a humanoid robot loses power, it can fall and crush you." This is not just a hypothetical scenario. Some companies have already reported serious injuries and fatalities involving humanoid devices.
Safety Technologies: From Chips to Infrastructure
Nvidia has introduced a safety system for humanoid robots based on Blackwell chips. The company's Senior Director of Robotics, Amit Goel, explains: "The safety system and the functional system need to interact with each other frequently, and in a much broader context. We created this operating system layer and software stack so you can run these two things together." The model can interpret sensor data about potential hazards and stop the robot in unsafe conditions.
Another level of control involves the robot relying not only on its own cameras and sensors but also on the surrounding infrastructure. Philadelphia-based Fort Robotics develops controllers and software that gather information from multiple sources. CEO Samuel Reeves explains: it's no longer just about visually detecting people in the machine's work area, but about more complex data — where the person is, what posture they are in, and whether that information can be trusted enough for the robot to make decisions based on it.
Standardization and Alternative Designs
The issue of stability loss has become so sensitive that it is being studied separately by an expert group from the International Organization for Standardization. Publication of the relevant requirements is expected by mid-2028. Until uniform rules exist, manufacturers are developing their own scenarios.
Germany's Neura Robotics produces the bipedal robot 4NE1, weighing about 80 kg. Company founder David Reger states that the design minimizes risk: if the robot detects a problem, such as a knee joint failure, it will attempt to regain balance, and if unsuccessful, it will fall "like a collapsing building," folding downwards.
Some developers have decided to remove the source of risk altogether. Dexmate creates robots on wheeled platforms with long arms that can reach items on warehouse shelves. Co-founder Yuzhe Qin explains that the battery and electronics are placed in the platform, giving the machine a low center of gravity so it cannot fall.
Cobot founder and CEO Brad Porter suggests assessing the threat without exaggeration. His company makes wheeled robots with manipulators that push carts in hospitals or sort parts in factories. They move at walking speed and do not have super-strong grips. "We don't need to put a lot of energy into the necessary actions. We're not trying to crush watermelons or anything like that," Porter notes.
My analysis: The safety problem of humanoid robots is not a technical nuance but a fundamental barrier to mass adoption. Until the industry resolves the issue with probabilistic AI systems that can act unpredictably, we will not see a full-scale deployment of such machines in warehouses and offices. Current approaches — from Nvidia chips to wheeled platforms — are temporary solutions. The real breakthrough will come when we learn to create AI that not only predicts but also guarantees safety.