Safety of Humanoid Robots: From Viral Falls to Multi-Layered Protection
Viral videos of humanoid robots kicking children or losing control while dancing have exposed a critical problem for the entire industry: how to deploy a machine in a warehouse or public space without injuring a person. The industry is moving from euphoria to pragmatism, implementing multi-level safety systems — from specialized Nvidia chips to abandoning bipedal designs in favor of wheeled platforms.
Incidents involving humanoids, including severe injuries and even fatalities, are forcing developers to fundamentally rethink safety approaches, which in turn slows mass adoption. According to functional safety expert Michelle Silva, the main threat lies in basic physics: "If a humanoid robot loses power, it can fall and crush you." The problem is compounded by the fact that these machines are probabilistic AI-based systems, not deterministic industrial robots operating on rigid algorithms. Their behavior is based on statistical probabilities, requiring fundamentally different control mechanisms.
Technical Solutions: From Chips to Architecture
A key player in this field is Nvidia, which introduced a safety system for humanoids based on Blackwell chips. According to the company's Senior Director Amit Goel, the model can interpret sensor data about potential hazards and forcibly stop the robot in unsafe conditions. "We created an operating system layer and a software stack so that the functional safety system and the main control system can work together in a much broader context," he explained.
However, safety is not limited to the robot's "brain." Philadelphia-based Fort Robotics is developing controllers that gather information from multiple sources: not just detecting a person in the work zone, but also analyzing their posture, position, and the confidence level of this data, allowing the robot to make informed decisions.
The problem of instability is so serious that the International Organization for Standardization (ISO) has created a dedicated expert group. Publication of unified requirements is not expected until mid-2028. In the absence of standards, manufacturers are finding their own paths. German company Neura Robotics has built a "controlled collapse" scenario into its 80-kilogram robot 4NE1: if, say, a knee joint fails, the machine falls "like a collapsing building," minimizing damage.
Radical Approach: Abandoning Legs
Some developers have decided to eliminate the root of the problem — the bipedal design. Dexmate creates robots on wheeled platforms with long manipulators. According to co-founder Yuzhe Qin, the battery and electronics are placed in the base, providing a low center of gravity and preventing falls.
Cobot founder Brad Porter suggests viewing the threat without excessive dramatization. His wheeled robots, pushing carts in hospitals or sorting parts in factories, 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," he quips.
My analysis: The evolution towards wheeled platforms and multi-layered software is not just a technical necessity but a fundamental shift in the philosophy of robotics. While the industry awaits ISO standards, we will witness an "arms race" among manufacturers, where the winner will not be the one who makes the most agile robot, but the one who can convincingly prove its safety for humans. Investors should closely watch companies that bet on software safety stacks, not just on "hardware."