Viral videos featuring robots kicking children and dancing uncontrollably have highlighted a key issue for manufacturers: how to deploy machines without harming people. The industry is addressing these risks with multi-layered safety systems—from Nvidia chips to replacing legs with wheels, according to WSJ.
https://www.youtube.com/watch?v=RrbfIxpdxv0
Various incidents involving robots have compelled developers to enhance safety mechanisms, thereby slowing down widespread adoption. Some companies have already reported serious injuries and fatalities involving humanoid devices.
In a conversation with reporters, Michelle Silva from Reynolds & Moore, which specializes in functional safety for robots, described a simple threat:
“If a humanoid robot loses power, it can fall and crush you.”
Another layer of complexity arises from the architecture of AI-based machines. Traditional industrial robots—such as welding machines, palletizers, and forklifts—are deterministic systems that follow a specific set of rules.
Humanoid robots, on the other hand, utilize AI and are probabilistic systems. They operate based on statistical probabilities rather than strict certainties. This is why they require multi-layered safety measures when working alongside humans.
In line with this concept, Nvidia has introduced a safety system for humanoid robots based on its Blackwell chips. Amit Goel, the company’s senior director of robotics and edge AI ecosystem, stated that the model can interpret sensor data about potential hazards and stop the robot in unsafe conditions.
“The safety system and the functional system need to interact frequently, and in a much broader context. We created this level of operating system and software stack so you can run these two things together,” said Goel.
Another level of control suggests that robots will rely not only on their own cameras and sensors but also on the surrounding infrastructure.
Fort Robotics, based in Philadelphia, is developing controllers and software that gather information from multiple sources. CEO Samuel Reeves explained that this involves not just visually detecting people in the robot's workspace, but also more complex data: where a person is located, their posture, and whether this information can be trusted enough for the robot to make decisions based on it.
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. Relevant requirements are expected to be published by mid-2028. In the absence of unified rules, manufacturers are developing their own scenarios.
The German company Neura Robotics is producing a bipedal robot called 4NE1, weighing around 80 kg. Founder David Reger stated 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 that fails, it will fall “like a collapsing building,” folding downwards.
Some developers have opted to eliminate the source of risk altogether. Dexmate is creating robots on wheeled platforms with long arms that can reach items on warehouse shelves. Co-founder Yuzhe Qin explained that the battery and electronics are housed in the platform, giving the machine a low center of gravity and preventing it from tipping over.
Two-wheeled robot. Source: WSJ.Brad Porter, founder and CEO of Cobot, suggested evaluating the threat without exaggeration. His company makes wheeled robots with manipulators that push carts in hospitals or sort parts on factory floors. They move at walking speed and do not possess excessive gripping strength.
“We don’t need to invest a lot of energy into necessary actions. We’re not trying to crush watermelons or anything like that,” noted Porter.
As a reminder, in July, the company Humanoid introduced an approach to training humanoid robots through trial and error on real production tasks.