Safety is often overlooked in the rush to bring humanoid robots to the workplace. As high-profile corporations ranging from retailers like Amazon to carmakers such as Mercedes and BMW have announced humanoid pilots for factories and warehouses, conversations around worker safety are regularly buried beneath industry hype.
One Bay Area-robotics firm is hoping to correct that shortfall.
Figure AI announced Tuesday it is building a Center for the Advancement of Humanoid Safety, an in-house division wholly focused on the topic. Rob Gruendel, the former Amazon Robotics safety engineer who is heading up the project, revealed the plans via a LinkedIn post.
Earlier factory and warehouse systems addressed the safety problem by caging in big, heavy robots. Subsequent solutions have turned to technology, including advanced computer vision, software developed by Veo Robotics, and a safety vest designed by Amazon to help keep robots from crashing into people.
Amazon’s wearable is designed solely for internal usage and likely won’t appear outside of its own fulfillment centers. Even with that technology, the company still uses protective cages for many of its own robots. Veo Robotics, meanwhile, was absorbed into Symbotic in 2024. More recently, Symbotic has taken the reins of Walmart’s robotics division.
While valued for its adaptability, multi-purpose functionality, and ability to integrate into existing brownfield warehouses, the humanoid form factor introduces a new set of safety concerns. After all, one of the category’s key selling points is its ability to work alongside humans. Their big, metal bodies — ranging freely inside factories and warehouses — have the potential to cause injuries by colliding with their softer, fleshier colleagues.
Workplace safety organizations like Occupational Safety and Health Administration have a lot of catching up to do in terms of automation safety regulation. Today, there are no “specific OSHA standards for the robotics industry,” according to the federal agency. This broader concern needs to be addressed, but regulation specifically targeting humanoids is especially timely.
Figure AI is homing in on that gap.
“One of our recent successes was to finalize a formal plan with our OSHA recognized independent testing laboratory to certify our robot’s battery, functional safety control system, and electrical system to industrial standards,” Grudendel noted in the announcement post.
Fetch founder, Melonee Wise, has made the topic of humanoid safety a key focus since joining Agility Robotics as chief technology officer in 2023.
“With any humanoid robot operating in this space, safety is not clear,” she told me during a humanoids panel at last year’s Automate conference. “There isn’t an easy stop for [Tesla’s] Optimus. There isn’t a stop on many [humanoid robots], and that is against the safety standard. It’s very clear that many of the companies are not interested in it.”
The creation of the Center for the Advancement of Humanoid Safety is an important step on that journey for Figure.
“We recognize that the general population is often at the mercy of the technology experts to judge whether an AI-controlled robot can be safe,” Gruendel wrote. “We want to speak directly with our customers. We will test and communicate the robot’s stability while stationary, stability while moving, detection of humans, detection of four-legged pets, safe AI behaviors, and navigation to prevent injury. We will listen to suggestions from our customers on ways to test the safety of Figure 02 and beyond.”
Figure plans to publish quarterly updates offering more transparency around the process. The reports will include testing procedures and fixes for potential hazards.
“We will summarize our successes and failures,” according to Gruendel.
Those successes and failures could eventually close the gap on safety and prepare the industry for the next big leap: bringing robots into the home.
Humanoid safety standards are set to become increasingly essential as more firms push to bring these robots into the home.
