SafeWorld emerges from stealth mode to build and deploy robot safety simulation technologies

SafeWorld this week surreptitiously emerged with $12.2 million in seed funding. The startup aims to simplify safety testing on AI-based robots with its simulation technology.

“Our view is that if you want to have more capable robots, it makes sense that you need more capable ways to test and validate those robots,” said Kyle Wong, co-founder and CEO of SafeWorld The robot report. “Otherwise, you won’t be using those robots.”

“We’ve been using robots for maybe 100 years already,” said Ding Zhao, co-founder of SafeWorld and director of the Safe AI Lab at Carnegie Mellon University. “We founded the company today because, with the arrival of artificial intelligence, it has actually taken automation to the next level. Robots, for the first time, have been allowed to come out of the cage and share the same space with humans.”

Zhao said this level of collaboration requires a whole new level of understanding for the robot, and that’s where SafeWorld comes in.

Shine Capital and a16z Speedrun co-led the financing round. It also included participation from Box Group, Carnegie Mellon University Endowment, Innovation Endeavors, SV Angel and other leading venture capitalists and angel investors.

Ultimately, SafeWorld wants to make it easier for robotics companies to build safe systems. The Oakland, California-based company currently has customers in industrial, manufacturing, logistics, construction and even in some cases home use, Wong said.

“Robotics companies themselves must, of course, be responsible for safety,” Wong said. “The question, though, is more, ‘Should they build it from scratch every time?’ We don’t think it makes sense.”

Inside the SafeWorld platform

SafeWorld said it offers companies a scalable way to test robot behavior around people in rare, dangerous and unexpected situations, without putting anyone at risk.

“These are security scenarios that have not yet existed in the world,” Wong said. “So how do you train against something for which there isn’t a large data set? It’s a big challenge.”

“We are providing robot safety simulation software that robotics companies or companies that are building robots can use to provide better safety evidence, to test different types of safety scenarios before they happen in the real world, or to test things that are actually too rare, too dangerous, or too expensive to test in the real world,” he added.

Teams can create scenarios directly in the browser from past incidents, safety standards, and robot logs, with no simulation skills required. SafeWorld then runs the robot through thousands of variations with realistic, responsive human movements and measures safety performance.

Because each software update and new environment can introduce new risks, teams can repeat these tests continuously, not just once before launch. The result is a safety record that engineering, safety and operations leaders can sustain, SafeWorld said.

Wong noted that while many companies work with simulation, using it for security brings new requirements.

“Where this breaks down and becomes different is really in the nuance of features and functionality,” he said. “You have to understand what the testing criteria are for safety. Obviously there are different standards, but these standards are not all comprehensive. So there are different types of risk assessments.”

When should companies start working with SafeWorld?

For some robotics developers, it can be difficult to know when it’s time to start investing in security. If you start too early, your system may change significantly before you see the real world, rendering much of your work useless. However, if you start too late, you run the risk of delaying implementations.

“We work with about half a dozen companies and two-thirds of those are publicly traded companies,” Wong said. “I would say if your implementation timeline is less than a year, you probably need to start thinking about security. If it’s over a year, maybe it’s a little too early.”

The question of when to start thinking about security also differs depending on the type of company.

“There are two types of AI-based robotics companies. One is to bring AI to robots. Another is to bring robots to AI,” Wong said. The first category involves companies like Caterpillar, which already has sophisticated hardware and wants to bring AI capabilities to that hardware.

For these customers, security is important from day one, because they already have the hardware working all over the world. If you start with software, security comes later when you start working in the real world, but getting ahead in terms of security can still help these companies make things easier later.

“Security has been paramount to Anyware Robotics since day one,” said Thomas Tang, CEO of Anyware Robotics. “Advanced simulation tools like SafeWorld give us a scalable way to test challenging scenarios, strengthen our safety processes, and better prepare our robots for real-world deployment.”

Startup plans to promote security research

Looking ahead, SafeWorld said it plans to use its latest funding to deepen adoption, strengthen its team and advance safety research across the industry.

“We’re really trying to go very deep in terms of enterprise adoption and going very deep with some of our enterprise customers,” Wong said. “There’s so much going on here with safety, it makes sense for us to really focus on that delivery.”

“From a marketing perspective, we’re trying to do a lot in terms of promoting different research, whether it’s funding research or promoting research, and open source work that is at the frontier of AI security,” he added.

As the robotics industry grows, Ding and Wong agreed that having third parties like SafeWorld can help build trust with other stakeholders.

“Part of the reason I think a third party is very helpful here is that there’s an engineering part of security, and then there’s a trust part of security, and you need both to implement,” Wong said. “The fiduciary aspect of security is really about managing the stakeholders around security and who actually approves it.”

“This is really an implementation issue. So having a third-party company working on this is really necessary to win the trust of not only the company itself, but also the entire industry,” Ding said.