SafeWorld argues that traditional robot testing methods become less practical as machines rely more heavily on AI systems whose behavior can vary across environments and situations. Physical testing can cover only a limited number of cases, while safety measures such as cages and speed restrictions can constrain how robots operate alongside people.
The company’s platform lets teams create scenarios in a browser using information from previous incidents, safety requirements and robot logs. SafeWorld then tests the robot across thousands of variations using simulated human behavior and records how the system performs.
Those simulations can include circumstances such as blind corners, people carrying objects, different body positions or someone unexpectedly falling near a robot. The robot being evaluated can run its actual control software inside a digital representation of the environment.
“One of the biggest lessons from autonomous vehicles is that real-world testing alone can't cover every dangerous situation,” said SafeWorld co-founder Dr. Ding Zhao. “As robots move into factories, warehouses, and other human environments, simulation gives us a way to test those situations before they happen in the real world. The same rigor the AI community is bringing to models now needs to apply to the machines those models control.”
Because changes to robot software or operating environments can introduce different risks, SafeWorld is designed to let companies repeat testing throughout development and deployment rather than treating safety evaluation as a one-time process.
The company is already running early pilots with automotive manufacturers, warehouse automation companies and medical device makers. Anyware Robotics and Gritt Robotics are among the robotics companies described as working with SafeWorld on safety simulation.
“Safety has been foundational to Anyware Robotics from 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.”
SafeWorld was co-founded by Zhao, Kyle Wong and Simo Rachidi. Zhao directs Carnegie Mellon University’s Safe AI Lab and previously worked at Google DeepMind. Wong previously founded AI and user-generated content platform Pixlee and later served as CEO of Stanford startup accelerator StartX. Rachidi previously worked as a principal security and machine learning engineer at Salesforce Einstein and has founded multiple startups.
“As robotics moves from impressive demos to everyday deployment, safety becomes a prerequisite for adoption.” said Wong, SafeWorld’s co-founder and CEO. “We believe more modern ways to test and validate safety can help unlock the broader potential of robotics.”
The company is also positioning independent safety testing as part of its role. Its founders argue that robot makers may want outside validation of their systems in addition to the simulation tools they develop internally, particularly as similar safety problems emerge across different companies and deployment environments.
SafeWorld’s funding round also included Ovo Fund, Valkyrie, Zelda Ventures, Alpha Square Group, Founders Future and Brave Capital, along with founders and executives from companies including NVIDIA, Google DeepMind, Waymo, Meta, DoorDash, Together AI and Salesforce.
The company plans to use its simulation technology to help robotics developers and operators evaluate increasingly autonomous machines as they move into environments shared with people.
This analysis is based on reporting from Robotics Tomorrow.
Image courtesy of Safeworld.
This article was generated with AI assistance and reviewed for accuracy and quality.
