A new startup called Safeworld has emerged from stealth mode with more than $12 million in seed funding to address a critical challenge facing the robotics industry: ensuring that robots controlled by generative AI systems are safe to work alongside humans. The company was founded by Dr. Ding Zhao, who directs the Safe AI lab at Carnegie Mellon University, along with veteran startup executive Kyle Wong and machine learning engineer Simo Rachidi.
The funding round was led by Shine Capital and a16z Speedrun, with additional investment from Box Group, Carnegie Mellon University Endowment, Innovation Endeavors, and SV Angel. According to Jonathan Lai, a partner at a16z Speedrun, the timing is crucial for establishing industry safety standards while robots are still being designed and deployed, rather than waiting until safety incidents occur in real-world settings.
The core problem Safeworld aims to solve stems from the nature of generative AI itself. Unlike traditional algorithms, generative AI in robotics architecture is less predictable, making it difficult to guarantee safe operation. Safeworld’s approach involves evaluating robotic control systems in simulations populated with realistic human models, running thousands of scenarios to test how robots respond to various situations.
The company uses established simulation platforms like Genesis or MuJoCo to create digital environments where robots can be tested. These simulations are particularly valuable because they allow Safeworld to test dangerous or difficult scenarios without putting real people at risk. For example, the company can simulate people tripping and falling near robots to ensure the machines respond appropriately, something that would be impractical and dangerous to test repeatedly in the real world.
The challenge is compounded by the fact that robots operate in unstructured environments with different safety standards per facility. Common scenarios include robots navigating blind corners in factories, where engineers must determine appropriate speeds and stopping distances to prevent collisions with workers who might be carrying boxes or other objects that could obscure their presence.
Safeworld is already working with industry partners on practical applications. The company is partnering with Gritt Robotics, which develops AI for robots that help workers install photovoltaic panels at solar farms. For Gritt’s robots, which operate alongside human workers, safety verification requires considering countless scenarios involving different human postures, movements, appearances, and behaviors.
While many robot manufacturers are developing similar internal testing tools, Safeworld’s founders believe there is value in having a third-party validate safety systems. This independent validation could help share safety information between competitors and provide objective assessments that build trust with regulators and customers. The company is still determining whether to offer its technology as a platform for external users or as a services-based solution, but the team remains confident they are addressing a fundamental need in the rapidly growing robotics industry.