University of Pennsylvania and Built Robotics are partnering to develop AI safety systems for construction equipment by collecting real-world data from active jobsites.

What happened: Built Robotics and Penn's Safe Autonomous Systems Lab (xLAB) will collaborate to build AI models for construction robotics. Built will deploy survey robots equipped with sensors on active solar projects to gather data; xLAB researchers will analyze this data to improve AI detection of humans and hazards in construction environments. The initial phase will focus on Built's edge AI model for personnel detection across a fleet of construction survey robots.

Why it matters: Construction is one of the most demanding environments for autonomous systems because robots must safely coexist with hundreds of workers across thousands of acres in unpredictable conditions. By systematically collecting data on edge cases—unusual body poses, occlusions, odd lighting, unexpected human behavior—the collaboration aims to train AI models capable of detecting dangers that humans might miss, addressing a core safety challenge in physical AI deployment.

What to watch: The research will expand beyond Built's current piling and solar work to other vehicle platforms and construction activities. Built has amassed more than 50,000 hours of operations and installed more than 3 gigawatts of solar across 40+ deployed sites, giving the partnership a substantial real-world dataset to build from.