Robotic ’Dog‘ Assists in Construction of UW-Madison’s Newest Building
MADISON, WI – A quadrupedal robot, nicknamed a “robot dog,” is being utilized by University of Wisconsin-Madison engineers to enhance safety and efficiency during teh construction of a new campus building.The research, led by faculty member Xu, focuses on improving human-robot collaboration in challenging real-world environments, with potential applications extending to high-risk industries like infrastructure inspection.
The project aims to address critical gaps in robotic autonomy, ensuring these systems can safely coexist with human workers. Unless robots become context-aware and socially adaptive, they risk being ineffective or even creating hazards, Xu explained. This research is particularly timely as construction projects increasingly integrate automated technologies, demanding a focus on reliable and trustworthy robotic systems.
The engineering team is currently working with a Unitree B2 robot, observing its performance on an active construction site. Researchers have already noted the robot’s extraordinary mobility, successfully navigating stairs, uneven terrain, and confined indoor spaces – conditions typical of construction zones.This field testing has revealed navigation hazards that are impossible to identify through simulations or laboratory settings.
However, the B2’s geometry-based navigation system presents challenges. It frequently enough struggles to interpret contextual cues, such as fragile objects, warning markers like red tape, or designated human work areas. Insufficient processing power has also led to mapping failures and instances where the robot required human assistance to overcome arduous obstacles.
“Our research helps…ensure robotic systems are not just functional but trustworthy and safe,” Xu said. The team is focused on identifying specific hazards – including terrain instability, potential material contact, and boundary violations – to inform the design of more adaptive and context-aware robot autonomy systems. Ultimately, the goal is to support safer human-robot collaboration, particularly in high-risk sectors.
