FURI | Fall 2025

Uncovering the Dangers of Control Barrier Functions in Autonomous Robots

Autonomous robots are increasingly being tasked with more complex, precise, and important directives, yet many rely on insecure navigation systems. This research investigates a specific vulnerability in multi-robot coordination where an adversarial agent can influence, or “herd,” another robot to an undesirable location without making physical contact. Using control barrier functions within a simulated environment allows the team to model and explore how safety-critical control can be exploited. By varying parameters throughout test cases, the framework can identify the most vulnerable conditions, allowing programmers to safeguard against such attacks and improve the safety of these systems.