Integrity-Gated Eco-CACC adds safety checks to eco-driving at intersections

The Integrity-Gated Eco-CACC framework addresses a critical blind spot in cooperative adaptive cruise control (CACC) systems: they assume the internal world model stays valid, even when sensor outages or semantic mismatches occur. Researchers Lyes Saad Saoud and Moussa Ayyash propose monitoring three indicators—positional innovation (GPS vs. expected location), observability loss (degraded sensor coverage), and semantic inconsistencies (e.g., wrong traffic light state). These feed into a unified integrity metric that outputs a trust score. When the score stays high, the vehicle continues standard eco-driving that optimizes fuel use at signalized intersections. When the score drops, control authority gradually shifts to a conservative fallback maneuver that prioritizes safety over efficiency.

Scenario-based simulations demonstrate the method's practicality. Under normal conditions, the system maintains nominal energy efficiency. During integrity degradation—such as a sudden loss of V2X communication or inconsistent signal timing—the framework triggers early, conservative responses. This prevents unsafe planning while still allowing eco-optimization when data is reliable. The approach differs from robust control methods that try to maintain performance under uncertainty; instead, it admits whether energy-optimal control is admissible at all. The paper suggests this epistemic filter could be integrated into existing CACC stacks for safer, more trustworthy autonomous driving.