Planner-level fusion trust under thermal-infrared spoofing

Determine planner-level policies that weight RGB and LiDAR corroboration according to sensor reliability and evaluate threat-aware thermal-infrared signal processing against thermodynamically consistent heaters.

Background

Thermal-infrared sensing is especially important at night or when visible-RGB contrast is poor, but that same dependence can allow physically plausible heaters or wearables to dominate multimodal perception. The paper identifies a tension between preserving night-driving performance and resisting thermal-infrared spoofing. It leaves unresolved how planner-level fusion should account for sensor reliability and how robust signal processing should be assessed against realistic thermal decoys.

References

Night performance still conflicts with IR spoof resistance.

— Can Autonomous Vehicles Keep Their Cool? A Survey of Thermal Side-Channel and Perception Attacks  (2610.21034 - Aaryan et al., 15 Oct 2026) in Section 6, subsection “Open Problems,” item “Fusion trust under IR spoofing”