Excitation-Normalized Inertial and Wheel Speed Cues for Early Detection of Slippery Road Events
Bildiriler > Tam Metin Bildiri
Yazarlar: Hasan BULUT, Huseyin Karacali, Cagla Karahan
2026 International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA), , , Uluslararası, Tam Metin Bildiri, ss 1-6
DOI: 10.1109/acdsa67686.2026.11468067
Özet
Early and reliable detection of slippery road conditions is critical for driving safety; however, dynamics-based approaches are fundamentally limited by the so-called “excitation paradox,” in which friction-related observables typically appear only under high tire utilization, which occurs infrequently during normal driving. We introduce ENICE/AMI, a lightweight, non-machinelearning detection framework that integrates two physicsconsistent observables, with optional augmentation by wheelspeed asymmetry, within a shared excitation-gated hysteretic finite-state machine. The two primary signals are: (i) an IMU-based coherence-drop score indicating local decoupling between lateral acceleration and yaw rate, and (ii) a speed-aware lateral consistency metric comparing measured lateral acceleration with the value inferred from observed longitudinal speed and yaw rate. Wheel speed measurements provide an additional left-right asymmetry cue, which is fused using a conservative logical-OR rule. All thresholds are fixed and exposed as hyperparameters, and the overall computational complexity is linear with respect to the number of samples. In a dataset of 30 runs with 23 annotated low-friction events, the IMU-only configuration achieved a micro-averaged F1-score of approximately 0.19. Including vehicle speed increased this to approximately 0.82, and further incorporation of wheel speed information yielded a precision, recall, and F1-score of 1.00, with no false positives observed in this dataset.