"DriverDistraction_NonIntrusive data sets"

"Phone induced driver distraction is a persistent safety risk, particularly at intersections and in dense traffic. Many production systems only intervene when advanced driver-assistance systems (ADAS) are active, and most phone mitigation approaches require pairing with the device. We present a non intrusive system that detects and mitigates phone induced distraction without connecting to the phone. The system fuses in cabin vision (gaze and hand\/phone posture), steering wheel infrared (IR) hand presence, and vehicle dynamics with environmental context (speed, intersection proximity, traffic, weather) to estimate a Distraction Probability Score (DPS). A two mode policy (Normal vs. Aggressive) adapts thresholds for teen driver\/child present conditions and escalates interventions from visual\/auditory prompts to seat vibration and a calibrated brake module pulsation that creates a perceivable rocking motion while preserving vehicle stability. We detail the sensing pipeline, DPS model, and safety interlocks, and we propose a two stage evaluation: a simulator study and a closed track pilot. Planned metrics include detection ROC\/PR, reductions in off road glance time and lane variability, reduced missed sign events, intervention latency, and driver acceptance. Because the design leverages hardware common in modern vehicles, it provides a practical path to deployment."