Modeling Drivers' Risk Perception via Attention to Improve Driving Assistance
URL: http://arxiv.org/abs/2409.04738v1
archive: archived pipeline: cataloged verified
Abstract
Advanced Driver Assistance Systems (ADAS) alert drivers during safety-critical scenarios but often provide superfluous alerts due to a lack of consideration for drivers' knowledge or scene awareness. Modeling these aspects together in a data-driven way is challenging due to the scarcity of critical scenario data with in-cabin driver state and world state recorded together. We explore the benefits of driver modeling in the context of Forward Collision Warning (FCW) systems. Working with real-world video dataset of on-road FCW deployments, we collect observers' subjective validity rating of the deployed alerts. We also annotate participants' gaze-to-objects and extract 3D trajectories of the ego vehicle and other vehicles semi-automatically. We generate a risk estimate of the scene and the drivers' perception in a two step process: First, we model the movement of vehicles in a given scenario as a joint trajectory forecasting problem. Then, we reason about the drivers' risk perception of the scene by counterfactually modifying the input to the forecasting model to represent the drivers' actual observations of vehicles in the scene. The difference in these behaviours gives us an estimate of driver behaviour that accounts for their actual (inattentive) observations and their downstream effect on overall scene risk.
Summary
Driver-modeling study testing whether incorporating models of drivers' attention and scene awareness into Forward Collision Warning (FCW) systems reduces nuisance alerts. The authors note that 92.2% of conventional FCW alerts were judged 'nuisance' or 'crash unlikely' in deployed-vehicle data and argue that driver state and gaze context must inform alert timing. Real-world cabin and world-state recordings are used to learn risk-perception representations that gate when warnings should fire.
Key finding
Modeling driver risk perception via attention substantially reduces redundant FCW activations relative to context-blind alerting, suggesting that ADAS warnings benefit from joint reasoning over scene and driver state.
Methodology
Data-driven driver-modeling on paired in-cabin (driver state, gaze) and forward-scene recordings of safety-critical events. Models learned to predict the driver's current risk perception and were used to gate FCW activation. Performance compared to a context-blind baseline using nuisance-alert rate as the headline metric.
Sample size: Real-world driving dataset of safety-critical events; specific participant count not extracted
Quality score: 5 / 5