Non-Visual Human Eye Gaze Tracking Using Radar and Artificial Intelligence for Robust Driver Monitoring Systems

Sammanfattning

Estimating gaze direction from in-cabin sensors is central to driver distraction monitoring.Existing methods rely on cameras, which degrade under adverse lighting andcollect identifiable facial imagery. Millimetre-wave radar avoids both limitations,and prior work at 60 GHz has demonstrated sensitivity to eye-region micro-motionsuch as blinks. Whether radar can support directional gaze estimation, rather thanbinary event detection, remains an open question.This thesis develops a radar-only gaze estimation pipeline built on a 60 GHz FMCWsensor. Amplitude and inter-receiver phase-difference cues are extracted from theradar return, and candidate eye-movement events are detected from the radar signalalone, removing the need for camera or stimulus timing at inference. A lightweightdual-stream temporal architecture, DualStreamGazeNet, encodes the two cue typesseparately, fuses them through cross-modal self-attention, and produces both a directionlabel and a continuous gaze angle through a modality-aware hierarchicalclassifier and an auxiliary regression head.Experiments on a multi-session dataset show that the model achieves 85.7% balancedaccuracy for four-direction classification with azimuth and elevation errors of 7.82◦and 3.15◦. Under strict cross-session evaluation, few-shot calibration with three tofive labelled events per direction yields 82.1% balanced accuracy, demonstrating thatthe learned representation generalises effectively with minimal target-session adaptation.These results establish that close-range mmWave radar is a viable modalityfor event-level gaze estimation and can serve as a privacy-preserving, illuminationindependentcomplement to driver monitoring system.