2024GSoC 2024 — AI/ML Driver Safety Analytics

Eye-Tracking in Driving Simulators

A Google Summer of Code 2024 project analyzing driver behavior and risk using eye-tracking and vehicle telemetry from driving simulator sessions.

Source

Eye-Tracking in Driving Simulators screenshot

Driving simulators generate rich behavioral data, but turning raw footage and telemetry into actionable safety insight requires dedicated tooling. This GSoC 2024 project builds that analysis layer.

Using OpenCV, simulator video is processed frame-by-frame to extract steering angles and detect lane departures, while a companion notebook performs statistical analysis over driving telemetry — identifying harsh acceleration, braking, and sharp turns, and computing time-to-collision (TTC) values. The result is a set of heatmaps and risk scores visualizing driver input intensity and safety-relevant events across a session.

highlights

Frame-level steering angle extraction and lane-departure detection via OpenCV
Statistical detection of harsh acceleration, braking, and sharp-turn events
Time-to-collision (TTC) computation and driver risk scoring
Heatmap visualizations of steering, acceleration, and braking intensity over time

built with

Python
OpenCV
Pandas
Jupyter Notebook

Next projectLinguaSynth