Projects

This project accelerates the real-world impact of a smartphone-based telematics system designed to reduce road accidents in West Africa. Key activities include developing an analytical dashboard for fleet managers, implementing a drivers-requested "hijack detection" feature, and formally engaging the Federal Road Safety Corps (FRSC) and transport unions (NURTW) in Nigeria and Cameroon. The hijack detection feature models three specific scenarios using existing sensor data: high G-Jolt events, suddenly vanished sensor vibrations, and abrupt vehicle motion state changes. The dashboard enables fleet operators and insurance companies to monitor driver behaviour and measure safety improvements. Ethically, the project prioritises privacy-by-design and equitable engagement, working with driver unions to ensure the technology empowers drivers and enhances personal security rather than acting as punitive surveillance.

A dedicated Android telematics data-collection app tailored for Nigerian driving conditions. The app captures smartphone-sensor data and driver-reported external influences for each trip. First deployment yielded 400,000+ data points across multiple Nigerian cities. Source code is publicly available for reuse and adaptation by other researchers.

An end-to-end architecture encompassing trip detection, on-device behaviour analysis, and a dual-component NLG pipeline designed for robustness to intermittent connectivity and noisy sensor data. The system uses a two-step "reflection" pipeline for higher-quality, psychologically and culturally appropriate feedback delivered after trips. Pilot deployment with 90 drivers demonstrated promising evaluation results.

A continuous, smartphone-sensor approach to detecting alcohol-influenced driving suited to low-resource contexts. The model was trained on a Nigeria-collected dataset and achieves recall 100%, precision 60%, F1 75%, and accuracy 90.91%.

An Android app published on Google Play for evaluating how different presentation styles affect driver understanding of harsh-braking factors. Part of the broader research programme on explanation and feedback for driving behaviour change.