Work

Android application-based automated prediction for green transportation mode detection: architecture, challenges and results

Year: 2025

Type: article

Abstract: Implementing an efficient Green Intelligent Transportation System (ITS) is essential for modern cities. An effective approach is to target users’ Transportation Mode Detection (TMD) using smartphone sensors combined with Machine Learning (ML) models. We focus on green urban transportation modes, excluding cars and motorcycles, to emphasize eco-friendly alternatives, and to improve urban mobility planning. In this paper, we outline the development of an Android application, designed for two complementary purposes: (1) to collect data from smartphone sensors (accelerometer, gyroscope, magnetometer, and Global Positioning System (GPS)) and (2) to predict four green transportation modes (walk, bike, public transport and kick scooter) based on accelerometer and gyroscope data only. GPS data are omitted in this paper, to address privacy concerns and energy consumption. Unlike existing works that focused solely on modeling and classification in TMD, the main contribution of the paper goes beyond and provides moreover a detailed overview of the implementation of the mobile application, as well as the challenges encountered during data acquisition and practical solutions we applied and the entire automatic cycle from data acquisition to classification, prediction and display of results on the application. Developing this application and its automated pipeline with a reliable client-server communication presented many technical challenges, requiring reliable data transfer and robust design to ensure good performances. The results obtained show good predictions, providing the user with an efficient tool to evaluate their sustainable mobility habits. Encouraging such modes contributes to ease traffic congestion, and more sustainable urban lifestyles.

Source: Journal of Intelligent Transportation Systems

DOI: 10.1080/15472450.2025.2576935

Authors: Ilhem Gharbi, Denis Faure-Vincent, Hassen Fourati, Nicolas Vuillerme