Trip phase recognition and transport mode classification through mobile sensing technologies

As urban populations grow and sustainability challenges increase, the ability to automatically detect transport modes and recognize trip phases using mobile sensor data has become critical for urban planning, public transport management, and environmental monitoring. This thesis presents a comprehensive approach to transport mode detection and trip phase recognition, utilizing GPS data from mobile devices. The core objective is to develop robust machine learning and deep learning models capable of classifying transportation modes such as walking, cycling, driving, and transit use (bus, metro, etc.) with high accuracy while simultaneously identifying trip phases, including access, egress, and waiting times. By leveraging large-scale datasets such as the GeoLife and Sussex-Huawei Locomotion datasets, the study applies advanced data preprocessing techniques and designs multiple classification algorithms, including Random Forest and Convolutional Neural Networks (CNNs), to effectively distinguish between different transportation modes. A key contribution of this research is developing a novel framework for trip phase recognition, which segments journeys into distinct stages, enabling the automated calculation of critical transit metrics such as waiting time at public transport stops, access and egress times, and distance to and from transit stations. The results of this study have wide-ranging implications, including optimizing public transportation systems, improving commuter experience, reducing carbon emissions by encouraging sustainable transportation choices, and providing policymakers and urban planners with actionable insights based on real-world mobility patterns. Furthermore, the thesis presents new key performance indicators (KPIs) to evaluate the accessibility level of public transit stations. This work advances transportation research and lays the groundwork for future developments in smart city initiatives and digital mobility services.