Abstract This paper introduces a framework that leverages Large Language Models (LLMs) to answer natural language queries about General Transit Feed Specification (GTFS) data. The framework is implemented in a chatbot called TransitGPT with open-source code. TransitGPT works by guiding LLMs to generate Python code that extracts and manipulates GTFS data relevant to a query, which is then executed on a server where the GTFS feed is stored. It can accomplish a wide range of tasks, including data retrieval, calculations and interactive visualizations, without requiring users to have extensive knowledge of GTFS or programming. The LLMs that produce the code are guided entirely by prompts, without fine-tuning or access to the actual GTFS feeds. We evaluate TransitGPT using GPT-4o and Claude-3.5-Sonnet LLMs on a benchmark dataset of 100 tasks, to demonstrate its effectiveness and versatility. The results show that TransitGPT can significantly enhance the accessibility and usability of transit data.

Abstract The design of transit signal priority (TSP) systems requires knowledge of dwell-time distributions at bus stops within the block. Dwell-time trends are not well established in the literature despite the ubiquity of large sets of automated passenger count (APC) data. Additionally, the impact of dwell-time variability on TSP performance is not well studied, particularly with field-collected dwell-time data. This study first analyzes trends and distributions inherent in dwell-time data deduced from APC data. Dwell times vary from stop to stop and for each stop by time of day (TOD). Stops with the highest proportions of non-zero dwell time also had the highest dwell-time magnitudes and variability. For most stops, dwell-time data was closely fitted by inverse Gaussian, log-normal, power log-normal, Fisk (log-logistic), and Johnson’s SU distributions. The second part of the study used a simulation environment to evaluate the impact of dwell-time magnitude and variability on TSP performance at both far-side and near-side bus stops. For far-side bus stops, dwell time significantly impacted the bus arrival profile at the check-in detector and thus the selected TSP strategy. Higher dwell-time magnitudes and variability led to a higher share of an Early Green Phase (EG) which is not as effective as a Green Phase Extension (GE). At near-side bus stops, dwell-time variability induced more uncertainty in an estimated time of arrival (ETA) and significantly reduced TSP effectiveness especially for GE. TSP performance in terms of GE success, bus travel time, and side-street traffic delay was significantly better at far-side stops compared to near-side stops.

Abstract Public transport (PT) is crucial for enhancing the quality of life and enabling sustainable urban development. As part of the UK Transport Investment Strategy, increasing PT usage is critical to achieving efficient and sustainable mobility. This paper introduces Machine Learning Influence Flow Analysis (MIFA), a novel framework for identifying the key influencers of PT usage. Using survey data from bus passengers in Southern England, we evaluate machine learning models. Subsequently, MIFA uncovers that easy payments, e-ticketing, and mobile applications can substantially improve the PT service. MIFA’s implementation demonstrates that strength and importance lead to specific insights into how service characteristics impact user decisions. Practical implications include deploying smart ticketing systems and contactless payments to streamline bus usage. Our results suggest that these strategies can enable bus operators to allocate resources more effectively, leading to increased ridership and enhanced user satisfaction.

Abstract Over the past few decades, intelligent transportation systems (ITS) have emerged as an effective way to improve the performance of transportation systems. ITS provide innovative services, enhance travel safety, provide travellers with more choices, and make transportation systems more efficient. Multi-agent systems (MAS), which define autonomous interacting entities, are suitable for modelling distributed and intelligent systems in general and ITS in particular. This paper provides an in-depth review of multi-agent systems applied to Advanced Public Transportation Systems (APTS), a subclass of ITS dedicated to managing public transportation networks. We carefully analysed 38 papers in this study, published in 19 journals during 31 years (1990–2020). We perform a synthetic analysis of the trends in this domain and a qualitative analysis focused on multi-agent systems’ dimensions and properties. We show that the MAS approach is well suited to the real-time management of disturbances thanks to their delegation process, and their pro-activeness and autonomy properties.