Abstract
Road networks are fundamental to the functioning of modern society, enabling the movement of both people and goods across local and global scales. In recent years, the rapid growth of e-commerce has significantly increased demand for commercial transportation, particularly in last-mile delivery systems. With this shift, logistics companies, government organizations, and other stakeholders are faced with the challenge of maintaining the safety and efficiency of road networks while accommodating rising delivery demands. Commercial vehicles, such as semi-trucks and delivery vans, are essential to supply chains, yet they significantly contribute to congestion and roadway risk. For example, urban logistics account for approximately 20–30% of road traffic and up to 53% of total delivery costs, highlighting both their prevalence and inefficiency. At the same time, crashes involving large commercial vehicles are more likely to result in severe or fatal outcomes, with driver fatigue identified as a major contributing factor due to long and irregular working hours. While infrastructure limitations and other risk factors also shape this issue, they fall outside the scope of this research. Addressing this broader problem requires both technological solutions to improve driver safety and systemic strategies to better manage delivery-driven traffic patterns.
To address the safety risks associated with driver fatigue, a research team developed the Intelligent Roadway Insight System (IRIS). IRIS is a wearable driver monitoring device in the form of a hat and was designed to detect drowsiness in real time to ensure users stay attentive while driving. Unlike existing in-vehicle driver monitoring systems, IRIS provides a low-cost, portable device that can be used across vehicles while preserving user privacy. The system is powered by a rechargeable lithium-ion battery and captures data using a camera and an accelerometer. This data is processed on an embedded microcontroller before being discarded. A neural network accelerator on the microcontroller enables the use of a convolutional neural network to classify the state of the wearer’s eyes, while a threshold-based algorithm analyzes accelerometer data to detect head motion associated with drowsiness. The results from the individual sensors are combined on the microcontroller to generate a final drowsiness classification. Based on this classification, a signal is sent to a haptic module to alert the driver when drowsiness is detected. The final prototype successfully demonstrated real-time multimodal sensing and on-device processing, with drowsiness detection through both head movement and eye state exceeding the target accuracy of 75%. However, limitations in hardware integration caused a reliance on evaluation boards and reduced nighttime performance. Despite these challenges, the results validate the feasibility of a wearable, privacy-conscious driver monitoring system and highlight key areas for future refinement, particularly in custom hardware integration.
To address the growing impact of last-mile delivery vehicles on urban traffic congestion, an investigation on how the increased use of last-mile deliveries has shaped traffic patterns and regulations in urban environments was conducted. To carry out this investigation, a comparative analysis of New York City and London was conducted using government documents, policy reports, and journalistic sources to evaluate both regulatory intent and public response. New York City and London were selected in part because both have implemented congestion pricing as a central regulatory response and because of their comparable roles as major urban centers. In London, congestion pricing reduced traffic by approximately 15% and delays by 30%, while New York’s program eliminated over 2.1 million monthly vehicle trips into its central business district. However, public acceptance of congestion pricing initially lagged due to concerns over cost and economic impact, highlighting the importance of stakeholder perception in policy success. The findings also show that regulations have been designed not only to reduce traffic volume but also to restructure how and when deliveries occur. New York’s off-hour delivery program serves as an effective way to keep large vehicles off of the road during peak traffic hours. Ultimately, this research argues that effective last-mile regulation requires a combination of economic incentives, operational adjustments, and public buy-in to sustainably reshape urban traffic systems.
