Track: Masters Thesis Competition

Abstract

Urban rideshares have become a cornerstone of urban transportation, offering unparalleled convenience and presenting a significant market opportunity. However, safety concerns are escalating as their usage proliferates. By harnessing advancements in computer vision and mobile technologies, alongside the widespread adoption of shared vehicles, urban rideshare systems can now deliver real-time road information and safety advisories. A critical gap remains in predictive capabilities and the translation of image data into human-understandable language or actionable safety recommendations. This paper investigates the potential of creating a real-time generative AI-based intelligent cognitive assistant for updating road information, specifically designed for urban rideshare safety recommendation services. Utilizing computer vision to collect road data during rides, the platform employs a generative AI database to comprehensively store and analyze this information. Users receive immediate safety updates through intuitive interfaces using a retrieval-augmented generation system. To demonstrate the platform's viability, a simplified test case is simulated, showcasing the system's effectiveness and potential impact on urban scooter safety. This test case highlights how the platform can enhance the safety and reliability of urban rideshare services by providing timely and accurate safety information.