Mr. chaitanya gunupudi

AI and Data Science in Autonomous Systems: Campus Shuttle Optimization using Reinforcement Learning

Abstract:

Autonomous transportation systems face significant challenges in dynamic campus environments, where traditional solutions often fall short in adapting to fluctuating demands, real-time constraints, and diverse objectives. This research presents a novel AI-driven approach to campus mobility—an intelligent autonomous shuttle system that learns and evolves through continuous real-world interaction. We introduce a multi-agent deep reinforcement learning framework that transforms fixed- route transportation into a self-optimizing, adaptive network. The system leverages autonomous decision-making, predictive analytics, and real-time data processing to proactively serve student needs rather than reactively following static schedules. At the core, Temporal Convolutional Networks forecast passenger demand by analyzing historical trends, while unsupervised learning identifies latent mobility patterns across campus zones. A robust feature engineering pipeline integrates heterogeneous data streams—from IoT sensors to academic calendars—enabling highly context-aware shuttle responses. Our agents, trained using Deep Deterministic Policy Gradient (DDPG) algorithms, learn coordinated fleet management policies that dynamically balance multiple objectives. These include minimizing wait times, optimizing route efficiency, and adapting to variables like weather events or exam schedules. With transfer learning and continuous policy updates, the system remains effective across academic seasons and unforeseen disruptions. Real-world deployment has shown a 34% reduction in average passenger wait times, 28% improvement in fleet utilization, and 42% reduction in operational costs, highlighting the tangible impact of integrating AI into autonomous mobility. This research contributes to the field of intelligent transportation by demonstrating how AI and data science can drive truly adaptive, safe, and efficient transit systems for complex environments.