Smart Traffic Management System using Edge AI and V2I Communication

Project Abstract

This research presents a comprehensive design and evaluation of a Smart Traffic Management System (STMS) that leverages Edge AI and Vehicle-to-Infrastructure (V2I) communication to optimize urban traffic flow, reduce congestion, and enhance road safety. The system integrates heterogeneous data sources including connected vehicle streams, roadside sensors, camera feeds, and weather and event data to enable real-time decision-making at strategically deployed edge nodes. The core innovation lies in distributing computational workloads across edge devices to minimize latency in perception, prediction, and control, while preserving data privacy and reducing backhaul bandwidth requirements. By implementing lightweight deep learning models and adaptive inference techniques on edge hardware, the STMS performs robust object detection, traffic state estimation, and anomaly detection with high accuracy under varying network conditions and illumination. The V2I layer facilitates bidirectional exchange of actionable information between vehicles and infrastructure, enabling cooperative adaptive traffic signal control, dynamic speed recommendations, and hazard warnings that are synchronized with central management dashboards. A hierarchical control framework is employed, combining local ramp/intersection controllers with a centralized optimization engine that ingests aggregated traffic metrics to compute globally beneficial signal timing strategies and routing suggestions. The research evaluates multiple modalities for sensor fusion, including camera-based vision, LiDAR proxy data, and inductive loop signals, to ensure resilience in the presence of partial sensor outages. A novel edge-aware optimization module formulates a multi-objective problem balancing travel time, emissions, and safety risk, solved through a real-time heuristic complemented by model-predictive components to anticipate short-term traffic evolution. The evaluation uses a multi-city dataset synthesized from real-world traffic patterns and augmented with synthetic scenarios to stress-test peak-hour conditions, incidents, and adverse weather. Key performance indicators include average travel time reduction, queue length, intersection throughput, energy consumption, emission estimates, and incident clearance time. Results demonstrate significant improvements in corridor-wide throughput, with reduction in average delay by up to 25-40% under peak loads and a substantial decrease in stop-and-go behavior. The system maintains sub-100 ms perception-to-action latency for critical safety messages and sustains robust performance with intermittent connectivity through edge-local decision logic and intelligent caching. A comparative analysis against centralized cloud-based control and traditional signal timing benchmarks highlights the benefits of edge decentralization in scalability and resilience. The research also includes a comprehensive assessment of privacy-preserving techniques, such as on-device feature extraction and policy-driven data sharing, to address regulatory and societal concerns. Finally, the study discusses deployment considerations for urban environments, including interoperability with legacy traffic systems, standardization of V2I protocols, and a phased rollout strategy with pilot corridors. The findings provide evidence that an edge-assisted, V2I-enabled STMS can deliver measurable improvements in mobility, safety, and environmental impact while maintaining robust operation under real-world constraints.

Project Overview

What This Project Is About

---

The project studies how traffic flow can be improved by using smart devices at intersections and vehicles to make better decisions in real time without relying on central servers. It combines lightweight on?device intelligence with car?to?infrastructure communication to ease congestion and reduce delays and emissions.

---

---

The Problem It Addresses

---

City roads often suffer from jams due to outdated signal timing and limited information sharing between vehicles and traffic lights. This leads to wasted time, higher fuel use, and increased pollution. The project seeks practical ways to coordinate signals and vehicles more efficiently in real time.

---

---

Objectives of the Project

---

1. Understand how edge computing can run simple traffic-optimization tasks near the street lights.
2. Explore vehicle-to-infrastructure (V2I) communication to share speed and arrival data.
3. Prototype a lightweight system that improves signal timing on the chosen intersections.
4. Evaluate performance improvements in terms of wait times and emissions estimates.

---

---

What You Will Do Step by Step

---

1. Review basic traffic concepts and edge computing ideas.
2. Design a simple traffic-light control logic that uses local data and V2I signals.
3. Set up a small testbed or simulator with vehicles and sensors.
4. Collect data on traffic flow, wait times, and fuel use under different scenarios.
5. Analyze results to compare with traditional fixed-timing signals.

---

---

Expected Outcome

A practical, easy-to-implement approach showing how edge AI and V2I can reduce delays and emissions at intersections, with clear steps for future expansion.