Autonomous Intersection Management with Cooperative Control

Overview

This project develops a decentralized autonomous intersection management (AIM) system that enables self-driving vehicles to navigate intersections safely and efficiently without traffic lights. The system uses vehicle-to-vehicle (V2V) communication and cooperative control to coordinate crossing maneuvers and maximize intersection throughput.

Key Features

• Decentralized coordination without centralized infrastructure
• Real-time trajectory planning for collision-free intersection crossing
• Handles heterogeneous vehicle types and dynamic arrival patterns
• Provably safe coordination protocols
• Up to 3x higher throughput compared to traditional traffic signals

Technical Approach

The system employs a reservation-based coordination mechanism where vehicles negotiate crossing times through V2V communication. Each vehicle computes optimal trajectories that minimize travel time while respecting safety constraints and previously confirmed reservations. The approach uses model predictive control (MPC) for trajectory optimization and formal verification methods to guarantee collision-free operation.

Results & Impact

Simulation studies demonstrate significant improvements over traditional intersection management:

• 200% increase in intersection throughput during peak hours
• 40% reduction in average vehicle waiting time
• 25% decrease in fuel consumption and emissions
• Zero collision events across 10,000+ simulation runs
• Robust performance under V2V communication delays up to 100ms