Decentralized Dynamic Path Optimization for Enhanced Efficiency and Safety in Autonomous Vehicle Networks

Abstract

In this paper, we propose a decentralized algorithmic framework designed to enhance both the efficiency and safety of fully autonomous vehicle networks (AVNs) through dynamic path optimization. By using a peer-to-peer communication paradigm, vehicles exchange real-time positional, velocity, and environmental data to collaboratively predict and preempt potential collisions. By integrating advanced predictive modeling with dynamic rerouting strategies, the framework mitigates traffic bottlenecks and reduces collision risks without reliance on centralized control. Simulation-based evaluations indicate significant improvements in latency and throughput (the rate of vehicle flow throughout the network), particularly in dense urban and high-speed highway environments. This work represents a notable advancement in autonomous vehicle coordination, addressing key challenges in traffic management and collision prevention.