A low-cost Framework for Decentralized Autonomous Intersection Management (2311.17681v2)

Abstract: This paper addresses the traffic management problem for autonomous vehicles at intersections without traffic signals. In the current system, a road junction has no traffic signals when the traffic volume is low to medium. Installing infrastructure at each unsignalled crossing to coordinate autonomous cars can be formidable. We propose a novel low-cost solution strategy where the vehicles use a harmony matrix to find the best possible combination of the cars to cross the intersection without any crashes. The harmony matrix defines the connection between different vehicle maneuvers and is queried online for intersection management. We maximize the throughput of the intersection by solving a maximal clique problem formulated based on the vehicles present at the intersection. The proposed algorithm relies on the intent perceived by the autonomous vehicles. We compare our work with a communication-based strategy that uses V2I communication protocols, and through extensive simulation, we showed that our algorithm is comparable when the traffic volume is less than 500 PCUs/hr/lane.

• The paper introduces an algorithm that leverages a harmony matrix to enable autonomous vehicles to determine safe crossing sequences.
• It applies graph theory and solves a maximal clique problem to maximize simultaneous crossing and reduce waiting times.
• The framework prevents deadlocks and performs effectively in simulations for unsignalized intersections in low to medium traffic conditions.

Decentralized Autonomous Intersection Management: A Low-Cost Strategy

In recent advancements of traffic management for autonomous vehicles, Rugved Katole and Arpita Sinha from the Systems and Control Engineering Department of the Indian Institute of Technology, Bombay, have developed a novel and cost-efficient framework for Decentralized Autonomous Intersection Management (DAIM). This article offers a comprehensive technical summary and evaluation of their framework, with particular emphasis on its practical and theoretical implications.

Overview

The paper addresses the critical issue of managing intersections without traffic signals, especially in areas with low to medium traffic volume. The authors propose a low-cost, infrastructure-independent solution where autonomous vehicles utilize a harmony matrix to preemptively determine optimal crossing sequences, thereby avoiding collisions. This strategy maximizes the intersection throughput by solving a maximal clique problem tailored to the current traffic scenario.

Key Contributions

1. Algorithm for Decision-Making: The core of the proposed method lies in an innovative algorithm that allows autonomous vehicles to make decentralized decisions at unsignalized intersections. Each vehicle uses on-board sensors to detect and interpret the intended maneuvers of other vehicles. The harmony matrix is pivotal in this process, forming an adjacency matrix that captures the connectivity between non-conflicting vehicle maneuvers.
2. Graph Theory Application: The decision-making process is formalized by constructing a graph where nodes represent vehicle maneuvers, and edges indicate non-conflicting actions determined by the harmony matrix. Solving the maximal clique problem identifies the largest subset of vehicles that can simultaneously and safely cross the intersection.
3. Deadlock Prevention Mechanisms: The algorithm is designed to prevent deadlocks by ensuring that vehicles adhere strictly to the computed crossing sequence. In rare occasions where a new vehicle might entering the intersection disrupts the decision-making process, a predefined safety buffer zone enforces proper right-of-way.

Practical Implications

The proposed framework demonstrates robust performance in simulations using the SUMO traffic simulator, thus validating its potential applicability in real-world scenarios. By eliminating the need for expensive infrastructures like traffic signals or V2I communication systems, DAIM serves as an efficient and economical solution particularly suitable for rural or low-traffic areas.

Evaluation and Comparative Analysis

The algorithm's efficacy was subjected to thorough evaluation in various simulation settings, including 3-way, 4-way, and 5-way intersections with different traffic distributions. Key performance metrics included average waiting time and travel time:

• Comparative Metrics: The DAIM algorithm showed lower waiting times and competitive travel times compared to fixed-time traffic signals (FTS) and adaptive traffic signals (ATS). It performed comparably to communication-based V2I protocols under low-traffic conditions but without the associated infrastructure costs.
• Balanced vs. Unbalanced Traffic: Interestingly, DAIM showed enhanced performance in unbalanced traffic scenarios due to its priority-based approach, which mitigates queues more effectively compared to balanced traffic conditions where no single lane predominates.

Theoretical Implications

From a theoretical perspective, the application of graph theory to solve the intersection management problem reflects an elegant synthesis of control systems and optimization. The deployment of a harmony matrix and maximal clique problem underlines the algorithm's robustness and adaptability to dynamic traffic scenarios.

Future Directions

Future work can delve into extending this framework to mixed-autonomy environments where both human-driven and autonomous vehicles coexist. Additionally, integrating advanced confidence models to manage uncertainties in vehicle intent detection will further enhance the algorithm's robustness and reliability.

In conclusion, this paper presents a significant stride towards efficient and cost-effective intersection management, exploiting decentralized decision-making and graph-theory-based optimization to ensure seamless and safe traffic flow at unsignalized intersections.