Autonomous Parking of Vehicle Fleet in Tight Environments (1910.02349v3)

Abstract: The problem of autonomous parking of vehicle fleets is addressed in this paper. We present a system-level modeling and control framework which allows investigating different vehicle parking strategies while taking into account path planning and collision avoidance. The proposed approach decouples the problem into a centralized parking spot allocation and path generation, and a decentralized collision avoidance control. This paper presents the hierarchical framework and algorithmic details. Extensive simulations are used to assess several allocation strategies in terms of total fleet parking time and queue length. In particular, we describe how Braess's paradox can be observed for parking vehicle fleets.

• The paper presents a hierarchical framework that divides parking tasks between centralized allocation and decentralized collision avoidance to improve system scalability.
• Efficient offline path generation paired with an occupancy grid approach reduces real-time computation and ensures effective collision management.
• Simulation results demonstrate that specific strategies, like Interval-first and Farthest-first searches, can significantly lower Mean Task Time and Maximum Queue Length.

Overview of "Autonomous Parking of Vehicle Fleet in Tight Environments"

The development of efficient autonomous vehicle (AV) parking systems is critical in urban areas where available parking spaces are limited. The paper "Autonomous Parking of Vehicle Fleet in Tight Environments" by Xu Shen, Xiaojing Zhang, and Francesco Borrelli addresses this challenge by proposing a hierarchical system for managing the autonomous parking of large vehicle fleets. The authors present a structured and computationally efficient framework that separates parking tasks into centralized and decentralized components, facilitating scalable solutions to congestion in parking facilities.

Key Contributions

The authors highlight three primary contributions of their work:

1. Hierarchical Framework for Parking Management: The problem of parking a fleet of AVs is split into centralized and decentralized components. The central coordinator is tasked with parking spot allocation and path generation, while individual vehicles manage collision avoidance autonomously. This separation allows for efficient processing, alleviating computational burdens on individual vehicles.
2. Efficient Path Generation and Collision Avoidance: The framework incorporates offline path generation for parking maneuvers, reducing real-time computational requirements. Collision avoidance is managed using an occupancy grid approach, which decentralizes decision-making and enhances system scalability.
3. Optimal Allocation Strategies: By running extensive simulations, the authors assess various parking allocation strategies. They analyze the impact of spot assignment policies on key performance metrics such as Mean Task Time (MTT) and Maximum Queue Length (MQL).

Numerical Results

The proposed framework was tested through rigorous simulations, demonstrating how different allocation strategies affect parking efficiency. Key findings were:

• Mean Task Time (MTT): Simulations indicate that the Interval-first Search (IS) strategy minimizes MTT, particularly when parking intervals are optimally set around four spaces. This strategy allows vehicles sufficient space for simultaneous maneuvering, reducing overall parking time.
• Maximum Queue Length (MQL): For minimizing queue lengths, the Farthest-first Search (FS) strategy was effective, particularly with both lanes open. This finding suggests parking vehicles further from the entrance distributes them better across the facility, reducing external queuing.
• Braess's Paradox Analogy: An intriguing finding was that opening more lanes sometimes degraded performance—akin to Braess's paradox. This counterintuitive outcome suggests that increasing parking avenues without strategic arrangement can exacerbate congestion and impinge on parking efficiency.

Implications and Future Directions

This work has multiple implications for autonomous parking, both in practical implementations and theoretical advancements. Practically, the proposed solutions can optimize parking operations in urban car parks, improving user experience and minimizing traffic disruptions. Theoretically, the paper introduces new dimensions to optimize AV interactions and path-planning strategies in dynamic, constrained environments.

Future work could explore implementing the framework in real-world settings, considering additional factors such as vehicle heterogeneity and communication latencies. Expanding the model to include dynamic parking fees or reward mechanisms could also be investigated as they present further opportunities for optimization in real-time scenarios.

Overall, the paper provides a comprehensive approach to tackling the complex challenge of autonomously managing vehicle fleets within restricted parking environments, presenting relevant insights and methodologies crucial for developing scalable AV systems.