Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DMV-AVP: Distributed Multi-Vehicle Autonomous Valet Parking using Autoware

Zubair Islam, Mohamed El-Darieby

TL;DR

This paper introduces DMV-AVP, a distributed simulation system for cooperative multi-vehicle autonomous valet parking built on the DMAVA framework. It adds two key components: the U-YOLO perception module for real-time, vision-based parking-spot detection and the Multi-Vehicle AVP Node for cross-host coordination, queuing, and reservation management, all synchronized via Zenoh across Autoware instances. The authors validate the approach on two- and three-host configurations, demonstrating deterministic coordination, conflict-free parking behavior, and scalable performance, while highlighting limitations such as centralized coordination bottlenecks, perception sensitivity to vehicle appearance, and hardware constraints. The work provides an open, extensible foundation for distributed AVP simulation, paving the way for hardware-in-the-loop validation and eventual real-world deployments.

Abstract

This paper presents the DMV-AVP System, a distributed simulation of Multi-Vehicle Autonomous Valet Parking (AVP). The system was implemented as an application of the Distributed Multi-Vehicle Architecture (DMAVA) for synchronized multi-host execution. Most existing simulation approaches rely on centralized or non-distributed designs that constrain scalability and limit fully autonomous control. This work introduces two modules built on top of the DMAVA: 1) a Multi-Vehicle AVP Node that performs state-based coordination, queuing, and reservation management across multiple vehicles, and 2) a Unity-Integrated YOLOv5 Parking Spot Detection Module that provides real-time, vision-based perception within AWSIM Labs. Both modules integrate seamlessly with the DMAVA and extend it specifically for multi-vehicle AVP operation, supported by a Zenoh-based communication layer that ensures low-latency topic synchronization and coordinated behavior across hosts. Experiments conducted on two- and three-host configurations demonstrate deterministic coordination, conflict-free parking behavior, and scalable performance across distributed Autoware instances. The results confirm that the proposed Distributed Multi-Vehicle AVP System supports cooperative AVP simulation and establishes a foundation for future real-world and hardware-in-the-loop validation. Demo videos and source code are available at https://github.com/zubxxr/multi-vehicle-avp

DMV-AVP: Distributed Multi-Vehicle Autonomous Valet Parking using Autoware

TL;DR

This paper introduces DMV-AVP, a distributed simulation system for cooperative multi-vehicle autonomous valet parking built on the DMAVA framework. It adds two key components: the U-YOLO perception module for real-time, vision-based parking-spot detection and the Multi-Vehicle AVP Node for cross-host coordination, queuing, and reservation management, all synchronized via Zenoh across Autoware instances. The authors validate the approach on two- and three-host configurations, demonstrating deterministic coordination, conflict-free parking behavior, and scalable performance, while highlighting limitations such as centralized coordination bottlenecks, perception sensitivity to vehicle appearance, and hardware constraints. The work provides an open, extensible foundation for distributed AVP simulation, paving the way for hardware-in-the-loop validation and eventual real-world deployments.

Abstract

This paper presents the DMV-AVP System, a distributed simulation of Multi-Vehicle Autonomous Valet Parking (AVP). The system was implemented as an application of the Distributed Multi-Vehicle Architecture (DMAVA) for synchronized multi-host execution. Most existing simulation approaches rely on centralized or non-distributed designs that constrain scalability and limit fully autonomous control. This work introduces two modules built on top of the DMAVA: 1) a Multi-Vehicle AVP Node that performs state-based coordination, queuing, and reservation management across multiple vehicles, and 2) a Unity-Integrated YOLOv5 Parking Spot Detection Module that provides real-time, vision-based perception within AWSIM Labs. Both modules integrate seamlessly with the DMAVA and extend it specifically for multi-vehicle AVP operation, supported by a Zenoh-based communication layer that ensures low-latency topic synchronization and coordinated behavior across hosts. Experiments conducted on two- and three-host configurations demonstrate deterministic coordination, conflict-free parking behavior, and scalable performance across distributed Autoware instances. The results confirm that the proposed Distributed Multi-Vehicle AVP System supports cooperative AVP simulation and establishes a foundation for future real-world and hardware-in-the-loop validation. Demo videos and source code are available at https://github.com/zubxxr/multi-vehicle-avp
Paper Structure (14 sections, 5 figures, 1 table)

This paper contains 14 sections, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Multi-Agent Simulations and Coordination Frameworks
  4. Vision-Based Parking Spot Detection
  5. System Design and Architecture
  6. Integration and Extension of the DMAVA Architecture
  7. System Workflow
  8. U-YOLO Module
  9. Multi-Vehicle AVP Node and State Machine
  10. Experiments and Results
  11. Experimental Setup
  12. AVP Lifecycle Execution and Perception Accuracy
  13. Distributed Validation and Scalability Testing
  14. Discussion and Conclusions

Figures (5)

  • Figure 1: Proposed seven-workflow layered architecture for the DMV-AVP System, extending the five-workflow DMAVA with additional Perception and Coordination workflows.
  • Figure 2: System-level workflow of the DMV-AVP System across two physical hosts.
  • Figure 3: AVP state machine diagram illustrating the vehicle lifecycle from arrival to completion across Drop-off, Parking, and Retrieval phases.
  • Figure 4: Custom AVP Panel in RViz displaying vehicle status, parking availability, and lifecycle controls for Drop-off, Parking, and Retrieval.
  • Figure 5: Three-host validation showing synchronized AVP operation across Hosts 1, 2, and 3, with all vehicles parked and system states aligned.