An arXiv preprint presents a robot-fleet toolkit through demonstrations that span virtual robots, a physical robot sensing a virtual environment and a mixed fleet combining both. One test placed a physical Jackal in a fleet with virtual robots; the machine was reported to navigate toward an assigned goal while avoiding obstacles and the simulated robots.
The system, called VirTooS, is a modular Python 3 and C# toolkit built on ROS 2, ChoiRbot and Unity. The paper describes it as a way to support fleet-management tasks in virtual and mixed-reality environments, but the evidence is a set of qualitative demonstrations rather than a quantified comparison.
A test bench for robot teams
VirTooS has three software layers. The first handles optimization-based team management; the second covers control, navigation and planning for an individual robot; and the third provides a Unity interface for virtual sensing and mixed-reality simulation.
It integrates SLAM Toolbox for mapping with Nav2 for local replanning and collision-avoidance support. It also supports customizable virtual environments, virtual and real sensors and containerized deployment across machines.
The reported setup used ROS 2 over a standard 5 GHz Wi-Fi network, Dockerized runs on Ubuntu 24.04, an AMD Ryzen 7 7800X3D workstation with an Nvidia GeForce RTX 5090, and Vicon pose tracking.
Three ways to run the system
In the first scenario, four simulated Clearpath Jackal robots performed dynamic task assignment in an indoor virtual environment with obstacles. Nav2 handled control and local replanning, while ros2-slam-toolbox updated the shared map.
The second scenario used a real Jackal in a laboratory while it sensed a virtual environment. Vicon pose information and virtual laser measurements were exchanged through the ROS-TCP-Connector.
The third scenario used four Jackals in a 4 m × 8 m room: one robot was physical and the other three were virtual. The experiment involved dynamic task assignment in mixed reality.
In that mixed-reality demonstration, the physical robot was reported to navigate toward its assigned goal while avoiding static obstacles and virtual robots. Its motion was limited between 20 and 30 seconds, followed by resumed navigation after target-assignment updates.
What the demonstrations establish
Taken together, the three scenarios cover virtual task assignment, physical-robot interaction with a virtual environment and mixed-reality task assignment. The authors interpret VirTooS as a potential bridge between algorithmic research and deployment.
The result is a feasibility demonstration, not a head-to-head test of one toolkit against another. Outcomes were described qualitatively, without standardized performance metrics or inferential statistical analysis.
The evidence is also limited to the described Jackal experiments in indoor virtual or mixed-reality settings. Open questions include how the system behaves as fleet size, environmental complexity or communication load increases, how accurate and timely virtual sensor and pose exchanges are relative to physical systems, and what failure modes or safety margins appear during prolonged mixed-reality operation. The demonstrations do not establish real-world fleet-management performance, safety or scalability beyond these settings.
Code and publication status
The source code was stated to be planned for public release on GitHub, but no repository URL or release status was provided in the supplied record. A video of an experiment was reported as available in the supplementary material.
The document is an arXiv preprint, version 1, dated 26 August 2026. The Space It Up project was funded by the Italian Space Agency and the Ministry of University and Research.
Paper data and sources
Original title: VirTooS: A ROS 2 - Unity Virtualization Toolkit for Fleet Management of Autonomous Mobile Robots
Authors: Andrea Drudi, Lorenzo Pichierri, Andrea Testa, Giuseppe Notarstefano
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text