Fire as a Service: Augmenting Robot Simulators with Thermally and Visually Accurate Fire Dynamics

Anton R. Wagner1*, Madhan B. Rao2*, Helge Wrede1, Sören Pirk1, Xuesu Xiao2
1VCAI Lab, Kiel University, Germany   2RobotiXX Lab, George Mason University, USA
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2026
*Indicates Equal Contribution
A Spot robot in an Isaac Sim warehouse environment affected by multiple fires, shown from two viewing angles.

Fire as a Service (FaaS) augments existing robot simulators with thermally and visually accurate fire dynamics. Here we show a Spot robot in an Isaac Sim warehouse environment affected by multiple fires from two different viewing angles.

Abstract

Most existing robot simulators prioritize rigid-body dynamics and photorealistic rendering, but largely neglect the thermally and optically complex phenomena that characterize real-world fire environments. For robots envisioned as future firefighters, this limitation hinders both reliable capability evaluation and the generation of representative training data prior to deployment in hazardous scenarios. To address these challenges, we introduce Fire as a Service (FaaS), a novel, asynchronous co-simulation framework that augments existing robot simulators with high-fidelity and computationally efficient fire simulations. Our pipeline enables robots to experience accurate, multi-species thermodynamic heat transfer and visually consistent volumetric smoke without disrupting high-frequency rigid-body control loops. We demonstrate that our framework can be integrated with diverse robot simulators to generate physically accurate fire behavior, benchmark thermal hazards encountered by robotic platforms, and collect realistic multimodal perceptual data. Crucially, its real-time performance supports human-in-the-loop teleoperation, enabling the successful training of reactive, multimodal policies via Behavioral Cloning. By adding fire dynamics to robot simulations, FaaS provides a scalable pathway toward safer, more reliable deployment of robots in fire scenarios.

Video

Asynchronous Fire Co-Simulation

FaaS treats fire as an external service that evolves independently in a high-fidelity combustion solver, rather than baking it into the robot's control loop. We build on Fire-X, a GPU-accelerated hybrid solver with multi-species thermochemistry, validated against real compartment-fire experiments and NIST's Fire Dynamics Simulator. The robot simulator streams camera pose, depth, and scene geometry over ROS; the fire simulator rolls out thermodynamic dynamics, renders a viewpoint-consistent alpha-matted image of flame and smoke, and composites it back onto the robot's RGB observations. Because the two run asynchronously and non-blocking — each always using the latest available data — the robot's control loop is never paused, keeping end-to-end latency low enough for real-time augmentation and teleoperation.

FaaS architecture: a robot simulator and an external fire simulator exchange pose, depth, RGB, and composite renderings over ROS in a non-blocking loop.

The system asynchronously couples a conventional robot simulator (red) with an external fire simulator (orange) via a non-blocking ROS data bridge. Camera pose, depth, and RGB are published best-effort; the fire simulator renders a viewpoint-consistent, alpha-matted flame/smoke image that is composited back onto the robot's sensors.

FaaS fire renderings superimposed on Gazebo (left) and MuJoCo (right) simulations.

Engine-agnostic by design: besides Isaac Sim, FaaS superimposes scene-aware fire on established simulators such as Gazebo (left) and MuJoCo (right).

Key Capabilities

  1. Thermally accurate hazard modeling that quantifies heat radiation and cumulative thermal risk to robotic hardware.
  2. Visually consistent fire and smoke dynamics that augment camera-based perception in a physically grounded manner.
  3. A high-performance co-simulation architecture with temporal resolution sufficient for both human-in-the-loop teleoperation and high-frequency reactive safety controllers, enabling reactive, thermally-informed policies trained via Behavioral Cloning.
  4. Engine-agnostic interoperability, with seamless integration demonstrated across Isaac Sim, Gazebo, and MuJoCo.

Results

BibTeX

@misc{wagner2026faas,
  title         = {Fire as a Service: Augmenting Robot Simulators with Thermally and Visually Accurate Fire Dynamics},
  author        = {Wagner, Anton R. and Rao, Madhan B. and Wrede, Helge and Pirk, S{\"o}ren and Xiao, Xuesu},
  year          = {2026},
  eprint        = {2603.19063},
  archivePrefix = {arXiv},
  primaryClass  = {cs.RO},
  url           = {https://arxiv.org/abs/2603.19063}
}