Mr.MSTE: Multi-robot Multi-Source Term Estimation with Wind-Aware Coverage Control

TL;DR

The paper tackles multi-source term estimation in dynamic atmospheric dispersion by marrying a physics-informed, hybrid Bayesian inference framework with wind-aware, coverage-based robot deployment. It introduces a joint multi-source probability density over an unknown number of sources, with deterministic birth/death/merge transitions and diffusion-based source evolution, implemented via a sequential Monte Carlo (particle filter) approach. A wind-aware generalized Voronoi partition guides robot placement by incorporating plume anisotropy, enabling efficient, scalable sensing for multiple sources. The framework is validated through extensive Monte Carlo simulations and real-world CO$_2$ experiments with mobile robots, showing faster convergence, improved source separation, and reduced uncertainty compared to traditional coverage strategies and static sensor networks. These results indicate practical viability for emergency response and environmental monitoring using collaborative mobile sensing.

Abstract

This paper presents a Multi-Robot Multi-Source Term Estimation (MRMSTE) framework that enables teams of mobile robots to collaboratively sample gas concentrations and infer the parameters of an unknown number of airborne releases. The framework is built on a hybrid Bayesian inference scheme that represents the joint multi-source probability density and incorporates physics-informed state transitions, including source birth, removal, and merging induced by atmospheric dispersion. A superposition-based measurement model is naturally accommodated, allowing sparse concentration measurements to be exploited efficiently. To guide robot deployment, we introduce a wind-aware coverage control (WCC) strategy that integrates the evolving multi-source belief with local wind information to prioritize regions of high detection likelihood. Unlike conventional coverage control or information-theoretic planners, WCC explicitly accounts for anisotropic plume transport when modelling sensor performance, leading to more effective sensor placement for multi-source estimation. Monte Carlo studies demonstrate faster convergence and improved separation of individual source beliefs compared to traditional coverage-based strategies and small-scale static sensor networks. Real-world experiments with CO2 releases using TurtleBot platforms further validate the proposed approach, demonstrating its practicality for scalable multi-robot gas-sensing applications.

Figures (12)

• Figure 1: Example plot of gas concentrations distribution at different locations resulting from superposition of dispersion from three gas sources subject to a probability of detection
• Figure 2: Robot and sensor heading angles.
• Figure 3: Illustration of the proposed generalised Voronoi tessellation for different values of $\alpha$. The wind heading is $\psi=-90^\circ$ and a square marker is used to denote the robot locations.
• Figure 4: Illustrative run for the proposed wind-aware coverage control based MR-MSTE algorithm with $n=6$, $M_{\max}=4$, $\alpha=-0.75$, $v_w = 4m/s$ and $\psi=90^\circ$ and the black dashed line representing the generalized Voronoi cells corresponding to each robot.
• Figure 5: Variation of (\ref{['fig:rmse_uncertainity']}) GOSPA error along with $\sigma_{\Theta_k}$, and, (\ref{['fig:ObjectiveFunction']}) $\mathcal{H}_{\mathcal{V}}$ for the illustrative run presented in Fig. \ref{['fig:Illustrative_Run']}.

Theorems & Definitions (13)

• Definition 1
• Remark 1
• Remark 2
• Lemma 1
• proof
• Theorem 1
• proof
• Remark 3
• Theorem 2
• proof