Source term estimation of a hazardous airborne release using an unmanned aerial vehicle

Abstract

Gaining information about an unknown gas source is a task of great importance with applications in several areas, including responding to gas leaks or suspicious smells, quantifying sources of emissions, or in an emergency response to an industrial accident or act of terrorism. In this paper, a method to estimate the source term of a gaseous release using measurements of concentration obtained from an unmanned aerial vehicle (UAV) is described. The source term parameters estimated include the three‐dimensional location of the release, its emission rate and other important variables needed to forecast the spread of the gas using an atmospheric transport and dispersion model. The parameters of the source are estimated by fusing concentration observations from a gas detector on‐board the aircraft, with meteorological data and an appropriate model of dispersion. Two models are compared in this paper, both derived from analytical solutions to the advection–diffusion equation. Bayes’ theorem, implemented using a sequential Monte Carlo algorithm, is used to estimate the source parameters to take into account the large uncertainties in the observations and formulated models. The system is verified with novel, outdoor, fully automated experiments, where observations from the UAV are used to estimate the parameters of a diffusive source. The estimation performance of the algorithm is assessed subject to various flight path configurations and wind speeds. Observations and lessons learned during these unique experiments are discussed, and areas for future research are identified.

Particle Filtering With Soft State Constraints for Target Tracking

“…The developed algorithms in this paper can be applied to a wide range of areas where the systems under investigation are nonlinear and/or non-Gaussian, and the state vectors are subject to soft constraints. Examples may include ground vehicle tracking, air traffic monitoring, maritime navigation, and the other areas beyond target tracking such as networked systems [36] and source term estimation [37]. Finally, we point out that, for those applications where a continuous-time dynamic system is discretized, the information provided by soft state constraints can be applied in a higher sampling rate than the sensor measurements, and hence having a potential to further increase the filtering performance.…”

Unmanned Aerial Vehicle-Based Hazardous Materials Response: Information-Theoretic Hazardous Source Search and Reconstruction

“…In this work, a model derived from analytical solutions to the advection diffusion equation given various assumptions is used, which shall be referred to the Isotropic plume (IP) model as verified in [2]. The model is fast running and based on the assumption of a steady state plume with a consistent mean wind velocity, source strength, and turbulent conditions.…”

“…In addition, the experimental assessment of the information theoretic search algorithm using a UAV is the first of its kind, paving the way to a deployable system given some further improvements to a few of the components. The results show a significant reduction in search time of using the proposed information theoretic planner when compared to that of a pre-planned uniform flight path [2]. Finally, the value of the system is demonstrated in realistic scenarios in addition to further demonstrations of the inference engine using a manual and a preplanned flight for data collection.…”

“…An unmanned aerial system (UAS) can supplement an existing network, or replace it entirely, by intelligently gathering spatio-temporally indexed concentration measurements of the release. There is limited research in the area of STE using a UAS, however, it has recently been validated in an experimental study previously conducted by the authors [2]. In this study a systematic flight pattern was executed to gather the HAZMAT data which were used in a Bayesian inference based estimation algorithm to estimate the source parameters; taking into consideration the large amount of uncertainty in the measurements from the sensor and the atmospheric dispersion process.…”

“…Experiments in more realistic natural environments are limited. Most related to this article is the previous work conducted by the authors [2]. A UAV executed a pre defined flight path to collect measurements from the onboard atmospheric sensor which were used to estimate the source term of an atmospheric release using a sequential Monte Carlo algorithm.…”