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

Hazardous materials (HAZMAT) released into the atmosphere pose both an immediate and chronic risk to human health. Characteristic examples include the Sarin gas terrorist attacks in Japan (1995), the infamous chemical accidents of Bhopal, India (1984) and Seveso, Italy (1976), such nuclear disasters as Fukushima (2012), and the recent use of chemical weapons and nerve agents in Syria (2013-2018). A prompt and accurate prediction of the whereabouts of the HAZMAT and a forecast of its dispersion and deposition are important to enable responders to undertake appropriate mitigation strategies and extract people from affected regions. Hazard predictions, however, require accurate knowledge of the release parameters (the so-called source term), as well as the local meteorological information. In many situations, this information is unknown or highly uncertain. HAZMAT sensor readings will indicate the presence of HAZMAT, and this must be turned rapidly into a warning to ensure the safety of personnel in the vicinity. This currently requires either a static network of pre-deployed sensors, which can be costly and necessitates substantial planning, or the manual collection of sensor measurements, e.g., using handheld devices and HAZMAT suits, which can be time-consuming and places personnel at risk.