An automated heterogeneous robotic system for radiation surveys: Design and field testing

Abstract

During missions involving radiation exposure, unmanned robotic platforms may embody a valuable tool, especially thanks to their capability of replacing human operators in certain tasks to eliminate the health risks associated with such an environment. Moreover, rapid development of the technology allows us to increase the automation rate, making the human operator generally less important within the entire process. This article presents a multirobotic system designed for highly automated radiation mapping and source localization. Our approach includes a three‐phase procedure comprising sequential deployment of two diverse platforms, namely, an unmanned aircraft system (UAS) and an unmanned ground vehicle (UGV), to perform aerial photogrammetry, aerial radiation mapping, and terrestrial radiation mapping. The central idea is to produce a sparse dose rate map of the entire study site via the UAS and, subsequently, to perform detailed UGV‐based mapping in limited radiation‐contaminated regions. To accomplish these tasks, we designed numerous methods and data processing algorithms to facilitate, for example, digital elevation model‐based terrain following for the UAS, automatic selection of the regions of interest, obstacle map‐based UGV trajectory planning, and source localization. The overall usability of the multirobotic system was demonstrated by means of a 1‐day, authentic experiment, namely, a fictitious car accident including the loss of several radiation sources. The ability of the system to localize radiation hotspots and individual sources has been verified.

Real-Time Gamma Radioactive Source Localization by Data Fusion of 3D-LiDAR Terrain Scan and Radiation Data from Semi-Autonomous UAV Flights

“…UGVs are more flexible, yet compared to UAVs, they are relatively slow and unfavorable terrain conditions might limit their operational scope. Especially in outdoor environments, this implies either frequent human intervention or elaborate schemes to identify traversable areas as, for example, demonstrated in [ 7 ]. Moreover, airborne systems for ground observation missions typically carry radiation detection equipment underneath, so that shielding effects by the vehicle itself can be eliminated.…”

“…Localization, on the other hand, also requires a (statistical) model to estimate source location(s) based on spatially scattered observations. It can be performed by exhaustively surveying the search area and estimating a posteriori likely source locations [ 4 , 5 , 7 , 19 , 20 , 21 ], or by applying greedy algorithms updating the current belief about source locations iteratively and adjust the measurement device’s trajectory accordingly [ 22 , 23 ]. The main advantage of greedy algorithms is their potential to significantly reduce mission duration.…”

“…Explicit localization methods, on the contrary, address the problem of finding a point source directly [ 1 , 2 ]. Existing systems and methods utilize micro UAVs [ 3 ] and UAVs [ 4 , 5 , 6 ] at low altitudes <10 m or in combination with ground vehicles [ 7 ] to detect and localize point sources.…”

“…For practical applications, a quick result is of great importance. However, data are often analyzed after the survey has been completed, e.g., in [ 5 , 7 ], which inevitably leads to delays. So far, only a few research results with real-time ability have been presented, e.g., in [ 3 , 4 ].…”

Development of a mobile robot for safe mechanical evacuation of hazardous bulk materials in industrial confined spaces

“…These works are applied research with problem solving nature for performing risky tasks to prevent threats to human health in unstructured outdoor environments. Some of these robots can be mentioned such as robot for nuclear radiation detection (Christie et al, 2017; Gabrlik et al, 2021; Miller et al, 2015; Oka & Shibanuma, 2002; Oomichi et al, 2007; Qian et al, 2012), fire‐fighting robot (Miyazawa, 2012; Pack et al, 2004; Wang et al, 2017), robot for urban search and rescue (Bishop et al, 2005; Enayati & Najafi, 2011; Habibian et al, 2021; Hirose & Fukushima, 2012; Kamegawa et al, 2011; Kamegawa et al, 2020; Li et al, 2009; Masuda, 2012; Matthies et al, 2002; Park et al, 2012; Zhang et al, 2006), cleaning robot for large diameter sewers (Walter et al, 2012), rescue robot for nuclear sites (Guzman et al, 2016; Hosoda et al, 2002; Nagatani et al, 2013), search and rescue robot for coal mines (Wang et al, 2014; Zhai et al, 2020), robot for sampling and radiological characterization in nuclear fields (Ducros et al, 2017; Kobayashi et al, 2012), robot for inspection and monitor of nuclear facilities (Bird et al, 2019; Fulbright & Stephens, 1995; Isozaki & Nakai, 2002; Yuguchi & Satoh, 2002), military rescue robot for casualty extraction task (Choi et al, 2019) and rescue robot for explosion accidents in coal mines (Li et al, 2020). A general unified outcome has been gained by the former cited research indicates existence of limitations for application of various types of sensors and actuators in robot development based on nature and magnitude of hazards (Morris et al, 2006).…”