Mobile Compton Cameras for Critical Infrastructure Surveillance – RED-7D

The objective of RED-7D project is to design a multi-task mobile Compton camera loaded on a rover aimed at reconstructing 7D scenes (3D visible environment augmented with 3D nuclear source localization in space and 1D in energy). Such augmented facility could be considered in deploying solutions for the challenging problem of decommissioning of nuclear facilities, surveillance of critical infrastructures and assessing environmental contamination. Reconstruction of time-varying radiation maps, while the rover is moving, is a challenging signal processing task. The Compton imaging is prone to artefacts and source localization ambiguity. Optimal motion control and robust statistical signal processing are needed in order to ensure a high-resolution reconstruction of the radiation map.

Based on the concept of temporal Compton camera, the main scientific challenge of RED-7D project is to design an integrated approach in order to efficiently mapping the 7D-environment in real-time, by fusing the following aspects:

o Compton camera measurements and nuclear source reconstruction while moving the robot: novel algorithms should be designed in order to process mobile cones output by Compton camera detectors.

o A coarse-to-fine approach to reconstruct the 3D radiation map: In a first step, the mobile Compton camera covers the whole investigated region with a predefined trajectory. The predefined trajectory is the output of a multi-objective optimization algorithm aiming at minimizing the inter-visit durations, maximizing the Gamma ray acquisition time and ensuring the return of the robot to its charging desk before battery runs out. In a second step, the mobile robot will focus on detected radiation sources considered as region of interest (ROI), with an optimized trajectory around the detected source for precise and high-resolution reconstruction.

o Fusion of visible camera and Compton camera information in order to output an augmented 7D map where the radiation data are projected on the visible 3D map, which is of great help for operators. Also, considering the Compton camera information to automatically control the robot and better cover the contaminated zones makes our approach different from the conventional SLAM approaches.