Abstract
This paper presents a novel Gaussian Processes - Peak Suppression Particle Filter (GP-PSPF) method with adaptive weighting corrections, so as to identify sources in the multi-modal radiation field under some tough conditions, e.g. spatially sparse measurements and sources with large strength differences. As the radiation cumulative effect and ambiguous source number, most existing methods fail to localize the hotspots clustered in narrow regions, and PSPF scheme overcomes these difficulties through multi-layer structure and peak-suppressed correction. In contrast to our earlier work, the proposed algorithm mainly focuses on more severe and practicable conditions, as well as accuracy and robustness improvement. Firstly measurement biases are adopted as the correction feedback through Gaussian Processes technique, and then strength deviation for each particle can be inferred and utilized in two dynamic modules. The dynamic peak-suppressed correction is implemented to achieve more accurate estimations, while the location correction focuses on the solution of location dilemmas, consisting of redundant source identification and less swarm clustering. In addition, scaling adaptation policy and sequential swarm reordering are specially conceived and developed for more stable and accurate optimization. Finally, extensive simulations and physical experiment are conducted under above-mentioned intractable situations, validating the accuracy improvement and practical effectiveness of the algorithm.
Highlights
Owing to the irreversible damage of ionizing radiation to human body, autonomous robots have gradually become substitutions for intervention tasks in facility decommission and nuclear accidents [1]–[4]
The GP-based weighting modules are incorporated into the Gaussian Processes - Peak Suppression Particle Filter (GP-PSPF) scheme to address the following problems: i) the co-existence of radiation sources with large strength differences; ii) location mismatching and redundant source identification under extreme sparse measurements; iii) less swarm clustering for the cancellation effect among multiple sources
The prediction procedure for the GP-PSPF framework is illustrated in Fig. 15, and the following facts can be observed: drastic state changes and belief fluctuations always occur in initial steps; but once two swarms aggregate in the local region, the confidence score and estimations accuracy would be promoted in a rapid speed
Summary
Owing to the irreversible damage of ionizing radiation to human body, autonomous robots have gradually become substitutions for intervention tasks in facility decommission and nuclear accidents [1]–[4]. The peak-suppressed offset and location correction factor for each particle can be dynamically adjusted by the strength deviation priori Through these varying correction modules, more accurate and robust results can be obtained in the multi-source estimation scheme, as well as rapidity performance achieved. The GP-based weighting modules are incorporated into the GP-PSPF scheme to address the following problems: i) the co-existence of radiation sources with large strength differences; ii) location mismatching and redundant source identification under extreme sparse measurements; iii) less swarm clustering for the cancellation effect among multiple sources. Location factor plays little effect on weighting correction, while strength factor works well to improve the estimated accuracy, as shown, 8(d) and 8(f) The Pseudo code of the GP-PSPF algorithm is shown in Algorithm 1
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
