Realizable Deployment of Limited-Knowledge Robotic Inspectors for Nuclear VerificationDownload PDF

Anonymous

19 Apr 2023 (modified: 25 May 2023)ICRA 2023 Workshop Lab-to-Real Gap Blind SubmissionReaders: Everyone
Keywords: verification, radiation detection, statistical methods
TL;DR: Verifying the absence or presence of sources subject to information constraints.
Abstract: This extended abstract underscores the prospect and fundamental challenges of deploying robotic radiation detectors in nuclear safeguards and arms control applications, where safety and security concerns are of utmost importance. We propose a performant, minimal-knowledge approach which addresses information security constraints, with specific emphasis on the challenge of confirming the absence of undeclared radioactive sources. We demonstrate a random walk process which requires no a priori knowledge of the environment and certifies the absence or presence of sources without revealing any a posteriori information. We improve the convergence of this method by incorporating directional radiation measurements. The finite time and physical safety limitations may be addressed by limited, rather than minimal, knowledge approaches. The methods discussed illuminate a spectrum of limited-information contexts which have not yet been thoroughly explored in this application space.
0 Replies

Loading