Artificial Intelligence, Machine Learning and Robotics to Support International Nuclear Safeguards In-Field Inspection Activities

Year
2024
Author(s)
Heidi Smartt - Sandia National Laboratories
Sydney Dorawa - Sandia National Laboratories
Shiloh Elliott - Sandia National Laboratories
Philip Honnold - Sandia National Laboratories
Zahi Kakish - Sandia National Laboratories
Adithya Ramakrishnan - Sandia National Laboratories
Tania Rivas - Sandia National Laboratories
Nathan Shoman - Sandia National Laboratories
Kyle A. Williams - Sandia National Laboratories
Abstract
Sandia National Laboratories has designed and is developing Inspecta (International Nuclear Safeguards Personal Examination and Containment Tracking Assistant), an artificial intelligence (AI)-enabled smart digital assistant with robotic support. The goal is to provide inspectors an in-field digital assistant that can perform tasks identified as tedious, challenging, hazardous, or prone to human error. Sandia has selected a high-impact inspection task, seal examination, for the initial prototype and is developing skills that will reside on a smartphone, Boston Dynamic’s Spot robot, or both, in support of this task. Skills developed for the on-device component include optical character recognition (OCR) for seal identifiers, task tracking, OCR to ingest documents, information recall that supports natural question/answer structure of the OCR-ingested documents and transcription. The addition of the Spot robot allows capabilities like direct voice commanding of the robot to perform tasks, autonomy for specific tasks, object detection to locate containers and their seals, and facility mapping. This community-organized session summarizes the Inspecta project, and provides detailed progress to date on Inspecta’s skills, including both on-device capabilities and Spot capabilities. Demonstrations (including some hands-on participation from the audience) will showcase OCR, information recall, direct voice commanding of Spot, Spot object detection, Spot task autonomy and mapping.