Daniel Wurmser
U.S. Department of State, AVC/VPO
Washington, DC 20520

Bottom line: As Department of State COR and Program Manager, I provide sponsor approval for publication of the referenced article.

More complete review of the article and the research it describes:

The scope of arms control and non-proliferation treaties and agreements has historically been constrained by need to reliably verify compliance. In other words, cheating has to be detected before clandestine nuclear activities advance to the point where they can adversely impact United States’ national security equities. It is extraordinarily difficult to reliably monitor most nuclear activities using standoff detection techniques, so the ability to verify compliance with nuclear non-proliferation and arms control agreements has depended on the level of cooperative access that could be negotiated and the technology that could be deployed once such access is achieved. As new managed access arrangements are conceived and the accompanying enabling technologies are developed, policymakers are empowered with new arms control options.

Onsite inspections by trusted human inspectors have been a key verification component of a variety agreements ranging from the Nuclear Non-Proliferation Treaty (NPT) to a series of bilateral arms control agreements between the United States and the Soviet Union and subsequently Russia. In the future, robotic systems may possess advantages over human inspectors, since they can be permanently deployed to a facility, and be programmed and designed to only observe and/or report specifically permitted types of information. Furthermore, robot inspectors can be deployed under adverse conditions that may compromise the efficacy of and/or raise safety concerns regarding human inspectors. The U. S. Department of State supported the efforts of Princeton University and the Princeton Plasma Physics Lab (PPPL) to advance the foundational technology base that will bring this vision closer to reality. This investment has paid off handsomely. This article does a very good job of describing the progress made to date and the expected next steps.

As the title of the article implies, a key goal of future nuclear arms control agreements may be to limit tactical warheads and warheads in storage. Onsite inspection regimes would help verify compliance with such agreements by locating and counting nuclear devices at a given facility. There is also a need for similar smart detectors to provide reliable and cost-effective implementation of nuclear safeguards agreements under the NPT. The Princeton University/PPPL effort developed innovative highly-sensitive directional radiation detectors, mounted them on a robot, and demonstrated the ability of the assembly to “identify the source of nuclear radiation and whether it has been shielded to avoid detection.” This represents a significant technical milestone.

The effort has set the table for transition work to proceed (at least partly) under Department of Energy funding. One significant initiative now in process involves a visit to DOE’s Savannah River Site nuclear fuel fabrication plant to explore the possibility of testing the inspector robot under realistic conditions to demonstrate its utility as a “cost-effective means to provide effective and efficient verification” of compliance with fuel-cycle operations under the NPT.

As noted in the article, “The key challenge in looking for undeclared activities is to make a wise selection between continuing to explore versus stopping to exploit small signals that could be indicative of misbehavior.” A swarm of inspector robots capable of addressing this challenge will depend on the “design of machine learning (artificial intelligence) software to guide the robots, and technology to enable the robots to communicate with one another during inspections.” Principal Investigator Naomi Leonard, a highly-respected expert in the field, will lead this effort. This promises to be an exciting study in multi-agent communication and task coordination in an important (and challenging) robotic monitoring environment. Advances in such cooperative agent algorithms will find application in a variety of areas, and this research may inform a wide variety of future arms control and verification approaches.

The article therefore provides a nice snapshot of progress in a very successful research initiative, and it foreshadows solid future transition work.