Newswise — The U.S. Department of Energy’s (DOE) Office of Science has selected four National Laboratory-led projects to drive breakthroughs in advanced robotics and automation for scientific infrastructure. The DOE National Laboratories house and operate some of the world’s most advanced scientific instruments, computing systems, and experimental facilities. These selected projects aim to rapidly accelerate progress in advanced robotics and automation, tailoring advancements to these unique research settings and creating solutions that can be applied across a broad spectrum of autonomous scientific operations.
DOE expects these projects will create open software and interfaces, reusable robot skills, digital-twin environments, benchmark tasks, datasets and trained models, safety practices, provenance records, and training resources. The objective is not simply to produce four impressive robotics demonstrations, but to establish capabilities that can be reused and extended by the broader DOE scientific community. DOE prioritized projects that would build reusable infrastructure, advance learning-enabled autonomy, demonstrate transfer across instruments or laboratories, integrate robotics with Super Intelligence (SI) and advanced computing, and complement the other selected projects.
The selected projects are:
Modular Autonomous Experimentation through Self-improving Testbeds for Robotic Operations (MAESTRO) – Led by Argonne National Laboratory, MAESTRO will develop modular, self-improving infrastructure for laboratory robotics. Its approach combines digital twins and world models, reusable robot skills, agent-based orchestration, safety mechanisms, and learning from operational experience. A central robotics sandbox and multiple science-facing proving grounds will be used to evaluate whether robotic capabilities can improve through use and transfer across instruments, workflows, and laboratory environments.
DOE AI Robotics Testbed to Generalize Autonomous Science (DART) – Led by Brookhaven National Laboratory, DART will focus on determining when robotic autonomy is reliable, transferable, and scientifically valid. The project will develop adversarial digital counterparts, a verified failure atlas, multimodal robot-learning methods, and approaches for transferring task knowledge across different robotic platforms. DART will evaluate not only whether a robot completes a motion, but whether the resulting physical and scientific outcomes meet the requirements of the experiment.
Testbed for Robotics and Autonomy in Connected Experiments (TRACE) – Led by Oak Ridge National Laboratory, TRACE will build on Oak Ridge’s in-house INTERSECT automated laboratories ecosystem to connect robots, scientific instruments, SI agents, digital twins, data systems, and computing resources through standardized interfaces. The project will develop typed capability contracts, reusable sensorimotor skills, end-to-end provenance, safety and mixed-fidelity testing, and independently rerunnable evaluation methods. Its emphasis is on enabling autonomy solutions to be tested, reproduced, and deployed across different laboratories and scientific domains without rebuilding the complete software stack.
A Source-to-Discovery Platform for Instrumentation, Robotics, and Embodied AI in DOE Photon-Science Facilities (SPIRE) – Led by SLAC National Accelerator Laboratory, SPIRE will integrate physical sample manipulation, intelligent detectors, accelerator and instrument controls, persistent experiment state, and edge-to-high-performance computing. The project is designed to demonstrate source-to-discovery autonomy at photon- and electron-science facilities, where decisions must be made across timescales ranging from microseconds to hours. SPIRE will also develop methods, interfaces, digital twins, and benchmark tasks intended for adaptation at other DOE scientific User Facilities.
Collectively, the four projects address different but connected stages of the autonomous-science lifecycle:
- MAESTRO: develop capabilities that learn and improve through experience;
- DART: stress-test those capabilities and determine whether they transfer reliably;
- TRACE: connect, package, and reproduce capabilities across laboratories; and
- SPIRE: deploy and validate autonomy in consequential scientific-facility operations.
Funded through the Advanced Scientific Computing Research program under the solicitation on Robotics and Automation Testbeds for Autonomous Scientific Discovery (LAB 26-3601), the total funding for these projects is $30 million, with $2 million in Fiscal Year 2026, and outyear funding contingent on congressional appropriations. See more information about the selections on the Office of Science funding page.
Selection for award negotiations is not a commitment by DOE to issue an award or provide funding. Before funding is issued, DOE and applicants will undergo a negotiation process, and DOE may cancel negotiations and rescind the selection for any reason during that process.
Leave a comment