US Air Force Taps ARM Institute for Robotics

This new five-year cooperative agreement between the Air Force Research Laboratory (AFRL) and the ARM Institute has a combined funding ceiling of $87.66 million. The collaboration is designed to advance dual-use robotics and automation for both the commercial industrial base and military sustainment operations. The partnership will focus on specific technology areas including robotic mobility, multi-robot/multi-human teaming, advanced visualization, and manufacturing process informatics. This initiative will be executed through project calls led by the ARM Institute, which will facilitate collaboration among its consortium of over 450 members from industry, academia, and government. One ARM Institute-funded project, in partnership with the AFRL, involves Machina Labs' RoboCraftsman™ platform. This project will use AI and machine learning to automate the tool path generation for producing aircraft parts like airframe skins and panels, addressing a critical bottleneck in military aircraft readiness. The Air Force is already deploying robotics for maintenance tasks, such as the A5 robotic system, a 22,000-pound mobile robot that uses real-time sensor feedback for tasks like aircraft coating removal, which is expected to cut maintenance times by up to 50%. Other initiatives include using robotic systems for high-precision drilling and fastening on aircraft like the Boeing 787. A key focus for the ARM Institute is the application of multi-modal AI, leveraging foundation models that can process various types of data. This enables robots to better understand and interact with their environment, moving beyond single-task programming to more flexible and rapidly re-taskable manufacturing. For example, projects are exploring how robots can use vision systems to identify and handle diverse materials, like nested fabric pieces on a cutting table, by adapting their gripping strategy based on the geometry of the piece. The push towards embodied AI, where AI directly controls a physical system, is transforming manufacturing. Instead of just processing data, these systems use sensor feedback to generate a sequence of actions to achieve a physical goal, such as a robot sanding a part to a desired finish. Companies like Covariant are developing robotics foundation models, pre-trained on massive datasets, to give robots more generalized capabilities for the physical world, similar to how models like GPT have transformed language processing. This collaboration provides a direct pathway for innovations from ARM's diverse members, which include major defense contractors like Lockheed Martin and Northrop Grumman, robotics suppliers such as FANUC, and tech startups, to be integrated into Air Force operations. The institute's projects often bring together organizations that would typically be competitors to solve broad manufacturing challenges.