US Air Force Seeks AI to Coach Tanker Crews on Aerial Refueling
A Small Business Innovation Research solicitation asks for software that would review refueling video and give crews post-flight feedback, starting with the KC-46
By AXY Defense Desk ·
Key points
- The US Air Force wants AI software that analyzes tanker camera footage of aerial refueling and gives crews post-flight feedback.
- The request comes through a Small Business Innovation Research solicitation focused on the KC-46 tanker.
- The system would track the tanker, boom and receiver and measure position, velocity, alignment and closure rate.
- The Air Force says the aim is to reduce bias in human assessment and to build a prototype for wider AI-based aviation training.
- The Air Force says most remaining problems with the KC-46 remote vision system will soon be resolved.
The US Air Force wants artificial intelligence to help train the crews who refuel other aircraft in flight. According to an Air Force Small Business Innovation Research solicitation, the service is looking for a system that would use cameras mounted on tankers to automatically analyze refueling operations and give the tanker crews what the solicitation calls “hot wash” feedback.
A hot wash is a debrief held soon after a mission, while the details are still fresh. Crews talk through what went well and what did not, and instructors use that conversation to correct habits. The Air Force wants the same kind of review to happen faster and with more detail, using data collected from the aircraft itself.
Why Aerial Refueling Is So Hard to Judge
Aerial refueling, sometimes called air-to-air refueling, lets a tanker aircraft pass fuel to another aircraft while both are flying. It extends the range and time that fighters, bombers and other aircraft can stay airborne, which is why the mission is central to how the US military projects power across long distances. It is also among the most demanding jobs in aviation. The two aircraft have to fly close together, at matched speeds, for long periods, while small corrections to position and speed keep the connection safe.
The solicitation describes how training and review work today. In its words, “Current aerial refueling training and post-flight assessment processes rely heavily on instructor observation, pilot self-assessment, manual video review, and subjective evaluation of performance.” In plain terms, a crew learns how well a refueling went mostly by watching, remembering and talking it over. Each of those methods depends on the people involved, and people see the same event differently.
The Air Force says it hopes an AI-based assessment system will reduce those differences. The aim is to limit the biases that come with human judgment, so that the feedback a pilot or boom operator gets is based on measured data rather than on who was watching or how they remember the flight. For a crew, that could mean a clearer picture of a small drift in position that would otherwise go unnoticed until it became a habit.
How the Cameras and Software Would Work
The solicitation asks for a system that can “leverage advanced computer vision, machine learning, and aviation (flight dynamics) analytics techniques to automatically extract operationally relevant performance metrics from aerial refueling events, quantify deviations from desired flight profiles, and generate actionable post-flight feedback to improve aircrew proficiency, reduce human-factor-driven mishaps, and enhance mission readiness.”
In practice, computer vision means software that can recognize objects in video and follow them from frame to frame. The system would track the tanker, the refueling boom that extends from its tail, and the receiving aircraft that takes on fuel. It would then measure a set of flight data, including position, velocity, alignment and closure rate. Closure rate is how quickly the receiving aircraft is moving toward the tanker as the two come together. The software would compare each of those measurements against what the solicitation calls nominal parameters, meaning the values that a correct refueling should fall within, and flag where they drift.
The result would be a set of visual tools for the pilot of the receiving aircraft. The solicitation mentions “graphic AI overlays, heat maps, event timelines, and immersive or interactive debrief environments.” An overlay would place markers on the live or recorded image. A heat map would show where most of the deviations happened. An event timeline would let a crew step through a refueling moment by moment, rather than trying to recall it.
The system is also meant to connect with other training tools. The solicitation says it could work with virtual reality and augmented reality, simulators, mission rehearsal systems and other computer-based training. That matters because a crew that spots a problem in a debrief could then practice the same maneuver in a simulator before flying it again.
The KC-46 Remote Vision Problem
The solicitation focuses on the KC-46 tanker. The aircraft is built by Boeing and uses a remote vision system. That system lets an operator control the refueling boom using cameras and a three-dimensional display rather than looking directly out of a window. The camera system is central to the new proposal, because the same cameras that help the boom operator could also feed the AI analysis.
The remote vision system has had problems. The source material says it has been affected by what it describes as teething problems, the kind of early faults that often appear in new equipment. The Air Force says most of those problems will soon be resolved. The Air Force has not tied the new training effort to a specific fix, and the solicitation does not single out any one fault. For now, the relevance is simple: the program depends on the tanker's camera system working reliably, so the Air Force is building a training tool on top of a system it is still improving.
The KC-46 is also the service's main new tanker, and any training tool that works well on it could later be used on other platforms. The solicitation makes that aim explicit, asking for something that can “support transition into existing aviation training ecosystems.”
What the Air Force Wants Delivered
The solicitation sets out a clear end point. It says “the desired end state is an operationally demonstrated capability ready to implement for at least one type of receiver aircraft and that can support transition into existing aviation training ecosystems and provide a scalable foundation for future AI-enabled aviation performance assessment across the Department of Defense.”
Breaking that down, the Air Force is asking for three things. First, a working system that has been shown to operate in real conditions, not just in a laboratory. Second, support for at least one type of receiving aircraft, which would be the aircraft that takes on fuel. Third, a design that can be expanded to other aircraft and other training programs across the Department of Defense, the federal department that oversees the military services.
The Air Force describes the work as a prototype for future AI-based training systems. That wording suggests the service sees this effort as a test case. If it works, the same approach could be used to review other tasks that depend on crews watching each other and judging their own performance.
Why This Matters Beyond the Tanker Crew
For the people who fly these missions, the change would be personal. Tanker and receiver crews spend long hours in tight formation, and a single mistake can mean an unsafe contact, damage to equipment or a mission that has to be repeated. A debrief that shows exactly where a flight drifted outside safe limits could help a crew fix a habit before it causes a mishap.
The solicitation names reducing human-factor-driven mishaps as a goal. Human factors is a term used across aviation for errors that come from how people perceive, decide and act, rather than from equipment failure alone. If the system can show patterns in those errors across many flights, it could help the Air Force decide where to focus training.
For taxpayers, the program is also a bet on how the military buys new technology. The solicitation is issued under the Small Business Innovation Research program, which is designed to fund research and development by smaller companies. A successful prototype could lead to contracts with businesses that are not among the largest defense firms.
Open Questions for the Program
Several questions remain open, and the answers will shape whether the idea works in the field. The first is whether the camera feed from the KC-46 is clear and consistent enough for software to read it accurately, especially while the remote vision system is still being fixed. The second is how the Air Force will handle the data. Recording every refueling in detail raises questions about how that information is stored, who can see it and how it is used in any personnel decisions.
The third question is whether crews will trust the feedback. An assessment system only helps if pilots and boom operators accept its findings and use them. The solicitation's own language points to that challenge. It describes the goal as improving proficiency, and proficiency depends on the people who fly the aircraft.
The next test is the prototype itself. The solicitation asks for an operational demonstration, which means the system must show it can track the tanker, boom and receiver and produce useful feedback under real flight conditions. Whether that happens, and how quickly, will show whether AI can become a regular part of how the Air Force trains its refueling crews.
Questions and answers
What does the US Air Force want AI to do for aerial refueling?
The Air Force wants an AI system that uses cameras mounted on tankers to automatically analyze refueling operations and give tanker crews hot wash feedback, a debrief held soon after a mission.
Why is the Air Force focusing on the KC-46 tanker?
The solicitation focuses on the KC-46, which uses a remote vision system that lets operators control the refueling boom through cameras and a 3D display. The Air Force says most of that system's problems will soon be resolved.
What happens next with the AI refueling project?
The solicitation asks for an operationally demonstrated system ready for at least one type of receiver aircraft, which would serve as a foundation for AI-enabled aviation training across the Department of Defense.
Topics: US Air Force, Aerial refueling, KC-46 tanker, Artificial intelligence, Military training, Small Business Innovation Research, Boeing
Sources: Military Times
https://axynewsglobal.com/article/us-air-force-seeks-ai-to-coach-tanker-crews-on-aerial-refueling