Lockheed Martin Skunk Works, Air Force Test Pilot School run sensor-to-intercept AI tests on X-62 VISTA

Lockheed Martin’s Skunk Works and the U.S. Air Force Test Pilot School have demonstrated what the company calls sensor-driven autonomy executing fighter-relevant intercept maneuvers using real sensor data in flight tests.

The effort, conducted on the X-62 Variable In-flight Simulation Test Aircraft (VISTA) at Edwards Air Force Base, paired an artificial intelligence (AI) agent with targeting information sourced from an operational sensor. Lockheed Martin said the AI agent then executed successful air intercepts against a live target during the flight test series.

In total, the X-62 VISTA completed eight flights that produced 27 AI-controlled intercepts. The company framed the run as a validation of a full test cycle, from development and simulation through training and then execution in the air, with the AI making decisions based on what it was sensing in real time.

Lockheed Martin credited its long-running X-62 partnership with the Air Force Test Pilot School and industry participants for expanding autonomy tests into a live mission system environment. Skunk Works, the company said, has been a key partner and integrator on the X-62 for decades, supplying open software and hardware architectures intended to enable pathfinding flight tests.

In comments accompanying the announcement, Ron Fehlen, vice president and general manager of Lockheed Martin Skunk Works, said the experiment demonstrated that its AI can close the “sensor-to-action loop” aboard an operational combat aircraft. He also said the AI agents consumed classified infrared search and track feeds and executed combat-critical maneuvers in real time during the intercepts.

Stacy Kubicek, vice president and general manager of Lockheed Martin Sensors and Global Sustainment, emphasized that reliable sensor data is necessary, but the advantage arrives when sensing connects seamlessly with autonomous AI to take action in complex environments. The company described the work as a way to combine sensing and autonomy as a “force multiplier” for faster, more informed action.

Beyond the flight test results, Lockheed Martin pointed to what it says is the next phase of the X-62 effort. It said the X-62’s Mission Systems Upgrade will enable the aircraft to demonstrate seamless integration of combat systems, sensors, and airborne AI agents within a next-generation mesh network, intended to support more distributed connectivity between systems and assets.

The Skunk Works work lands amid a broader defense push to move beyond pre-scripted automation toward autonomy that can react to changing conditions. Sensor-driven AI, when it is able to sense and then decide and act within a flight control and mission context, is seen by many programs as a potential enabler of faster decision cycles in contested environments, particularly where communication latency or bandwidth constraints complicate human-in-the-loop control.

Lockheed Martin did not disclose details such as the exact sensor model used, the definition of “success” for each intercept beyond stating that intercepts were successful, the timing of decision-making steps, or the specific training and verification methods for the AI agent. It also did not provide information about how edge cases, degraded sensing, or adversarial scenarios were handled during the tests, and the underlying infrared feeds were described as classified.