THE BIG BET: AUTONOMY FOR HYPERSONICS.

• Author: Burns, Michael
• Position: Sandia National Laboratories

Over the next six years, the National Nuclear Security Administration's Sandia National Laboratories will invest $35 million from the congressionally authorized Laboratory Directed Research and Development Program into the development of an autonomous hypersonic flight system.

Similar to a self-driving car, which uses cameras to perceive traffic, artificial intelligence to process that information and robotics to navigate to its destination, we will build a self-guided glide vehicle. This flight system will make automatic, in-flight course corrections that can compensate for atmospheric conditions or home in on a moving target, while traveling in excess of one mile per second. The technology we develop will also cut down mission planning time, sharpening our military's ability to respond to fleeting threats.

Sandia has a history of developing pathfinder flight systems, and we believe autonomy will overcome limitations of today's state-of-the-art hypersonics, which currently fly fixed, pre-computed trajectories. A flight plan is calculated in precise detail to include a host of variables--including defenses we will encounter, heat and resistance from the atmosphere, and, of course, the location of the target.

If any one of those variables changes, our warfighting ability can be diminished. This scenario is becoming increasingly likely. Other countries have reported aggressive development of their own hypersonics programs, and the nature of threats from rogue states and non-state actors is constantly changing.

In contrast, an agile hypersonic weapon that adapts to the unexpected will help the United States maintain its edge over adversaries, allowing us to effectively respond to threats anywhere in the world without putting our war-fighters at unnecessary risk.

Our task will require new computational tools. To teach a vehicle to react to its environment in real-time at hypersonic speeds, Sandia must develop algorithms that compress 12...