Artificial Intelligence could steer hypersonic vehicles

Sandia National Laboratories (SNL) has announced that it has set up an academic research coalition to help create artificially intelligent aerospace systems to control hypersonic missiles and other complicated vehicles in challenging environments. Called Autonomy New Mexico (NM), the organization consists of numerous US universities and aims at making hypersonic craft capable of autonomously controlling their own flight.

Flying at hypersonic speeds of Mach 5 (3,704 mph, or 5,961 km/h) is one of the hottest technological fields of the 21st century, with every major power and several middling ones pouring billions of dollars into research and development. Not only does mastering the ability to fly so fast open up the possibility of traveling from London to Sydney in a couple of hours, but it also holds out the promise of cheaper spaceflight and weapons that can penetrate any current defence system. There is, however, a fly in the hypersonic ointment.

Operating in the Mach 5+ environment is extremely difficult under the best of conditions, and controlling a hypersonic craft requires an enormous amount of planning and programming. According to Sandia, this makes testing hypersonic vehicles a slow, painstaking undertaking and makes turning the technology into something practical, problematic.

Currently, hypersonic flights by Sandia involve firing a hypersonic boost-glide vehicle into space using a sounding rocket, from which it detaches itself and plunges back to Earth, building up speed like a returning spacecraft. Unfortunately, like a returning spacecraft, its trajectory is largely ballistic, with little in the way of control. Additionally, the process takes weeks of programming and calculations that can take many hours to achieve.

What Sandia wants to do is use the pattern recognition capabilities of artificial intelligence to combine localization and data gathered about what is in front of the hypersonic vehicle to make predictions and select flight paths. This could be done in minutes while under the eye of a human pilot who reviews and approves the results. In a semi-autonomous vehicle, this could be achieved in milliseconds. In the latter scenario, the pilot could still overrule the computer and disengage the AI.

Because such autonomous technology could have wide applications in areas like self-driving cars, Autonomous NM will look at a wide range of support technologies, such as advanced computing, artificial intelligence and machine-learning algorithms, sensors, navigation systems, and robotics. To this Sandia will add its own knowledge of hypersonic flight.