New control system will allow satellites to 'think for themselves'
By Ben Coxworth
February 16, 2011
Situated as they are, high above the surface of the Earth, satellites are pretty much left to fend for themselves – if a piece of space junk is drifting towards one, for instance, no one is going to be there to push it out of the way. To address this type of situation, engineers from the University of Southampton have developed what they say is the world’s first control system for programing satellites to think for themselves. It’s a cognitive software agent called sysbrain, and it allows satellites to read English-language technical documents, which in turn instruct the satellites on how to do things such as autonomously identifying and avoiding obstacles.
The collision-avoidance system has been tested using models in a Southampton lab. Miniature satellites were created that glided across a perfectly-level smooth glass table on integrated ball bearings, to simulate the frictionless environment of space. Overhead visual markers stood in for celestial bodies, which the satellites would use for navigation via onboard cameras. Instead of the thrusters that actual satellites use for positioning, the models were able to move themselves using a set of eight propellers.
Using sysbrain, inertia sensors and additional cameras, the models were able to navigate their way across the table while simultaneously detecting and avoiding one another.
The system is programed and updated via special English-language digital documents, which sysbrain “reads” using natural language programming (NLP). The type of English used to write them is known as sEnglish, which is short for “system English.” Devices equipped with sysbrain could read such documents directly off the internet, which would allow their control system to be updated remotely – although outer space internet is still a work in progress, sysbrain has also been suggested for use in a variety of terrestrial vehicles.
All photos courtesy University of Southampton.