By Andrew Nusca
Posting in Aerospace
Japanese researchers are working to develop rockets that have artificial intelligence, making it easier than ever to send communications satellites into space.
Adam Hadhazy at TechNewsDaily writes that Japan seeks cost savings as the impetus for smarter rockets, allowing for greater automation in pre- and post-launch diagnostic tests.
If proven successful, a more intelligent rocket could even control its own trajectory.
Currently, rockets are automatic, but not intelligent. They have some degree of automation and are equipped with sensors that trip when malfunctions occur -- but the sensors can neither inform the operator what the problem is no offer a solution.
But the sensors in the Epsilon launch vehicle will interact, operating more like a rudimentary "brain" and less like a series of switches. It will be able to determine the cause of a problem and potentially fix it on-the-fly.
One example of this AI in action could be the regulation of the electrical current that controls the orientation of the thruster nozzle. Where the thruster is pointed determines the rocket's direction, and a surge or other irregularity in the nozzle's electrical current can send the rocket off course. Applying AI in this way is quite similar to its use in electrocardiograms that interpret the human heart's electrical signals in order to evaluate organ function.
JAXA, Japan's aerospace organization, is working to get the three-stage, solid-fuel rocket off the ground by 2013. The price: 3.8 billion yen, or about $46.4 million.
With fewer components, a lighter weight and a dose of intelligence, it may be simpler than ever to launch a communications satellite into space.
Artificially Intelligent Rockets Could Slash Launch Costs [TechNewsDaily]
Mar 21, 2011
Is it not so much the hardware, as it is the control strategy? If every little thing did not have to be first predicted, and then optimized and controlled, from the ground, it seems the hardware could be much more flexible and reliable. At the very least, it could TELL the engineers what was wrong with it, and supply its own estimates of short-term reliability. While this might not make launches themselves easier ("Go on liftoff; the AI says we have at least a sixty percent chance of not blowing up!"), it could conceivably make in-mission glitches a lot easier to deal with. But note, that such a system WOULD NOT have spotted ANY of the failure modes of Shuttles to date. Until we have the imagination to predict something (LIKE cold-weather O-ring failure, coupled with a "launch fever" attitude in the human intellectual capital base), we cannot teach it to an AI any more than we could learn it ourselves.