New Horizons: Pre-launch, launch and main cruise, Pluto and the Kuiper belt Options

[cIclops]

Yes it's happening after all these years, the mission to the last planet!

And maybe to celebrate the confirmation of budget, NASA approval preparations and the fueling of the RTG (radioisotope thermoelectric generator), there is an updated web site at http://pluto.jhuapl.edu/

Launch will be January 2006 with arrival at the Pluto Charon system July 2015 (mark your calender!) and then on through the Kuiper belt during 2016-2020 and beyond.

20.8-centimeter telescope for 100m resolution at closest approach
IR/UV spectrometers
2 x 8GB data recorders
data rate: 768 bps (sic) to 70m DSN
465kg including fuel
$650m

336 days to launch

[chris]

Ten years is a very long time in the software universe, there may be other better approaches such as neural networks available by 2015. To make the most intelligent use of spacecraft resources during encounter is it possible to significantly enhance the software?

In my experience of writing software, the main things that makes for robustness are good design, good developers, and testing, testing, and more testing. The increasing use of techniques like unit and functional testing in the commercial environments I work in is testament to this. And remember that robustness for a spacecraft is orders of magnitude harder than for anything on Earth. You can't push a reset button if it all goes wrong. (The fact that JPL got Spirit back after the flash filesystem cock-up is one of the things that has impressed me the most about the technology on the MERs).

I'm not an expert in neural networks, but a good friend of mine did his PhD on them, and from what he tells me they would be a very bad choice for spacecraft. They are impossible to debug, and often are not doing what you think they are.

This (http://neil.fraser.name/writing/tank/) is a famous neural network story. It may be apocryphal, but I'm told it gives a flavour of the kind of problems you can end up with.

Chris