Slashdot Mirror
The Forgotten Huygens Experiment
jdray writes "An experiment onboard the Huygens probe didn't run as planned because someone forgot to turn it on. The team lead for the experiment has put eighteen years of his life into the project, just to watch it not happen after a seven year ride to its destination on Titan."
Officials for the European Space Agency said last week they would investigate to learn what happened. They were not available for comment on Thursday, nor did NASA officials immediately respond to telephone messages...
Yeah, both groups know how to shut up and watch each others backs when it all goes f*@$ ways : )
By the way, I thought they ended up using channel B because they did not take the full effect of the Doppler shift for channel A into account. Can anyine correct me on this? ..
FTA: The probe was to transmit data on two channels, A and B, Atkinson said. His Doppler wind experiment was to use Channel A, a very stable frequency.
But the order to activate the receiver, or oscillator, for Channel A was never sent, so the entire mission operated through Channel B, which is less stable, Atkinson said.
I guess it makes it even worse that it was purely a human error.
D'oh comes from a Laurel and Hardey character. but Homer's version is a lot quicker because animation time is valuable.
David Southwood (the ESA head of science) was the one who said so - he said "That's scientists trying to screw the system. We don't have redundant systems to get more data down, we have redundant systems for redundancy." http://www.newscientist.com/channel/space/mg185248 33.700
The article isn't quite correct. A fuller description would take a while to type, so I summarise:
Two redundant radio channels were used to get data from the lander to the orbiter, which relays the data to earth. The signal for the orbiter to start listening on the high-sensitivity channel, channel A, was never given. The data was transmitted redundantly on both channels, except for images and the output of the Doppler wind speed experiment. Fortunately, all was not lost, as scientists donated radio telescope time around the earth to search directly for the A signal, despite it not being relayed via the orbiter. Thanks to this increase in sensitivity, the data acquired was good enough to fulfill all objectives of all experiments.
So everyone can relax and get one with the analysis...
The first scientific assessments of Huygens' data were presented during a press conference at ESA head office in Paris on 21 January.
Results include:
It's done that way specifically so that post processing gets the colour balance right. Other info is used to balance the colour. Hence the raw images are greyscale.
Did he inhale?
The problem with channel A is not related to the Doppler problem but due to a human error as reported in the article.
See also what I reported on saturday.
Roughly five to ten O's have been omitted. Homer's trademark expression was voice actor Dan Castellaneta's interpretation "[ANNOYED GRUNT]" (which is how "D'oh!" has always been written in scripts for Simpsons episodes). It's based on the "Dooooooooooh!" from the Laurel and Hardy routines, only shortened considerably. Source
qntm.org
They use monochrome images for a very good reason.
To capture a colour image using CCD arrays, there are the cheap, the expensive and economic ways. The cheap way (consumer cameras) is to place itty-bitty colour filters over the entire CCD array. In this way each cell captures either red,green,blue or white.
The expensive way is to have separate CCD chips for every wavelength of light you want to capture.
However, when an image is captured by a CCD array, there is a very small amount of bleed from one CCD pixel into it's neighbours. You can compensate for this by making use of image process techniques like convolution/sharpening. But these methods are completely useless with the cheap way of capturing colour images (each of RGBW will have blended with its neighbours of a different colour).
This can be done with the expensive way (professional digital cameras), but you are restricted to three wavelengths of light.
Alternatively, you can have one CCD chip, and a series of calibrated colour filters that can be swapped over. In that way you, have a low energy budget of one CCD chip, and the flexbility of analysing a scene in multiple light wavelengths, each of which can be processed separately.
Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
Charles Perrow has an excellent analysis of those type of accidents in Nuclear Plants, Petrochemical industries, Aircraft & Airways, Dams etc.
(Normal Accidents: Living with High-Risk Technologies, by Charles Perrow, Basic Books, NY, 1984.) http://oak.cats.ohiou.edu/~piccard/entropy/perrow. html
Most of these accidents and failures were not the result of lack of money or due to operator error. In this case, I doubt it was a simple as forgetting to push a button on a control panel. This is not an excuse, but a reasonable explanation for a whole range of accidents involving complex systems.
Oh, so if you don't mind, why exactly do airplanes crash and kill hundreds of people?
I'm going from memory here, but I think it's something like 80% pilot/crew error, 15% weather (which could still be considered pilot error), and 5% mechanical failure. So if those pilots/crew had been following the proper procedure and protocol, then the shit *wouldn't* happen (except in those rare cases where mechanical error is to blame, and even then, most of those cases can be traced back to a mechanic cutting corners and not following said checklists/protocol). The original poster is right. Checklists prevent the shit from happening. When the shit does happen, it's because someone wasn't following the checklists.
Like woodworking? Build your own picture frames.
(I was reminded of this by a story on NPR this morning.)
The french spoken in France and elsewhere is quite modern...in small town Quebec they speak a 200-year-old variant almost...it's quite unpalatable to the french ear really.
Well actually, I used to do software testing. You simply simulate the special conditions when things are supposed to happen. That's generally why satellites cost so damn much. Yes, there are some exotic materials that go into them, but the chief expense is testing.
"Learning is not compulsory... neither is survival."
--Dr.W.Edwards Deming
This is totally false, as an scientist will tell you. Quantity is a characteristic of quality. In this case, splitting the data stream is actually the best choice, because you get both redudency of the communications infrastructure, but you also get redundency of data. The thing to remember is that there was a limited communications window and increasing bandwidth meant that the quality of the data they were going to get back would be greatly increased. They didn't affect redudency because, while they didn't get exactly the same data, they got a working subset of it.
disclaimer: I'm a French Canadian, AKA Quebecer
Quebec French is simply the result of a different evolution of a common language. When Nouvelle-France was colonized, a language was also brought. But it hasn't evolved quite the same way as France's French. The words are pretty much the same, safe for the fact France tends to add English words faster than we do (ex. parking, shopping, week-end, etc.)
However, France's French evolved in terms of its accents, rhythms much more than Quebec's French. We speak a French that sounds more like French used to sound a few hundred years ago. But even to my ear (i'm 35) French here is changing too, it's less "coarse" than it used to be in the 60's and 70's.
Comparing Quebec French and France's French is similar to comparing British English and US English. They are based on the same language, are written very similarly, yet they sound very different because of how they are spoken.
Salutations de Montréal!
According to this BBC report they've recovered most of the lost data by receiving it directly from the Huygens lander using radio telescopes.
Perhaps it could be illustrated as follows:
Assume a probe has the bandwidth to send a total of 8 images on two channels, 4 on each channel, on the way down. Each image will be represented by the digits 1 through 8.
With exact redundancy:
A: 1357
B: 1357
With alternating redundancy:
A: 1357
B: 2468
With exact redundency, if you receive both channels, you only get 4 total images in the end. With alternating redundancy, if both channels work, you get up to 8 images in the end. But if one fails, you still get 4, just like under exact redundancy. Thus, it seems like the better choice because you get twice as many images if both channels happen to work, but both techniques still send only 4 images if one channel fails.
Table-ized A.I.
What would happen? Probably about the same as if you dropped a canister of methane into Earth's atmosphere and detonated it. A big fireball, yes, but a canister of methane isn't going to consume all the oxygen on Earth. Nor will that canister of oxygen consume all the methane on Titan.
On Titan, warning signs reading "flammable" are posted in the oxygen mines (they extract small amounts of oxygen from solid-rock water), not around the natural methane lakes.