Google Lunar XPrize

[Concord]

deploying 3 nodes would work, but its a hassle and it contradicts what I thought was a good motto, "The less moving parts the better"

For instance, what if one of the nodes fails? How can we repair it?

I think GPS is going to be a difficult route to take here.


I also think that its possible that the navigation is pre-planned and programmed, but that the Rover has the software and the capabilities to detect the immediate terrain and be under our remote control. The tasks we plan ahead of time are well, planned ahead of time, and therefore we should be able to code them into the rover. The variable is the terrain of the moon, ditches, boulders, etc. I think that this may be a more difficult problem as well.

[Lucifer]

We absolutely have to face the high probability that we won't land where we expect to land. The first Apollo mission to actually land on the moon didn't land where they expected to land, and it was several missions before they were able to land in a specific place repeatably (at least, demonstrate they could do that, probably they could do it before but hadn't). In that case, we can't preprogram too much. We can provide a program to handle it if we land where we expect to land, but we have to have a solid contingency plan if/when that doesn't happen.

You see, we have to go to whoever's launching and prove we can accomplish the mission, full stop. They'll be depending on us for their win, just as we'll be depending on them for our win.

According to my dad, the problems with the cold are a bit more numerous than it may at first sound. Lubricants in bearings have to resist freezing, the electronics we already know has to survive. The heat expansion of materials are going to be a big issue too. If our tolerances are too exact in bearings and stuff, then the bearings themselves might collapse under the cold. Our best bet, imo, is going to be providing a heat source somehow. If that's trapping solar heat during the day and slowly releasing it at night, or if it's some other thing.

Now, with all that said, there is the peltier effect to be considered. That's electrical heating, as far as we're concerned. We could put a significant mass in there, have it take all the heat it can get, and then pump it electrically and in a solid state manner into a heat sink system that provides heating to everything that needs it. It's fairly low in terms of energy requirements, but it's not as efficient in cooling systems as freon and crap. Still, it's useful for when we need certain things.

Now, I talked to my dad about batteries and ultracapacitors. One problem with ultracapacitors (besides price) is that they're horribly expensive. Batteries won't survive the extreme temperatures unless we can warm them. So what do we do?

The prize rules don't say anything about surviving in place, nor do they require the 5km distance be a radial distance for wheeled vehicles. We could stay within a radius of the home that lets us return to home for the night. That gives us more options in terms of what we can take, because the rover can stay small and we can increase the mass to whatever's needed to maintain the rover over night, so long as we don't go over the limit of what can be transported.

We need to start this damn simulator asap. That's what we need.

[Concord]

2 questions:

Do the rules require us to go 5 km non-stop?

(My guess, 'Yes')

If yes, Can we go 5 km in a day?

(My guess, 'No')

[Lucifer]

The answer to both is probably no. However, 5km in a day isn't as bad as it sounds, since a lunar day is approximately 15 days long. (We probably won't land near the morning terminator though)

We need a function that represents the temperature on the moon throughout an entire, er, month.

[Concord]

Allow me to clarify,

By 'Non-Stop' I mean 'Without returning to base. I'm going to assume that the answer is still no

[Jonathan]

The US military can get about a 20 m range of error, but this is do to the atmosphere, humidity and on ground interference, (buildings, mountains etc.). Supposing that we could get the 4 signals on the moon, we would have a much more accuracy. By my estimates it would be within 5 m. My friend's new algorithm reduces the error on earth to just about 5 meters, and on the moon we'd get about 2-3 m.

Some facts:
- All satellites are in roughly the same direction from the Moon.
- Even if the receiver is arbitrarily precise, the errors in the satellites are still there.
This means that precision in the plane perpendicular to the direction of the satellites is quite bad even with an arbitrarily precise receiver. How feasibly correctable are satellite errors? Note we're talking about nanoseconds.

GPS works by receiving signals from 3 satellites and finding the intersection point of all 3 and the earth's sphere, the intersection point gives the 2D location of the receiver. So for the 3D information we need, we'd need 4 spheres/signals+moon's sphere.

3 spheres is enough in practice, and that's not even using the Moon's exact shape, only a rough location that's easy to find by other means.

Celestial navagation is looking like a better option to me, and a short term video+ultrasonic composition of the terrain. Use the colors from the video and mash it with the depths of the ultrasonic to create a description of the short term terrain ahead for the rover.

Ultrasound on the Moon? Interesting. Explain.

Celestial navigation requires you to know the orientation of the camera very well. How about measuring gravity from the camera? It only can't tell you how you are rotated about the gravity axis, but you don't need to know that precisely. Still I'm concerned some factor will cause the slightest unforeseen rotation, which completely destroys precision.

How hard is it really to travel 5 km?

This total distance traveled may be a straight line displacement, or may be a journey connecting a series of waypoints approved by the Google Lunar X PRIZE Judging Panel.

[Concord]

As for Ultrasound, I was thinking maybe using it like bats do, I'm not too familiar with its properties, so maybe their is as basic trait that would prevent its use on the moon. My idea though, was to use ultrasound to find distances of, say, a ditch or boulder, and build a rough map of its form, then add the video to analyze what the form was. A software would be made to do the addition an analysis and it should be able to give a pretty good description of the landscape.


Yeah, Jonathon makes another good point . . .I hadn't considered that.

Just for curiosity's sake, I wonder if as long as their not in the actual same spot . . .anyway

[QUARG]

Here is a link that answers some of the questions in your to-do list lucifer. Its a break down of the mars rover.
http://science.howstuffworks.com/mars-rover4.htm

[Concord]

I'm not very familiar with the market for Radioactive pellets,
good link there.

[Lucifer]

I've started the plan. QUARG and Jonathan, go get google docs accounts so I can put you down as collaborators.

The plan document I've written isn't terribly detailed, but it does show how we're going to get there from here in some general terms. It also has enough information to start coding. I'm going to try to lay out each stage in detail before we start them, but there's a lot that we can't know right now and won't until we finish the previous stage.

I've still got to put down research goals and describe what knowledge we need to have at various points, specify milestones within each stage, etc. But there's enough to start coding while I work on the rest of that, and really we need to talk more about where we're going. So give me google docs accounts so I can put you in as collaborators and you can read. If you really don't want to do that, say so and I'll share the documents in a way that you can still read them, but in that case you won't be able to edit them. This doesn't matter a whole lot for the plan document, but it does matter for research documents and spreadsheets.

I'll setup a subversion repository for code in a little bit, probably tomorrow or over the weekend. If kyle and jonathan want to start coding before then, one of you take the simulator and one of you take the rover, and be ready to import to the repository soon as I've created it.

Edit: Concord, this means you too, but you indicated you were a mechanic and not a programmer? If that's so, you're probably still waiting for me to come up with the research and shit we'll need.

Edit2: For languages, the simulator is either python or C++, and I'm not being picky. I'd prefer python, with C++ as needed to make things run fast when python can't do it. If you pick one of those, no discussion is needed (nor is agreement ). If you don't want to work with either of those, then we need to have our language argument. For the rover, I think it's python or we need to talk about it more. Mind you that if we start in python and later learn of a language that would be more suited to this application, it may be difficult to port. On the other hand, without actually getting our hands dirty with the rover control software, how can we properly evaluate other languages?

[Lucifer]

ODE claims not to be accurate enough for quantitative engineering applications. Here's another library, this one's in java;

http://www.impulse-based.de/

Edit: Er, I was wrong, apparently it's C++, don't know why I thought it was java.

[Lucifer]

I've created an svn repository for code and stuff. I'm wondering if we should just use that instead of google docs. (I'm strongly wondering, I have a preference for wiki-formatted text files and using txt2tags to convert them)

I'll create user accounts for everyone so far interested. I have a web panel for administering your user account now, so I'll just pm passwords that you should change. If you want a username that is different than what you have here, please say so. Also, you should put in your real name at some point, because we actually need folks' legal names. There's no big hurry on that, just something to keep in mind.

There's no code there yet, largely because I need to rearrange the default usvn setup that it made so we can have separate modules for all the software.

[Concord]

I'm giddy
mostly because of the Patriots
Victory
Victory
Victory

But its also
nice
to see the wheels
set in motion

[Jonathan]

I wouldn't really mind even raw HTML if it's written cleanly. Just have to get svn. Google Docs isn't liked much so far but it could do.

This username is fine. Maybe lowercase it if I get to type it a lot.

If you have an RCX or NXT or whatever, you might want to try to program it to get accustomed to programming bots that accomplish things in the real world. A 'real' development environment is probably best; I use brickOS.

[Lucifer]

Ok, got the docs transferred into svn. They're in the documents directory, predictably.

They're in txt2tags format, so you should probably take a brief look at txt2tags. But you should be able to get away with just following what's there already, there's already headings and lists. **this is bold.** //This is italics.// Txt2tags files are quite readable as they are, so you don't actually have to do any fancy formatting. If you want to print, I recommend installing latex and outputting to a good print target like pdf or dvi. Then it's a two-step process, txt2tags to make it a latex file, then latex to make it pdf or dvi.

Edit: I probably didn't make this very clear. You do not need to install txt2tags unless you want to make fancy output with it. If you want to edit (which is the reason it's where it is), you only need a text editor. Notepad will do in windows, Programmer's Notepad is better, and if you go to the txt2tags site you'll find links to editors that have syntax highlighting. Similarly, I'm currently using Kate in Linux for editing. Since you can read it in a text editor just fine, it's not necessary to make html or latex or any sort of output. If you want to print for some reason, or fool with making the output itself look nice, then you need the other program(s). Otherwise, just a text editor is all you need.

And subversion to check out the files, but I already pm'd everyone with accounts (and my current setup with usvn prevents me from allowing anonymous access, so the setup might change).