Slashdot Mirror
"Part-Time" Scientists Aim To Build Autonomous Moon Rover
First time accepted submitter ziegenberg writes "The lunar rover 'Asimov' developed by the Part-Time Scientists, and due to land in 2014, will be the first autonomously navigated rover on the Moon. Its autonomous navigation system is a major technological leap. While the Russian Moon rovers Lunokhod 1 and 2 in the early 70s were fully controlled from Earth, today's Mars rovers like NASA's Mars Exploration Rover 'Opportunity,' which has been tirelessly exploring the Red Planet since 2004, are autonomous. However, Opportunity requires nearly three minutes to process a pair of images — a delay that causes it to move at an average speed of just 1 cm/sec or less. New developments by the technology partnership between the DLR Institute of Robotics and Mechatronics and the PTS have created, for the first time, an autonomous navigation system for a rover that has the capacity to process multiple images per second. The technology boosts a stereo camera that Asimov will use to calculate its own motion, generate a 2.5-dimensional environmental model, evaluate the site and determine a collision-free path — all in real time."
It seems like every month now that I wake up in the morning and see another amazing application for GPUs. It is incredible to see the progress that a multi-billion dollar entertainment industry can bring to other markets like space exploration that would normally run on super expensive first-generation prototype hardware.
Blog
Granted, the insolation will be better on the Moon than on Mars (closer to the Sun), but... just how much power their GPU is going to require for computing the 2.5 D env data?
Questions raise, answers kill. Raise questions to stay alive.
How come they didn't just use a laser range finder to create a 3d map of the terrain around the robot?
Rocket Surgeon.
... Simultaneous Localization and Mapping.
We learn from history that we learn nothing from history - Tom Veneziano
Why not to build on the moon a large telescope-digital-photo-camera and make the HD images of Earth surface in almost real time?
Not images of space, but images of Earth, of what is interesting and vital to us.
Shooting cloudless areas on Earth constantly and feeding data into digital maps.
I have an impression that moon pass over every piece of land on Earth. But nowadays the satellite images of Earth on digital maps are often obsolete, 4 - 5 years old, especially for small towns and villages.
There will be no cost of stabilizing an artificial satellite.
Yes, but will it observe the Three Laws of Robotics?
Smivs on the intertubes!
On a only somwhat related topic; could GPS work on the earth-facing side of the moon?
the current MER software allows faster moving than the "launch load" written back in 2003.
however, how fast do you need to move? The MER rovers have a maximum speed of about 5cm/sec, due to motor power limits. How far do you really need to go in a day? Do you need to cover multiple km in a short time, or can you just lollygag along, and get there when you get there. MER has basically the same motors as Sojourner did.
Moving fast costs resources: power, mass, volume. You need bigger motors, bigger drive electronics, faster vision processing, etc. You have to have more radiating surface to reject the heat form motors and motor drive electronics. All that costs mass and volume. What are you going to give up in terms of science instruments or other functions to move faster? Do you get more value from driving farther, or having more instruments, perhaps with more capability?
there are also some other practical problems. Does that GPU handle radiation? Upsets aren't a big deal, you can just rerun the calculations, and the upset rate on the moon will be pretty low. But does it have issues with latchup or total dose?
What about lunar dust? Run faster, and you throw more dust into the "air" (it's a vacuum of course) (ever seen the video of the previous Lunar rover?). the dust is electrostatically charged, and may stick to the rover, reducing the output of the solar panels. Do you just size them bigger to compensate, or do you spend the mass on some sort of cleaning scheme.
There's a lot more to rover design than just finding better nav algorithms.
You are so right! That's why I like to send nVidia a check for around $800 every year or so. I want to do my part to support science!
You are totally blocking my view of the wall. - Dogbert
On a only somwhat related topic; could GPS work on the earth-facing side of the moon?
Theoretically, yes; but practically, I doubt it. There would be signal strength issues for one, especially considering that the GPS transmitter antennas are pointed at the Earth. Furthermore, you'd need a special receiver, since off the shelf units have software that are trying to calculate latitude and longitude coordinates on the Earth: You'd either get the lat/long of the spot on the Earth where the moon is directly overhead, with an altitude reading of 384,400 km (the distance between the Earth and the Moon), or an error. Finally, there may not be enough resolution of location to be usable for lunar navigation.
You are in luck, however. NASA is researching a lunar GPS constellation for use in future missions. In the mean time, I'd suggest dropping a few beacons and using tried and true triangulation and dead reckoning techniques.
When our name is on the back of your car, we're behind you all the way!
You mean they stole it from somewhere?
Proverbs 21:19
Woosh!
"Think about how stupid the average person is. Now, realise that half of them are dumber than that." - George Carlin
Now just think about what could be done with the trillion dollars wasted every year on the Pentagon . . .
"Think about how stupid the average person is. Now, realise that half of them are dumber than that." - George Carlin
You could probably do very well with measurements of the sun's and/or Earth's location above the (lunar) horizon, combined with a very accurate clock. That is, after all, how latitude (and, to a lesser extent, longitude) were measured out on the open sea for centuries.
Is there any reason why they aren't using the rocker-bogie suspension? It seems stupid to send a rover to the Moon and then get it stuck. This is especially true since it will be autonomous.
Um... Spirit had a rocker-bogie system and it got stuck.
I'd rather see them use an Athlete-styled system. Instead of a passive suspension, put the wheels on the ends of powered legs. That way it can roll on its wheels over smooth terrain, and walk around on rough terrain.
When our name is on the back of your car, we're behind you all the way!
It always bugs me how people seem to often use the term Scientist for someone who is actually doing Engineering. There's no shame in engineering. I'm an engineer, and engineering is no trivial job. But as I understand it, it is the scientists that figure out how the world works, and the engineers use that knowledge to design new things. Both require serious insightfulness and creativity.
Like in Venture Brothers, Rusty Venture calls himself a "super scientist." Indeed if this were reality, he would have to do a heck of a lot of science. But in fact, the majority of what he does is engineering, building huge mechanical whatzits and stuff.
Why do scientists get all the credit?
(P.S. I think technically I'm a scientist also, because I have a Ph.D. (optional) and conduct and publish research. But I don't think I'm as good a scientst as many of my colleagues. What makes me competitive is that I have very strong engineering skills, which makes my experimental systems and experiments more robust.)
"The technology boosts a stereo camera that Asimov will use to calculate its own motion, generate a 2.5-dimensional environmental model, evaluate the site and determine a collision-free path — all in real time." So what they're saying is they are putting a large robot vacuum cleaner on the lunar surface... Neato! It's about time that space tech catches up with home tech.
As a physicist I am familiar with fractional dimensions in dimensional regularization (a process whereby infinities in Feynman diagrams are tamed by reducing the number of space-time dimension by a small fraction), however I have never heard of applying such advanced theoretical concepts to an environmental model. Bravo!
They were called chronometers, and were developed in the 18th Century for use in navigation. A couple governments had prizes like the X Prize for the most accurate chronometers. IIRC, they had to keep variances within something like a couple seconds a year to be certified as a chronometer.
Understanding the scope of the problem is the first step on the path to true panic.
As long as I can try it
You can, at least in your mind's eye. Just read science fiction and you ARE on the moon! I love the rover's name, Asimov is my favorite author. They should have named the probe that visited Vesta last year "Asimov", someone probably was too superstitious (imagine the flak if it had been marooned). Heinlein would have been a better name (The Moon Is a Harsh Mistress).
I'm surprised there weren't rovers on the moon twenty years ago.
Free Martian Whores!
OK, so here's my incredibly stupid idea.
If you had access to a weather balloon, how much rocket would you need to eject an extremely small rover (RC car sized) out of Earth orbit onto an eventual moon landing?
Imagine the tiny rover is in a bouncy shell thing that will unfurl when it lands. Imagine you only care that the rover lands in under 5 years.
Call it a semipro moon bounce. Is this at all feasible?
-l
Help cure AIDS, cancer, and more. Donate your unused computer time to worldcommunitygrid.org. Join Team Slashdot!
In the mean time, I'd suggest dropping a few beacons and using tried and true triangulation and dead reckoning techniques.
Off-the-shelf star-tracker of the type used on everything from deep-space probes to the Dragon capsule, plus an atomic-clock-on-a-chip, plus a big bumper book of star tables.
(Dropped beacons would quickly fall below the close horizon on the moon. Plus you'd still have to determine the location of the beacons, using a method small and cheap enough to throw away. So why not put that method on the rover?)
Science is all about firing a drunk pig out of a cannon just to see what happens.
Hopefully they are aware enough of Asimov's work to print the robot ID number "ASM-05" on the rover.
(No, actually, I'm just glad they didn't create yet another fucking backronym... "Autonomous Science and Investigation MOon Vehicle".)
Science is all about firing a drunk pig out of a cannon just to see what happens.
Maybe being a Part-Time Scientist is all about what the name of your rover building team is, and has nothing to do with either being part time or full time, or even being a scientist at all.
When our name is on the back of your car, we're behind you all the way!
Opportunity is a mere 3km short of Lunokhod's 37km record, and now that Spring has returned to Endeavour Crater, Opportunity is once again on the move. (The rover was parked for the winter.) Look for Lunokhod's record to be broken soon.
When our name is on the back of your car, we're behind you all the way!
No, a being a scientist is about doing science. Just having science education doesn't make you a scientist.
The Tao of math: The numbers you can count are not the real numbers.
Actually, I was somewhat ignorant about them as well. I knew the Russians had landed probes on the moon, and I believe I knew that some were rovers. What I did not know was that it had travelled anywhere near that far. Still, mcgrew has a point: The moon has been rover free for an awfully long time. Nobody has landed a probe or a rover on the Moon since 1976. No lunar landings since before the Walkman was invented, before the Apple II ignited the microcomputer craze.
When our name is on the back of your car, we're behind you all the way!