Now that the rules allow for laptops to be utilized as an on-board system on the robot, will any of you utilize it? Well I know I will. Computer Imaging is going to eat up a lot of computation power. I am gonna look into CUDA and see if that will help with the image processing. :mad: Now that it is the build season no time to slack off:p I hope parallel processing is the answer to autonomous robots.
Possible Uses For A Laptop:
1.)Signaling system:
The screen can display the color of the tube that the robot needs.
2.)Computer Vision:
The laptop can bring in images captured from web cams and process the various objects on screen. To make things even easier, OpenCV can be used. Honestly, I do not have the guts to write the whole thing.
3.)Artificial Intelligence:
The laptop can crunch numbers and data to generate the best route to take to score or pick up an object. That can be either transmitted to the driver or the robot can carry out the assignment itself.
4.)Decoration:
You can just stick the laptop on the robot and have a screen saver of your team logo or some shout outs or something like that. I wouldn’t be surprised if some team did that.
Well, after floating around ideas with a mentor from MaxBotix (They’re not even five miles from our school, and yet we’ve never heard of them or contacted them for the past three years), I’d have to say no to this one. A mix of cheaper sensors, mainly the line following devices and ultrasonics, can do a far better job at mapping the whole field, and far quicker, than any camera system for a field of this size. As for artificial intelligence, the Crio is still capable of plotting routes, the best routes, mind you, on the field hundreds of times faster than necessary with the right code.
This form has some discussion of weather or not it is even legal to put a laptop on your robot.
For me as per R19 “No individual item shall have a value of over $400.00. The total cost of COMPONENTS purchased in bulk may exceed $400.00 USD as long as the cost of an individual COMPONENT does not exceed $400.00.”
It says VALUE not cost. If you jack this $200 dollar laptop with $1000 of software in my eyes this would put the value of the laptop at $1200 and would be over the limit as per R19.
It depends, really, on where you place it. Undoubtedly, hard drives will be a no-no. I think it’s quite possible with a bit of shock absorption on the laptops body that a regular laptop with a solid state drive would work.
Really the only advantage to an onboard laptop would be image processing. I’m still not convinced that we’re going to have to do so much image processing in order to be competitive in auton, so it seems like putting a laptop on your robot is a big risk for little pay off.
If you’re thinking of a notification system for the feeder, I doubt a laptop screen is going to be bright enough to be seen across the field (especially when compared to an LED array).
Wouldn’t you be stuck with the $400 price limit from <R19>? With that type of a budget… you’ll still be stuck with a little Atom powered netbook (and probably not with a SSD). I don’t think it will give you much of an advantage. I tried video processing on a little dual core HP Atom powered netbook, and it struggled to keep up with everything else. We have a really powerful cRio that is designed to do so much more than we’re throwing at it right now. With the correct tuning, we can almost do anything with it.
As per R19 “No individual item shall have a value of over $400.00. The total cost of COMPONENTS purchased in bulk may exceed $400.00 USD as long as the cost of an individual COMPONENT does not exceed $400.00.”
It says VALUE not cost. If you jack this $200 dollar laptop with $1000 of software in my eyes this would put the value of the laptop at $1200 and would be over the limit as per R19.
In response to:
“Technically one can buy a very low end, but upgradeable laptop priced just below $400 and buy the upgrades necessary to make it more like a $600 laptop. There were no rules about that. Just buy them in separate orders.”
Section 4.3.5 of 2011 Rulebook
"G. If the item is part of a modular system that can be assembled in several possible configurations or applications, then each individual module must fit within the price constraints defined in Rule <R19>. If the modules are designed to assemble into a single configuration, and the assembly is functional in only that configuration, then the total cost of the complete assembly including all modules must fit within the price constraints defined in Rule <R19>. "
Well… as I think about it… the laptop is modular. You can separate the RAM, battery, and SSD. Still… you’re going to be stuck with a fairly low end laptop. If I were to even consider this, I would attempt to make a PC from scratch with a mini-ITX board or something. Still, I don’t see the big advantage.
Exactly my thought sir, but desktop computers run off of AC power. Is the robot able to produce 110v AC? If it is, I am dropping the computer Idea and going straight into PS3(s). They are cheap, $250 for the refurbished ones. 6 SPUs available to the developer and homebrew broke into the scene. If the budget allows, we can use 2 or more as a distributed memory system. But that would take lots of weight to do. I am dreaming big here.:rolleyes:
If you are trying to create a robot that works well, and that is all you’re trying to do, offboard image processing (offboard the crio) is un-needed for most tasks.
Vision processing works phenominally well with high contrast differences, or with self-lit objects. Thus the white on black rings last year, the 2007 lights etc. The pink over green of lunacy were problematic much of the time due to lighting. The colors of the tubes and changing shape depending on viewing angle even more so.
I love the idea of a robot that utilizes high amounts of vision process to automate tasks. However, I believe you will struggle to consistently track game peices, etc over the varying lighting levels seen on FIRST fields.
Last year you could adjust the robot to see the targets from close to 30 feet away at a framerate more than high enough to aim and shoot balls. You only need ONE FRAME to calculate your angle and distance to target, then use the gyro to turn and fire the ball. Your target wasn’t moving.
That is much removed from Lunacy where you needed at least several images and a timestamp to calculate the relative motion of the trailer to your robot and calculate some rather simple leading-angles and distances.
Likewise, this year, the rack is not moving. You need only 1 image to calculate your position and distance and get the angles you need to drive. The crio is more than powerful enough to do that.
I would really only want to put a desktop onto it, but even then the benefits are far outweighed by the inconvenience. Firstly, you would need a powerful battery to power a desktop. Laptops are too integrated to mess with and take apart AND put onto a robot.
I like the CRIO. It’s a nice system, as buggy as our old one may be.
You just need to convert the DC to AC; the 12v DC would convert to 110v AC.
Oh it seems like we are not doing autonomous this year. It seems that we lack the funds. It is the first time we would go to 2 regionals for the first time this year and we already invested too much money for that. There will be less money for the actual robot this year. Lets just hope my idea gets implemented, I do not like conformity.
