So anyways I've always been curious about robotics and robots and now I have the urge to make one. I want it to be intel based, and probably using wifi as the main communication method. However, there are a vast number of problems related to this and so I was wondering if anyone has done this before or knows anyone who has?
I'm avoiding pre-built kits because they are rather lame and vastly limited in what they can do. Linux on a robot would be far superior to most of these kits. I want to be able to make a NFS on wheels or an autonomous robot and anything inbetween should I so desire.
I've got ~$200 as seed money for this project, but clearly thats not nearly enough. However, will be searching for the best method and deals. Also seed money could double or even quadruple in a few weeks so we shall see.
I haven't made a robot before, but I have looked it up. It's going to be a bitch to make, but best of luck to yeh.
Edit:
Zorm, are you going to have it go off of Voice commands or like, are you going to have it respond to you typing commands into a console and sending them to it?
What kind of material are you going to be using?
How big of a robot are you going to make?
What are some of the features that it will have?
just some questions to help get you started. :)
SRSLY DEADLY? :D (<3)
I don't know how much of this is publically available, but you should check out stuff from the international science fair. I saw some amazing things there.
Quote from: deadly7 on March 30, 2006, 11:50:00 PM
I haven't made a robot before, but I have looked it up. It's going to be a bitch to make, but best of luck to yeh.
Edit:
Zorm, are you going to have it go off of Voice commands or like, are you going to have it respond to you typing commands into a console and sending them to it?
What kind of material are you going to be using?
How big of a robot are you going to make?
What are some of the features that it will have?
just some questions to help get you started. :)
Definitely not voice commands, main idea for right now is controlling it via the wifi connection. Possibly with something like a joystick to start off and then as the project progresses it can get more advanced. My idea is that I'll be able to start out very simple and work my way up to something more advanced.
I'm not really sure on the material thats going to be used, most likely aluminum and plastic. For now I'm thinking something slightly bigger than your typical RC car, with 6 wheels(3 on each side).
Featurewise, I'm going to say having 6 wheels and possibly a camera to start with. If I can get that far I'll be well off for the future.
Some concerns/idea/questions:
Power is going to be a big deal, the cpu will have to be something with low power consumption. I suspect I'll end up using something like lead-acid batteries for power. However, what should be done about storage for files on the computer? Something like flash media to lessen power consumption?
Wifi, what will get me the best range/bandwidth? Currently I have no wifi in my house so an access point will need to be acquired. Suggestions?
Cost is the other huge concern, I'm thinking that I should be able to raid some local thrift stores for things like motors/parts for the base and what not. Any ideas of specific electronics/appliances I should look at for motors? I also need to check out the black hole and see what it has, I may be able to acquire some goodies from there.
I didn't read your second post, but I'll say what I know from my school.
You might want to look into getting a microcontroller such as an Atmel AVR development board. I have written small programs for it (it's written in C, interfaces over serial with Linux, Mac, or Windows), and it's fairly easy to program. I'd recommend similar boards for robotics, I know the main robotics prof at my school uses Atmel. The board I have was about $120. You can probably figure out which board is best by checking around robotics websites.
In terms of control, at my school we use infra-red. At school, the tiny little controllers on-board our tanks understand approximately 4 commands (speed up/slow down left/right treads) It is cheap and consumes little power. Another option is just to use a modulated wave, probably in the 2.2 - 2.4ghz range since that's the only legal frequency. Using standard WiFi is typically wasteful in terms of CPU and power, and it is usually way overkill. The idea of robotics is to implement exactly what you need, and nothing else.
In terms of movement, I'd suggest starting with something small, like wheels. At school, we used a tank with treads. If you're planning on doing autonomous agents, you're going to need to track down collision sensors or something similar, but I think that's beyond what you're trying to do. If you want something more complicated, you have to start getting into cervos (servos? I've never seen it spelled..), which are quite pricy.
Hopefully that helps a little.
iago, I'm not sure if it's just not legal in Canada, but there's 802.11.b out as well, which is 5.4 GHz. You'll have a lot less interference because it's not used often, but the range dies if you try controlling it through walls. It also doesn't allow as much data transfer. Zorm, I'd recommend reading this (http://en.wikipedia.org/wiki/802.11) entire article to help you decide what you want.
EDIT: Oops, it's not 802.11.b that's 5.4 GHz, it's something else, but I totally forgot what it was.
EDIT 2: I remember what it was, it's 802.11.a that's in the 5 GHz range.
Quote from: iago on March 31, 2006, 12:31:05 AM
I didn't read your second post, but I'll say what I know from my school.
You might want to look into getting a microcontroller such as an Atmel AVR development board. I have written small programs for it (it's written in C, interfaces over serial with Linux, Mac, or Windows), and it's fairly easy to program. I'd recommend similar boards for robotics, I know the main robotics prof at my school uses Atmel. The board I have was about $120. You can probably figure out which board is best by checking around robotics websites.
In terms of control, at my school we use infra-red. At school, the tiny little controllers on-board our tanks understand approximately 4 commands (speed up/slow down left/right treads) It is cheap and consumes little power. Another option is just to use a modulated wave, probably in the 2.2 - 2.4ghz range since that's the only legal frequency. Using standard WiFi is typically wasteful in terms of CPU and power, and it is usually way overkill. The idea of robotics is to implement exactly what you need, and nothing else.
In terms of movement, I'd suggest starting with something small, like wheels. At school, we used a tank with treads. If you're planning on doing autonomous agents, you're going to need to track down collision sensors or something similar, but I think that's beyond what you're trying to do. If you want something more complicated, you have to start getting into cervos (servos? I've never seen it spelled..), which are quite pricy.
Hopefully that helps a little.
The biggest thing for me is the ability to move into new areas if I so desire which to me means power and overkill for now. I've looked at the Amtels before but it doesn't really seem to me like they'd have enough power to handle something like video processing?
The other advantage of using a x86 based computer is usb support and all the goodies that come along with it. Should I decide I want to add a GPS receiver or such its a matter of buying it and plugging it in vs. having to solder a chip onto a board and wires, etc. I'm sort of in favor of using off the shelf parts because if I end up getting in way over my head they will atleast have ebay value.
Bump sensors and such will be easy enough to get for a start on autonomous work, the only question is how to get these inputs? USB can probably only support so much and theres the issue of I doubt something like bump sensors will come USB ready.
The idea behind using standard wifi is that it will be easy and I'll be able to do things like stream video/other information down for debugging.
For movement I'm definitely started with wheels, but of course if it all works good and is cool I won't rule out wings someday. Its servos btw, and yeah most of the parts needed are way up there in the price range which is why I'm considering improvising on them for now.
Well, are you thinking of eventually making something like the spycamera used in American Pie: Band Camp? (Basically it was an "RC" Car with a camera attached that sent a constant feed to a computer)
In a way yes, but that isn't the only goal/objective and so as such I don't really want to make use a specific hardware just to accomplish that goal.
What other goals/objectives do you have?
I know the ones my school ordered were really basic, used laptops for brains and webcams for eyes. Then everything else was I think special crafted USB hardware and such. It had things like color recognition and shortest path AI.
Quote from: zorm on March 31, 2006, 01:28:11 AM
The idea behind using standard wifi is that it will be easy and I'll be able to do things like stream video/other information down for debugging.
It's really not that basic at all. You'll most likely need the wireless router and 1 access point in your house, to hopefully cover all of the gray and dead areas. Also, you'll need to find a way to make it so that when the bot loses the connection to one access point, it gains the connection to another, with minimal downtime (hopefully none).
"Bump sensors" are pretty easy to make. All you need is a photodetector (http://store.yahoo.com/webtronics/if-e10.html is a cheap one, and pretty good for this sort of thing). Detection is pretty easy. if(something_is_there) { move(); } I'd say stick with simple stuff. A breadboard is always a good way to go early on (no soldering needed! And they are pretty cheap).
Aside from that I can't say much right now (too early -.-).
Quote from: zorm on March 31, 2006, 01:28:11 AM
The biggest thing for me is the ability to move into new areas if I so desire which to me means power and overkill for now. I've looked at the Amtels before but it doesn't really seem to me like they'd have enough power to handle something like video processing?
The other advantage of using a x86 based computer is usb support and all the goodies that come along with it. Should I decide I want to add a GPS receiver or such its a matter of buying it and plugging it in vs. having to solder a chip onto a board and wires, etc. I'm sort of in favor of using off the shelf parts because if I end up getting in way over my head they will atleast have ebay value.
Bump sensors and such will be easy enough to get for a start on autonomous work, the only question is how to get these inputs? USB can probably only support so much and theres the issue of I doubt something like bump sensors will come USB ready.
The idea behind using standard wifi is that it will be easy and I'll be able to do things like stream video/other information down for debugging.
For movement I'm definitely started with wheels, but of course if it all works good and is cool I won't rule out wings someday. Its servos btw, and yeah most of the parts needed are way up there in the price range which is why I'm considering improvising on them for now.
USB is way overkill. There's proper ways to do that, but I don't really know tham. Things like USB, WiFi, and Ethernet are rarely used in microcontroller systems. They require way too much power and are expensive.
I don't know much beyond what I said in my last post, though.
Quote from: rabbit on March 31, 2006, 09:33:26 AM
"Bump sensors" are pretty easy to make. All you need is a photodetector (http://store.yahoo.com/webtronics/if-e10.html is a cheap one, and pretty good for this sort of thing). Detection is pretty easy. if(something_is_there) { move(); } I'd say stick with simple stuff. A breadboard is always a good way to go early on (no soldering needed! And they are pretty cheap).
Aside from that I can't say much right now (too early -.-).
I hate breadboards.
I took robotics at my highschool. It was kind of fun, but also a bit lame. For one thing, being a hoby robotist is extremely expensive. My partner and I spent about $1,000 of our school's money on just the peripherals for our robot.
The best road to take would probably be to join a competitive team that is being sponsored by some big company.
Quote from: deadly7 on March 31, 2006, 10:52:27 AM
Quote from: rabbit on March 31, 2006, 09:33:26 AM
"Bump sensors" are pretty easy to make. All you need is a photodetector (http://store.yahoo.com/webtronics/if-e10.html is a cheap one, and pretty good for this sort of thing). Detection is pretty easy. if(something_is_there) { move(); } I'd say stick with simple stuff. A breadboard is always a good way to go early on (no soldering needed! And they are pretty cheap).
Aside from that I can't say much right now (too early -.-).
I hate breadboards.
I have a nice training board that isn't a breadboard, but is functionally equivalent.
I left the picture big so you can see details, so I won't hotlink it:
http://www.javaop.com/~iago/atmel2.jpg
Notice that you can easily add/remove CPUs and other bits. It's a really great board to learn on.
That's hot. Did you solder it yourself?
There's no solder anywhere on the board.
I've done the breadboard stuff before, and used handyboards for robotic development at school. My only problem with these is that they weren't very powerful and didn't interface with anything else well.
I don't really know of anything aside from USB/Firewire that will provide modular components like I'd like. I also suspect that if I wanted to go down that road it'd be even more expensive. Thats one advantage of using something more common in everyday use like USB that it will be cheaper.
Feanor: $1000 on just peripherals seems excessive.
On wifi: being able to quickly switch access points or what not isn't a huge deal. I mainly plan to use it for information relay/debugging to avoid cables. It will probably also be used for remote control in the start.
So heres my list of goals for now atleast:
1. Remote control drivable, w/video
2. Autonomous with simple sensors, and simple tasks like avoid light or just move around avoiding obstacles, etc.
3. Autonomous and traveling along GPS waypoints, the idea being that I can give it GPS waypoints down my street and it will be able to travel there based only on its knowledge.
Edit:
I came across http://www.epiacenter.com/modules.php?name=Sections&sop=viewarticle&artid=61 which talks briefly about making a PC based robot. I also found http://www.linuxrobots.org/ which seems helpful.
If you want sponsorship for this, you need to have some sort of goal, otherwise I doubt you'll get much cash.
Aside from that, good luck.
Not intending to get sponsorship or anything of the sort, this is just a project for fun.
Quote from: iago on March 31, 2006, 01:28:13 PM
There's no solder anywhere on the board.
What the hell? How do the LEDs stay in? :?
So just sort of looking around at various parts and I've come up with this list:
VIA EPIA ME6000 LVDS Fanless Motherboard $135 @ http://www.mini-itx.com/store/ (http://www.mini-itx.com/store/)
Compact Flash to IDE Adapter (Type 1/2) $34 @ http://www.mini-itx.com/store/ (http://www.mini-itx.com/store/)
Kingston 512MB Compact Flash (CF) Flash Media $24 @ http://www.newegg.com/Product/Product.asp?Item=N82E16820160013 (http://www.newegg.com/Product/Product.asp?Item=N82E16820160013)
Crucial Technology 1GB 184-Pin DDR SDRAM Unbuffered DDR 266 $106 @ http://www.newegg.com/Product/Product.asp?Item=N82E16820145085 (http://www.newegg.com/Product/Product.asp?Item=N82E16820145085)
Still requires wireless card and wireless AP, as well as motors + way to drive them + batteries.
I would suggest maybe an older laptop battery off eBay...Charging would definately be an issue though. Also you are going to need to take weight into consideration, I really don't know where you are in school but if you don't know the physics then just ask.
Solar panels!
Okay, maybe not. Litium hydride batteries are pretty good for technical stuff. Also, if you weren't going to send it down the street for a beer I would have suggested an umbilical. Hey! Maybe you can give it a plug and teach it how to charge itself when its batteries get low?
I was going to say teach it to map WiFi hot spots around your house.
Looking at a couple of http://www.radioshack.com/product/index.jsp?productId=2049668 (http://www.radioshack.com/product/index.jsp?productId=2049668)s to be the batteries. Think I should have a seperate battery dedicated solely to the computer?
The local newspaper sells aluminum plates that are 35"x 23-1/16"x.007" thick for $0.20 each. It doesn't really sound thick enough but I'll have to investigate more.
I dug up an old webcam, which appears to have drivers made for it already on Linux. The webcam does 640x480 @ 30fps so it should be more then plenty for my needs.
Now I just have to figure out which linux distro to try and use. Any suggestions from the crowd?
Also I saw somewhere online a suggestion of using windshield wiper motors, if only I knew where to find a junk yard because I could probably salvage some of those for rather cheap.
[edit]
The mapping Wifi hot spots is a neat idea actually, but I'm unsure how useful it'd be around here. Solar panels seem like a cool idea but unless my robot is going to be outside for extended periods of time on its own they aren't very useful. Making it know to return home to recharge however is a very good idea and I can probably be done.
[/edit]
I just hope you realize those batteries are realllly heavy. IIRC they are about the same as a car battery. I highly doubt windshield wiper motors could carry a couple of them around. I think that you would be fine with running the computer off a few 9v's, especially if it's not a whole tower (which from the sound of it, it won't be).
A problem with aluminum is that it doesn't like to solder. Even with an acetylene torch I have a hard time making a good bond. Unless you have a plasma torch (IE: for welding), I doubt you want to use aluminum.
I doubt it will be powerful enough, but picotux is SMALL. Another one which might be a bit more well suited is Damn Small Linux which is about 50mb, which leaves lots of room left for sensory storage or just lets you use a smaller (cheaper) drive.
Those batteries are so fucking huge like wtf? I say find a common old laptop battery and gerry rig a few into the robot as your powersource.
I went to Digikey to get better information and they actually have a better price on them too. They weigh about 5.5lbs so they are no where close to a car battery and are probably very much similar to a laptop battery. http://dkc3.digikey.com/PDF/T061/1591.pdf (http://dkc3.digikey.com/PDF/T061/1591.pdf) for detailed specs. The thing about the battery is it needs to be rechargable and cheap is preferred. I also suspect that most motors can handle this weight with ease. You have to remember the plan is to have rather large motors, i.e. bigger than what you'd find in most RC cars and such.
My neighbor who lives behind me built a battlebot type deal, which used like 1/4" thick steel plate as a base which weighed more than the batteries he used to power it. I'm not exactly sure on the types of motors he used, I guess I should find out.
Not really planning to solder the aluminum, I plan to bolt everything to it basically. Have better suggestions for a base material thats cheap and can be easily acquired?
The problem with things like Damn Small Linux is that its made to be used as a workstationish environment and so not really suitable for this project. I've also realized that my cf size and memory may be on the overboard side. I could scale them back to way less probably.
Regular steel is a good base material. If you can afford it, titanium is stronger and lighter. The problem with windshield wiper motors is that they are designed for only a limited range of motion (150-175° or so, depending on the type) and aren't that powerful. I'm not too sure about a power source, but jury-rigged laptop batteries will definitely work.
And as to DSL, just don't run X! You can install what you need, run it, and you'll be set.
Ubuntu Server may be a good choice for distro.
EDIT -
Also, DSL is a pain in the ass to install. Thought you ought to know.
Actually, I could just use Windows for the whole thing. It'd make life really easy.
I wouldn't recommend it, though. Windows will use a lot of the available power and memory on, well, a whole lot of nothing.
I'm not so sure I couldn't strip down a win2k to make it nice. The advantage of all of this is that I'd get to program on a platform I know a lot about, and I'll break the hearts of a lot of geeks who ask what it runs when I get it all working :p
A quite powerful motor, and in this case quite cabable, is automatic window motor, they do full rotation also. Go to a local junk yard you could probably pick up a few of thoes on the cheap, or free.
MIT is about 10 miles away from my house, and they often have open houses showing off thier latest technologies, this includes their rebotics department. I'd think that many colleges do this also.
Quote from: zorm on April 04, 2006, 09:17:53 PM
I'm not so sure I couldn't strip down a win2k to make it nice. The advantage of all of this is that I'd get to program on a platform I know a lot about, and I'll break the hearts of a lot of geeks who ask what it runs when I get it all working :p
Robot .NET 2.0: OOP with serial ports, and XML web services all rolled into a small rolling package. :)
This post has thoroughly inspired me, many thanks! :D
Quote from: zorm on April 04, 2006, 09:17:53 PM
I'm not so sure I couldn't strip down a win2k to make it nice. The advantage of all of this is that I'd get to program on a platform I know a lot about, and I'll break the hearts of a lot of geeks who ask what it runs when I get it all working :p
You know, Microsoft is always trying to market themselves. You might try to exploit that by trying to get them to license you a copy of Windows XP Embedded. Tell them you'll advertise that they donated to your project.
Quote from: MyndFyrex86] link=topic=5395.msg63624#msg63624 date=1144251691]
You know, Microsoft is always trying to market themselves. You might try to exploit that by trying to get them to license you a copy of Windows XP Embedded. Tell them you'll advertise that they donated to your project.
Thats a great idea, I should hit up Newby since he has Microsoft connections.
I played around with XPe today and its actually very cool. Now I just need to figure out the best way to get a free license for it and I'm all set!
Also, add a compactflash reader/writer to my list of needed items. I figure this might help when I want to load the OS onto it.
Quote from: zorm on April 05, 2006, 10:38:11 PM
I played around with XPe today and its actually very cool.
Out of curiousity, how did you get it? Torrent?
No, they actually offer a free 120 trial. Then I googled around and found out how to run it inside vmware. Haven't done much aside from that since it really needs a hardware profile, but the generic one will work for now.
Quote from: zorm on April 05, 2006, 10:52:47 PM
No, they actually offer a free 120 trial. Then I googled around and found out how to run it inside vmware. Haven't done much aside from that since it really needs a hardware profile, but the generic one will work for now.
What I highly recommend is putting together a project proposal and see if they'll give you an internship. I'll clarify.
I found out the Microsoft recruiter for my university is Kim Bremmer (you can find this out at http://www.microsoft.com/college/). I didn't get the internship I applied for a while ago; however, when I was last on my let's-get-a-Microsoft-gig hype time, I sent her an e-mail asking, essentially, "If I put a project proposal together for something that would add value to a current Microsoft product, could I get an internship working on that project?" She told me, essentially, "Put the project proposal together and then we'll talk."
That was more work than I had time for at the time, but I am fairly certain that they're not averse to looking into new and creative ways. Their internships have great package benefits, and who knows? You could be building a Microsoft battlebot running XP Embedded, while they're paying you to do something that you wanted to do (basically) anyway. Then you could get parts and stuff for free too, *and* have the advantage of hardware pros like the people at the Xbox 360 division.
Of course, then you trade-off the notion of not being singularly identified with the project. But it's still not a bad gig.
That might be way more in-depth than you had in mind, but it's probably worth considering. ;)
Quote from: MyndFyrex86] link=topic=5395.msg63755#msg63755 date=1144293621]
What I highly recommend is putting together a project proposal and see if they'll give you an internship. I'll clarify.
I found out the Microsoft recruiter for my university is Kim Bremmer (you can find this out at http://www.microsoft.com/college/). I didn't get the internship I applied for a while ago; however, when I was last on my let's-get-a-Microsoft-gig hype time, I sent her an e-mail asking, essentially, "If I put a project proposal together for something that would add value to a current Microsoft product, could I get an internship working on that project?" She told me, essentially, "Put the project proposal together and then we'll talk."
That was more work than I had time for at the time, but I am fairly certain that they're not averse to looking into new and creative ways. Their internships have great package benefits, and who knows? You could be building a Microsoft battlebot running XP Embedded, while they're paying you to do something that you wanted to do (basically) anyway. Then you could get parts and stuff for free too, *and* have the advantage of hardware pros like the people at the Xbox 360 division.
Of course, then you trade-off the notion of not being singularly identified with the project. But it's still not a bad gig.
That might be way more in-depth than you had in mind, but it's probably worth considering. ;)
Interesting idea but I wouldn't really know where to start since I've never done any sort of real project proposal before. Also I won't be in college until the fall but that shouldn't be a huge deal.
My basic idea for now is to start coding some of the basic software things and to see where I get. I also realized that I have a spare IR mouse that can be used for detecting/verifying movement so I'll add that in too.
On software:
My basic idea is that the robot will run a server which will control everything as well as relay information to the clients. Since I'm using C++ I'm not really sure of how modular a design I should start with. Perhaps having it load DLLs for all the different sensors and such?
The clients will run on remote PCs and will basically be there for debugging/information purposes. As well as remote control when running in such a mode.
So I started coding and this is as far as I've gotten for now:
(http://www.valhallalegends.com/zorm/RobotServer.jpg)
Is that from a webcam or an actual robot you're starting to assemble? I love how you aimed it right at the picture, that's pretty cool (I did that once with a 3-plug monitor and a video camera once, it just went blue).
Its from a webcam that will eventually go on my robot. I'm just working on basic software for now.
You should over complicate this robot just for fun. Put it in a sphere that has a gyroscope in it, then put 1 track that goes all the way around the sphere with a wieght on it, then another one that's perpendicular and goes around the sphere as well with another wieght on it, and use that to control the way it moves. That'd be really hawt, and probably isn't *too* compliated. Only problem with this that I can think of is if the robot is ever on a slant of some sort.
Combined with a gravity sensor, that might be interesting. :)
So I walked out of my awards ceremony for jrotc last thursday with a check for $300. I also believe I'm in line for another $200 but I don't have cash in hand yet.
So this pushes seed money up in the $500 to $700 range.