Cheapest Possible Robot?
prof_braino
Posts: 4,313
What's the cheapest possible robot for somebody just getting started?
For example, there's a kid that might be interested, but the parents don't want to shell out $650 for a VEEX or a Mindstorms and have it just sit on a shelf after two days.
Please take a look and see it this looks like a possibility.
http://code.google.com/p/propforth/wiki/CheapestPossibleRobot
The idea is the "key components" are ordered directly from China. You supply your own favorite micro controller and software.
You can collect the remaining parts from around the house, or make them etc.
This would not be a fast or sexy robot, but it would be sufficient to test if an individual would enjoy the robotics activity or not. The cost is low enough that an entire classroom of kids could be equipped at $20 to $50 each, depending on choice of micro controller.
Thoughts?
For example, there's a kid that might be interested, but the parents don't want to shell out $650 for a VEEX or a Mindstorms and have it just sit on a shelf after two days.
Please take a look and see it this looks like a possibility.
http://code.google.com/p/propforth/wiki/CheapestPossibleRobot
The idea is the "key components" are ordered directly from China. You supply your own favorite micro controller and software.
You can collect the remaining parts from around the house, or make them etc.
This would not be a fast or sexy robot, but it would be sufficient to test if an individual would enjoy the robotics activity or not. The cost is low enough that an entire classroom of kids could be equipped at $20 to $50 each, depending on choice of micro controller.
Thoughts?
Comments
With a cap of $20 to $50, it would be a DIY project. Purchase the microcontroller and a couple servos, and homebrew the rest. One could scavange a second hand remote controlled car for the wireless transmitter, receiver, motors, wheels, wire, battery holder, plastic parts, gears, and driver board. Look for used parts from radios, TVs, VCRs, and othe electronic devices, and collect an assortment of containers, tin cans, wire, etc. Of course for the classroom, bulk purchases of components could be a factor.
That is what I am doing with 10 year Taiwanese student as he really isn't ready to study PWM and I personally believe that introducing him to specific component construction is an important first step. It is all too easy to hand a kid a black box toy and tell him he is really learning computers and electronics while big portions of the subject are ignored. Lego Mindstorms is a good example of WOW factor over and above any real learning experience.
The Japanese have a toy manufacturer that provide a 2 motor rear axle and other components to build a tiny robot from scratch. That is what we are using as a robot platform with the Parallax Homework board attached and translating IR remote control into directions. Soon we will add a Ping for object avoidance and whatever else comes to mind. But it is about the student asking for another feature rather than me telling him to stick to a lesson plan and that he must read and understand something to get to the next level.
It is important to not daunt the student and to empower exploration. But it is also important to avoid buying an expensive toy set that wows everyone for an hour and then wants you to buy more.
Just being cheap is rather absurd. Spend money with a purpose and don't make it the whole solution. Parallax with PBasic offers very good value in that context.
Interesting you chose that example. I taught my daughter to make lemonade from scratch when she was four. We had people at the park BIDDING for the last couple glasses at her lemonade stand, it was so good. She had fun and now she believes she can do anything.
Like Humanoido sez, all the non-purchase parts can be scavenged. THAT has always been the fun part for me. And figuring out how to stick them back together. I think that is the "skill" needed to spark the interest.
And no, this will NOT be beyond college juniors, or even high school juniors. (We'll MAKE the project appropriate, by design). I already got a response from a grade school teacher, we might be able to test "how low can you go" at the middle school or grade school. Any kids above that level should be successful as well.
Any input on the "just buy it" price point? (From the link). At some price, you have to really think about it before you buy it. At another price, you just go ahead and buy it when it interested you. I think $20 for a bunch or parts, or $50 for an assembled unit or prearranged kit. What's your "just buy it" level?
-Edit- fixed humorous typo by request
The fastest way to lose a kid who's shown interest in robotics is to have him or her struggle to get the thing working.
I think you're better off raising the bar, and spec out the least expensive quality parts that have a decent support network outside your Google page. I've never been fond of stepper motor bots, but you can certainly get some higher class gearmotors with wheels, for about $15 a set. Unlike some of the similar motors I've gotten straight from China, where the gears were literally falling out when I received the package, these are pretty well constructed. Build the base out of foam board or wood or other inexpensive material. Add an MCU already on a board -- don't have them breadboard everything as they'll make mistakes and/or it will look too complicated.
A price of $50 is more reasonable for parts that are manufactured with a little more care, and supported beyond the eBay sale. That's still within most people's comfort zone. You'd be surprised how much the extra $30 or $40 or whatever will go to actually *keeping* someone interested in robotics.
-- Gordon
Good points, but the idea is the parts cost $20, so somebody can sell a kit that costs $50 and not loose their shirt.
Personally, I would prefer the brushless DC motors, 300 oz torque at 24 volts, that the guy showed me yesterday, but they cost $25 each FROM CHINA. That's already outside the "just buy it" zone that we are looking for, targeting "first time just checking it out" individuals. If a person is prepared to spend more than the total target cost on just one part, person probably beyond the "first time just checking it out" stage.
I'm not assuming this challenge is easy. Making it simple and cheap is the hardest part, that's why its cool.
Lets do the impossible, yet again! Give it a try, it won't hurt. The worst that could happen is we find some more cheap stuff for erco to buy.
Do you want to teach robotics, or soldering and electronics? There is a huge difference, but yet the two seemed to get intermixed constantly. Soldering and electronics may lead to robotics, but it doesn't start there. Wiring up a remote control receiver, an R/C stepper servo, and an H-Bridge is electronics. Determining how to navigate (navigate, not move) around an obstacle, or the IK of an arm is robotics.
For the younger crowd I suspect you did indeed mean electronics, and you can start with some of the suggested basic wiring projects. If you are trying to introduce younger children to robotics, the LEGO systems, at $250 is your best bet. Trying to introduce wiring a CPU at the same time as introducing basic electronics is not going inspire anyone, young or old.
https://www.sparkfun.com/products/10825
http://www.robotshop.com/dagu-magician-dual-robot-controller-6.html
These are both fine, though I'm not a fan of the base because the motors they've been supplying don't have enough torque. There's no ultrasonic sensor on this, but it does have an accelerometer, plus it'll run both DC motors or servos. No Bluetooth, but as I detest Bluetooth that's not a showstopper for me!
So, yeah, it could be done, but I'd rather raise the bar, make the chassis more rugged (like the BOE-Bot), maybe switch to rechargeable LiPoly. I'm for anything that puts robots into more people's hands, but like I said, striving for the cheapest just invites junk, which in turn destroys the expectations for first-learners. If those $2.50 ultrasonic sensors turn out to be reliable, then I don't have an objection.
-- Gordon
This BEAM stuff is GREAT! I have a LOT of reading to do!
Crazy, Yes. Your assessment is accurate.
Yes. I'm shooting for the whole ball of wax. I didn't say it was going to be simple.
Yes, this is conventional wisdom. It also has not gotten us very far. It only works on a very small percentage of potential engineers. The kids that like LEGO, they should use them, that is fine. I'm looking at the rest of the population the does NOT get inspired by Mindstorms. We had a team of 50 kids for first LEGO league. Only six kids did the bulk of the programming and building. The rest were just as enthusiastic, and did other stuff. So I want to find something for the other kids.
This DAGU stuff is pretty cool. It's in the right price range. Let's see. I think it lacks some kind of "kid level engineering" in some way....
I think I'm leaning toward the kids making more of the bot. Making a 3D printer from unfamiliar, seemingly random parts was the coolest thing I did last winter. True, I don't get out much, but that is the kind of experience I'm trying to capture for the kids. Just at a smaller scale, for a lower cost. They get to do some "kid level engineering" to select and assemble the parts. AND it still has to be easy enough so their parents can help when they get stuck. I think this experience, creating something and gaining understanding from the process, is what inspires; and not so much the LEGO-bot result. (Although clearly the LEGO method can delivery that same inspiration, to a certain set of individuals).
Just noodling here. What do you say?
Oh, yeah, the steppers have arrived. I now have all the parts on the list for the basic unit.
Perhaps your approach could be a combination of pre-built ready to program units, and ready to assemble units, and let the 'students' decide where they wish to explore, with both tracks available as a sequential pair. I can see something like this feeding itself; build it first, then come back to program, or just start programming one already built. It would divide up having to (try to) teach it all at once.
I don't think its valid to separate hardware and software at the beginner level. While professionals specialize, they are usually part of larger effort which covers the other specializations. From the beginner's stand point, its just "the stuff" and they can't play with one without the other. Must have both, just a simple, manageable set, is my hypothesis.
The reason I'm using propforth is precisely because "you just turn it on and get the prompt".
As a software guy, I think I've exhausted many of my options with getting kids to "use software". The BEAM bots linked previously re-inforce the "biomimicry" idea, the form must follow function, and the function should be built in to the hardware design. So now my base assumption is, "any software is probably too much". They being said, I'm still going to have software, as I always enjoy too much (in moderation, of course
Yes I agree, some way to get building blocks of functionality that are fun and interesting is the key. Balancing how much software, how much hardware, how its put together, how big are the "learning chunks", this is where art enters the decision process.
This is getting interesting.
We need standards here!!! Someone is going to recommend an altoids tin for the chassis. I have altoids tins... but in no way are they acceptable. All my wheels are too big.
Those cheap steppers manage right around 25RPMs...at 12V. So, peeps are going to need big (Skippy Giant Sized) wheels.
If fixed... what is the function or what is the parts list?
Either way... you need to run a contest. Winner gets a free Forth compiler?
Rich:)
Mobile, with many sensor options. (hopefully)
Not standardize, PARAMETRIZE! Ideally, one can use any size peanut butter jar etc lids, and the software will compensate at boot time.
Contest prize is Your Name for the baseline for the design used in the SWARM bot project at the local middle school if we get so far
Goal is to have an entire class build bots, and have them swarm!
And a free forth compiler! Good idea!
We're thinking slow for the first attempt, 5 volts. Indoors only. Bigger motors later!
5 volts it is. Slow we can do:)
Next we need dates... no not that kind... not that kind either...
...a target date
A guy from a local school wants to outfit an entire class this fall. I told him it might not be ready until January, but he could be the bleeding edge experiment. He wants something like a full demo by end of August. I don't even have the wheels on the motors yet!
I said sure, I'll give it a try.
Better contact erco about techniques for peanut butter jar lids and ply wood....
-Tommy
How to get the jar lids centered? I saw something clever on this before, please refresh my memory?
Or you can measure the distance from the horn points to edge of wheel...
-Tommy
There are tools to find the center of cylinders (see Tommy's center finder, above). Basically a 90 degree angle of some sort with a straight edge coming out at a 45 degree angle. You place the cylinder in the 90 part and scribe a line along the straight edge - do this a few times and where the lines intersect is the circle.
It seems like there is a geometric construction technique where you scribe arcs through the circle and where they intersect is the center....or something like that. Geometry class was a long time ago and there was a cute redhead that sat in front of me.......
http://forums.parallax.com/showthread.php?131098-Cypherbot&highlight=cypherbot
http://www.backyardrobots.com/osq/osq.shtml
http://youtu.be/gAr3EOZMq6Y
The design is open source.
I got it to be pretty cheap, but then I think to use the robot much you would start to want to replace the cheaper components with more expensive ones. For example, it's cheaper to use 4 AA batteries then a LiPO or LiFE battery, but if you use the robot much you will spend more on disposable batteries then a rechargeable one. A parallax servo controller is almost the cheapest and simplest way to drive eight servos, but it really isn't very good for wiring in sensors. So you would probably want to replace that with a more expensive micro controller card. As you start to do more with the robot you start to wish you put more in the components.
The cheapest way to make a robot is to hack a toy. The old WowWee robots were designed to be hack-able and you can find old ones on EBay. The problem with a lot of suggestions about how to make a really cheap robot is that they tend to work well if you know a good deal about electronics and programming. If that is the case, almost by definition your time is valuable. And if you are going to build and program a robot you will sink a lot of time into it.
The OSQ is actually sturdy enough to run for an hour. I have made other walking robots kits and while many of them are excellent most of them can not go that long without a screw coming out or a servo failing. It's really small and light and has the bare minimum of parts. Scurrier, the larger hexapod I entered in the Robogames walking contest is also pretty sturdy, but I had to add reinforcement to make it more reliable. I think with its new shoulders it is dependable but I need to run it more on tests like walking through parks or around the block. The CAD files for that robot are also open and available to download on my site if anyone is curious.