Six Wheel Tank Drive, Ten Pounds, Internet Controlled?

[darknrgy]

I'm really excited because I just finished building the chassis for my new 6WD robot.

<right click open in new tab to see full size, sorry still getting used to the forum>





Yes, there will be structural members installed to eliminate the flex in the frame. Waiting on dimensions for batteries, motors, etc.

Each side of the robot consists of three wheels directly connected via roller chain. I bought the chain, sprockets, shafting from mcmaster-carr, the bearings from some knock-off site. Acrylic I bought at $1/sheet at TAP plastics locally. Screws, shims, washers, etc from Lowe's. The tires and wheel hubs are from robot marketplace. I have a cheap Indian-made 5A motor controller, but I'm willing to bet I'll blow that thing to pieces the first time I try it.

I built these axle assemblies and screwed them into the main platform.



I have a running blog which I will continue to update here: http://darknrgy.typepad.com/darknrgys-blog/6wd-robot/. I have a ton of information about the robot, the design, my trials and tribulations.

This is my second major robot project. My overall goal is to build a solid telepresence robot that will withstand any kind of environment. One day I might even use 3/4G and roam around the city . Imagine that.

My first robot was a real piece of crap. It worked, but it was so underpowered, everything broke constantly. It was an amusing project though, and this video summarizes it. Please excuse my dorkiness.  

[youtube]bCbhMKv2wJY[/youtube]

I have a ton of ideas to improve the remote control of the new robot. I plan to use motor encoders to control motor speed for a more reliable tracking. I have some ideas on interface to get around the lag you get across the internet. I also plan to incorporate distance sensors and add some autonomy; automatic collision avoidance, etc. At this point though, I just have a platform, and I'm ready to do whatever.

I should probably use arduino, but I am just addicted to the raw connection to the atmel chips and building my own prototype boards. Plus, considering how much crap I want to put on this robot, I'm going to need multiple processors - motor control, sensors, management, etc. Yes, I'm cheating a bit by putting a laptop on the robot, but it's for a good cause. Streaming video, gps, wireless network, 3G, anything is possible. I'm planning on getting a netbook anyway so it works out.

SO, I do actually have a question. I'm trying to find the right gear-motor, which I'll be connecting through some more roller chain, preferably at 1:1. I'd like to have about 2ft/s and I'd like to be able to propel up to about 10 pounds ... straight up. Each motor will then need to lift 5 pounds. I calculate this to be 22 Newtons. My wheels are 4.25" so I calculate about 1.2 Newton-Meters of torque. Converted to oz-inches this is 170 oz-in. At 2 feet/second and 4.25" diameter wheels, I've calculated I need about 108 RPM (at 1:1).

First off, I would greatly appreciate if someone double checked my work. I already had my dad check it, but I'd like to get some more eyes on the problem. With these calculations, I've determined that this is the perfect motor: http://www.pololu.com/catalog/product/1445. It's overpowered in every respect (200oz-in, 150RPM), it fits within my 12-24V requirements, easy to mount, 5A stall to fit within my motor controller specs, and has an encoder for measuring speed. Sweet!

Are there any rules of thumbs on power to weight? I picked 10 pounds of force for a ten pound robot because I want it to at least be able to go up a 45 degree incline with plenty of power to spare. There are some drive train losses because of the chains, but it seems to be pretty negligible.

Otherwise, any feedback on the design? I'm a programmer by trade and I do this for fun, so I half expect to be digging myself into a hole.

Anyone have any experience using ATMega8 for motor encoder data? Looks like even if I look at only one phase I'll be dealing with around 40kHz of interrupts. I found some ICs designed for decoding, but are there favorites?

[nickc]

The torque listed in the spec sheet is the the stall torque not the running torque.  Maybe try the robot motor calculator to check the  motor.  You also didn't allow for any losses in your calculation. 
http://www.societyofrobots.com/RMF_calculator.shtml
http://www.pololu.com/search/compare/51
In your chain driven design, you might need a tensioner pulley to reduce backlash or make the mounting holes slots you can slide the mounts and tension the chain.

[photomark]

Hello darknrgy , now that is a nice looking bot you have made there .

I went over your calculations and you are correct but as stated by nickc you have not made any allowance for friction losses in your drive chain and motor torque rating is the holding torque of that motor .

by the time you take all this into account the true torque requirement could be as much as 3 X your calculated figure.

internet control sounds interesting and I like the idea of an on board computer as I plan on doing a very similar thing with a RC plane I am making (one day)

there are these great little motherboards available now , they are very small and light but are very modern and fully functional PCs , they may be preferable to using a lap top  and they would save you an awfull lot of weight . when I find the website for them I will post it

have fun with your bot     

[darknrgy]

Thanks for the replies. There will certainly be losses in the drive train and I don't know how to calculate those. However, I think I grossly overestimated how much force I would need. Ten pounds of force for a ten pound robot. Enough to go straight up into the air or accelerate at 9.8m/s/s. Half that power would be enough to go up a 45 degree incline (right?) which I'm sure would be pushing the limits of the rest of the robot anyway - traction, risk, etc.

One thing I don't know is rule of thumbs - the 3X torque for efficiency loss sounds a good rule of thumb to have handy. I'm hoping with my overestimation of force I should be ok. So, I did end up already getting the motors - the original post is from a couple weeks ago. I ended up with 2 X 200 oz-in @150 RPM motors. I think I calculated 12 pounds of force at stall - I had the stall != your actual torque rating conversation with some other folks as well, thanks for the clarification. In any case, I hope I made the right decision. I won't know for... month? Actual work is competing for my time and I've been spending most of my time rewriting all my serial comm stuff and now learning python for possible use as a user interface on the user side.

I'll definitely be looking for a good on-board micro-computer. I may actually end up using a netbook. They have hardware video codecs for their webcams (and maybe even external webcam?). Although one really cool benefit of a DC powered micro-computer is that it could share the main battery. Still need to research this quite a bit.

[darknrgy]

The torque listed in the spec sheet is the the stall torque not the running torque.  Maybe try the robot motor calculator to check the  motor.  You also didn't allow for any losses in your calculation. 
http://www.societyofrobots.com/RMF_calculator.shtml
http://www.pololu.com/search/compare/51
In your chain driven design, you might need a tensioner pulley to reduce backlash or make the mounting holes slots you can slide the mounts and tension the chain.

I really wish I had built in a tensioner, possibly even using a single chain for all the wheels. My contingency plan is to just modify the holes that hold the axle assemblies on to adjust chain tension. Also prepared to come up with some motor logic that will prevent sharp changes in forces to reduce backlash. No idea how well it's going to work .

[nickc]

if the chain causes too much hassle consider switching to timing belts and pulleys.  For a living I build large computer controlled milling machines and we use timing belts and have very little backlash.  Even Harley motorcycles don't use chain anymore.
You definitely want to control the jerk of the motors right from the beginning  (jerk is the term for change in acceleration with time or da/dt).
here's a  link to motherboard/computer
http://www.newegg.com/Store/SubCategory.aspx?SubCategory=446&name=Motherboard-CPU-VGA-Combo
Here's a robot using the above computer:
http://www.pirobot.org/