First personal project.

[azy]

anything invoking the use of large amounts of duct tape gets my vote...cant wait too see a vid when its all done

classic project

[Tommy]

Good work overall and particular with the lathe work.

If I have to do it again I'l use .5" alum. in place of the .5" cold rolled steel
Would help if I was a machinist, but on a plus note the mill I was using had a CNC control
which made it possible for even a non-machinists like me to mill the pockets for the
bearings. The shaft did work out sweet because I turned down the end to 1/4" so a
encoder can easily be installed.
 http://usdigital.com/products/encoders/incremental/rotary/kit/e8p/

.cant wait too see a vid when its all done

Think as soon as we get some snow I'l get some videos

While the original idea was and is to make snow removal remote Wifi control from
PC in house, I'l look at making the Mule do Autonomous snow plowing(with supervision)(just to give me head start on mowing next spring).

I could be wrong but it seems using the Mule WiFi networked to home PC for localisation is my only option.

I'm thinking about playing with wireless dog fences for localisation, but would need one fixed and connected
to the home PC to use as a reference, I can see the cost being over 1K(ouch, gotta check Ebay&Craiglist).
http://www.havahartwireless.com/store/wireless-dog-fence/5134g?gclid=CPyduN2ouKUCFce7Kgodoma2aQ

By continually ranging to the Collar to determine the time-of-flight, the Havahart® Wireless electric fence system is able to track the real-time location of your dog.

I can see mounting the base unit on and interfaced to the Mule's PC(analog would be sweet) , then place two/three collar unit in some fix location.


Tommy
All opinions expressed are mine alone, and subject to change as information is gained.

[Soeren]

Hi,

Think as soon as we get some snow I'l get some videos

If you can't wait, just swing by for a test run... 

Yesterdays snow (as of 3 a.m.) and they promised lots more sometime during this night.

[Tommy]

If you can't wait, just swing by for a test run...

Soeren
While you may have the snow to play with  , I just got my hands on one of these  
http://usdigital.com/products/interfaces/pc/usb/usb4/

with 4 MHz encoder input frequency I can use high count encoders for velocity feedback.
would be interesting to see how a 2usec sampling could be helpful for Dead reckoning, best I can
do now is @10msec. also sweet is the unit will continue to track encoder inputs if USB is disconnected
as long as power is supplied, seems like a good way to make a incremental encoder act as an absolute one.

Tommy

 

[Tommy]

I really enjoy following the progress

While progress is slow due to my inexperience and bad timing(reacting to seasonal tasks not perfecting root problems).
at some point I'm bound to get it, I just hope I do before it's too late.

Root problem #1: Rear Casters are not going to work.

the best answer I can come up with is replacing casters with steerable rear drive wheel.
The encoder on the steering motor would be analog feedback(0deg=0Vdc, 359deg=5Vdc). on the drive motor
a standard 2000Cpr encoder would be used(counting both leading and trailing edges would give me 8000Cpr)

Feel free to tell me I'm wrong...

Tommy


 

[Soeren]

Hi,

While you may have the snow to play with  , I just got my hands on one of these 
http://usdigital.com/products/interfaces/pc/usb/usb4/

Wanna swap? 

with 4 MHz encoder input frequency I can use high count encoders for velocity feedback.
would be interesting to see how a 2usec sampling could be helpful for Dead reckoning,

I think that would be serious overkill.
Assuming a top speed of 10km/h (= 2.78 m/s - just for illustration) and a 2µs sample period, you'd get a count for each 5.6µm(!).

A more realistic speed is probably around 1..2 km/h shoveling snow which would give you pulses each 555.6nm to 1.1µm - a lot of info that you have to deal with, far more than needed.

The 24 bit counter in the USB4 can count up to 2^24=16,777,216 pulses and would overflow after 33.5s with a 2µs sample period, so would need to be transferred to the PC and zeroed regularly - If you cannot transfer and reset it in less than 2µs (impossible over USB), you'll loose samples, unless there's some way of strobing the data to a "relief" register or something along that line - I cannot download the datasheet, so I'm not too sure about it.

I would think that a pulse each 1mm or each 1cm would be more than adequate.

best I can do now is @10msec.

Oh, that's a bit granular.

also sweet is the unit will continue to track encoder inputs if USB is disconnected
as long as power is supplied, seems like a good way to make a incremental encoder act as an absolute one.

But limited to the 2^24 samples mentioned - plenty if a reasonable sample speed is used though.

I just hope I do before it's too late.

I think snow is a yearly recurring event 

Root problem #1: Rear Casters are not going to work.

the best answer I can come up with is replacing casters with steerable rear drive wheel.

You could just use non-castered wheels - no reason to steer them.
Or perhaps shifting more of the weight towards them would help

The encoder on the steering motor would be analog feedback(0deg=0Vdc, 359deg=5Vdc). on the drive motor

If you want a better resolution (probably not needed though), you could let the 0..5V cover only the used range, as you would probably have under 180° steering.
I do that to keep grey coded disks to 8 bit while keeping a 1° resolution on steering.

a standard 2000Cpr encoder would be used(counting both leading and trailing edges would give me 8000Cpr)

Feel free to tell me I'm wrong...

You're wrong   
If your tires are 400mm (they seem to be around that size give or take), a resolution of 10mm would take 125 pulses/rev and I seriously think that this would be quite enough for a firm control.

[Tommy]

You could just use non-castered wheels - no reason to steer them.
Or perhaps shifting more of the weight towards them would help

Can't shift weight(need to lift plow, and maybe mower),

Soeren
I'd be interested in hearing how I could get by without steering a single rear drive wheel(other then caster). the Mule does zero radius turns.

Tommy

[Tommy]

You could just use non-castered wheels - no reason to steer them.

Or, use a non-caster wheel I already have.(or at least should have)

Example in picture below

I can see beefing up the spindle currently used for Q axis feedback to handle any loads needed
then removing the rear casters completely, using the payload wheels in the picture below as the
only rear suspension system. As long as the Mule and any attachments can pivot under the payload
box in picture below a 360deg turn is still possible.

Root problem #2: I still believe I can use Dead Reckoning as a primary positioning system  
This may be caused by the ability to use high count encoders to solve common Positioning problems in things
like manufacturing machines. IMO the Mule is just another machine and if using encoders(linear & rotary) lets
a milling or turning machine hold positions to .0005mm they should be able to position a Mule to within 50mm,
you would think...

Quote from: Tommy on November 25, 2010, 06:29:51 AM
I just hope I do before it's too late.

Quote from: Soeren
I think snow is a yearly recurring event  

I do feel a need to move things along because the Mule's job is going to be helping my mom, every day she's faced
with tasks that even ten years ago seemed easy and gave her joy, while she still enjoys them, they no long seem
easy physically. If all goes right the Mule could be moving potting soil by spring and mowing by September.

Tommy

 


 


 

[Tommy]

george_0001.wmv


Tommy

[Tommy]

After test review.

While I do like the way the test(snow plowing) went, some known and unknown shortcomings have been spotted.

First unknown: Drive motors speed controller shares a problem with the Netbook PC, below 32degF it fails to start up.

This problem was known in the Netbook because it's first job was a GPS for my car. the Netbook was mounted
under the front seat, the model I got had to have the ability to glue the On button closed and remove the battery
so that when power is supplied it will start up, and when power is removed it would shout Off(had to check out 6 different units before I found one that did). I came up with only three options to fix this problem, 1) replace with flash bootup drive.
or 2) bring PC's temp above 32degF before starting(this option works for both the PC and Speed controller),
3) keep netbook running(this also works for Speed controller).

current fix: small 110Vac heating pad.

open to ideas.

Tommy


 

[Tommy]

As the Mule gets heavier(added 200lb of sand as ballast) and the Mule gets farther away
from home (over 450ft), I get to thinking "what if", while I have programed it to stop if WiFi
is lost(ping the base Pc each second), and if the PC shuts down the drives do go into fault,
and the Mule has never runaway(yet this winter),as I start to venture farther it's harder to have that
warm&fuzzy feeling I love so much with all the what ifs.

So I added a remote Mule shutdown button.

I got a Cobra microTalk GMRS Walkie Talkie from Walmart that listed a 20 mile range(tested at 2 Miles
non-line of sight, Sweet!). I stopped at Staples and got one of them "Thats Easy" buttons (seemed cute)
which I wired to one of the Cobra's call buttons. From RadioShack I got a SPDT 5V micromini
relay and a 1000uf cap. the relay when energized sends the Mules drives into fault.

A long range remote warm and fuzzy feeling for under $50, is a sweet thing.

Tommy

[Tommy]

While I'm sure winter has not finished yet the warmer weather has got me thinking about mowing again.

Got one of the three reel mowers temperately mounted on the Mule for testing.

Tommy

[Tommy]

While this design was fast, easy and cheap to implement, it couldn't maintain constant ground
contact for all the reel mowers. it was also very compact while the mower deck was in the raised
position, but it seems all the mowers have to independent of each other to have good ground
contact for a friction drive system.

back to the drawing board.
Tommy

[Soeren]

Hi,

back to the drawing board.

You didn't consider a spinning blade in between the wheels (like in a regular gas driven mower)?


Just stumbled over this looking for something entirely different - perhaps it may have some ideas you can use.

[Tommy]

You didn't consider a spinning blade in between the wheels (like in a regular gas driven mower)?

Soeren
The original design called for all attachments(plow,mower etc) to share a common three point connection format.
Last fall I did use a standard rotary mowing deck(24Vdc Earthwise 20"), but the noise, footprint(size) and battery
power needed to do a 40+" cut path in that style forced me to look outside the box.

I'd be the first to admit that the fraction drive ideas has only a 20% chance of giving me the results I need, but
the energy savings is worth the effort, on top of that the noise generated by the friction driven reel mowers is
so low that the Mule could mow at night, leaving the days for charging batteries(solar cells).

Tommy   

[Soeren]

Hi,

The original design called for all attachments(plow,mower etc) to share a common three point connection format.

Well, I didn't consider that. Sounds like a good idea

power needed to do a 40+" cut path in that style forced me to look outside the box.

I have seen one made from scratch with a 40W motor IIRC. It was probably only half the width (or less) though, so yes, you'd need at least 160W, as the area being cut goes up by the square of the difference (i.e. 2 times the diameter covers 4 times the area). Further, the torque needed will go up the farther from the center it should cut.

I'd be the first to admit that the fraction drive ideas has only a 20% chance of giving me the results I need, but

Since they're hinged (I assume) where they are mounted, wouldn't a shock absorber from a car going to each side of each mower solve the issue by pressing down on the far forward part of them?

the energy savings is worth the effort, on top of that the noise generated by the friction driven reel mowers is
so low that the Mule could mow at night, leaving the days for charging batteries(solar cells).

If the solar cells are mounted onboard, it will charge the same if it's doing work or not while charging, but I reckon that the solar cells are huge, if they have to charge a car battery.

Hint: Even a fairly small wind driven generator is generally more efficient in giving Watts/$ and from what we've seen from your hand, you'd have no problem rigging up one.
Although wind is usually less at night, there should still be enough to spin it most nights.

[Tommy]

wouldn't a shock absorber from a car going to each side of each mower solve the issue by pressing down on the far forward part of them?

Soeren
I went with springs and locking collars in an effort to adjust the amount of push I apply to each mower while testing,
I really don't have a clue as to the amount needed, this setup should give me the flexibly needed to find out.

 

If the solar cells are mounted onboard

Soeren, I'm above the 44N latitude, IMO the Mule would not have the surface area needed for weekly mowing.
I pictured a mule storage shed with the roof being covered in solar cell, but with that said a permanent magnet
generator would be much better this far north.


IMO it's safe to say that spring is here, If I had to rate the Mule's efforts its first winter on a scale of 1 to 10
I'd give it a 15, the 5 bonus points for the countless hours of entertainment. not since I was a kid have I
looked forwards to snow as much(even .5" can be fun and educational).

If I had to rate my efforts this winter on the Mule I'd give me a 3, thats even with the 5 bonus points for the
cold, IMO putting the Wifi and network cameras on the Mule was a crackhead thing to do, yes it was easy, fast
and I stayed warm and toasty, but the Mule learned nothing.

Tommy 

[Tommy]

Tested the friction drive Reel mowers on the Mule this weekend, Cost of test $420us(parts picked up locally).
Test performed using wired Joystick control with me walking along with the Mule.

While I felt forced into this test(because of energy savings) and had little confident in a satisfactory outcome.
but now I believe it could be useful, and warrants greater effort(and costs), like prox switches on each mower
to monitor Rpm, even a small twig will stop the blade and cause friction wheels to slide(seems like the safest mower).


Tommy

[Tommy]

Tested a Phidgets USB compass on the Mule to see if it could keep the Mule tracking straight
over a 850ft test area, and it did a sweet job(~+-.3 deg.). but that was in an open field, it
was unable to maintain a bearing when large metal objects(cars, storage sheds, est..) are present.

Has anyone tested the possibility of using magnet fields distortions as a way to map an area?

Magnet wavefront mapping

Tommy

[corrado33]

Wow, this project is amazing.  I think you have the greatest work done/post ratio! 

Why is the compass being affected by large metal objects, normal compasses work fine around them...  I understand it's a magnetic disturbance, but you have to be really close to anything metal for a compass to change it's heading (significantly).  Right?

And magnetic wavefront mapping sounds like an awesome idea!

Good luck on your project!

[rbtying]

Magnetometers are really, really sensitive.  Normal compasses have a needle with a mass about 1000 to 10000 (or more) that of the MEMS magnetometer.  Thus, the distance at which the readings of the compass will deviate is much, much larger than it would be for a standard compass.

[Tommy]

normal compasses work fine around them...

corrado33, I was wondering the same thing, so I got a standard needle compass and walked by
a car and sure enough the needle was diverted, I have to agree with rbtying's post.

Back to the mowing deck.
I have finished testing the friction drive 38" cut path Reel mowing deck, and have scraped it,
too unreliable(too many things can stop it) also, with the mower decks six fixed wheels spring
loaded to the ground it required lots of energy to keep it tracking straight, while I'm sure with
proper alignment of the mower decks wheels the energy needed could be reduced, theres just
to many variables to be reliable.

Next test..38" electric motor driven Rotary mower deck(2 20" standard mowers for testing).


PS: With the deadline for the project looming, I sure hope my localization plan works.
Tommy


 

[Tommy]

Update:

Got a chance to test the Phidgets GPS.
does a good job in open areas.

Tommy

[Tommy]

Mule's first try at mowing in a confined area in CNC mode

george_6.wmv

Tommy

[Tommy]

tested motorized reel mowers this season.

[Tommy]

Just when you think,.

The Mule's On-Board PC controller is a Lenovo 20015, before it's current job that PC was used
as a GPS unit in my car, and in the winter I would have to run a heating pad rapped around the
PC for an half hour while the car was warming up, if not the PC would try to start but fail. I always
thought the hard drive was the problem, not able to spool up because of the low temperatures,
But last night after defragging and loading the system onto a SSD and having the same problem
booting from the SSD it seems that's not the case.

In the Mule I have a 400watt 12Vdc heater mounted in the electrical enclosure which needs to run
for half an hour before I can start the PC when temperatures drop below 32F.

would be nice not to have to use that much energy.

has to be the CPU Fan now(I Think).

Tommy




   

[Soeren]

Hi.

has to be the CPU Fan now(I Think).

I don't think so, but you can spin the fan manually in case you want to test it.

I'd think it was a capacitor related issue, as I've seen my share of PC's (even owned a couple of them) with worn down caps that needed a bit of warm-up before starting reliably.

To check this, a heat gun at a distance (to avoid damaging anything), or even a couple of warm fingers directly at the electrolytics may show (if any of them bulge at the upper end, replace them ASAP).

Next guess would be on semi-defect semiconductors (transistors/diodes) or chips not seated properly in sockets. If you have RAM modules with tinned contacts going into gold flashed sockets (or the other way around), extract (even halfway will help) and reseat.

The POST will (must) fail until all systems behave.

[Tommy]

I'd think it was a capacitor related issue, as I've seen my share of PC's (even owned a couple of them) with worn down caps that needed a bit of warm-up before starting reliably.

Only reason I don't suspect component failure is the problem has been there even when new.

Just to get an understanding I put five computers in the freezer last night and today I removed and booted each.
Lenovo 20015(netbook) with SSD, failed to boot, turned self off after 5 sec. no messages but fan making noise.
Lenovo Q150(mini desktop) with SSD, failed to boot, "CPU FAN Error" message.
Acer Veriton(mini desktop) with HDD, failed to boot, no messages but HD making bad noises.
Foxconn NT535(mini desktop) With HDD, fail to boot, turned self off after 5 sec. no messages.
Foxconn NT535 with SSD, booted.

Tommy

[waltr]

I have know of ICs that will not work properly at lower temperatures.
This is very common unless they are automotive temperature rated.

Testing PCs for operation at low temperatures may be your only lower cost solution.
Another may be buying a computer with an automotive rating. Here is one source I know of:
http://www.logicsupply.com/categories/automotive_computers
This place has a good reputation and a good customer and sale service.

[Tommy]

Testing PCs for operation at low temperatures may be your only lower cost solution.

waltr, With luck it did turn out that way, most times I don't have five small footprint PC's available to put in the freezer.

because the Foxconn NT535 with SSD did boot I'l give it a try.


Tommy