PDA

View Full Version : Money Pit V for Victory



jwatte
08-12-2017, 10:08 PM
I placed 2 of 20 at the local "autonomous mini-cars" meetup/race today!
The main limiting factor is the power of the motors -- they're geared 50:1, and if I gear them up more, they end up overheating and dying.

Fully autonomous training run:

https://goo.gl/photos/VJp3G81CiGQQzW9c9

Before the race, I turned up the sensitivity a bit so it didn't stop/go as much when the mid-line was intermittent (seen around 20 seconds.)
The hardest bit was before the hairpin curve; not because it's hairpin, but because it's driving straight into the glare from the windows up top!

But, on the plus side, I can climb stairs! (This is with RC control)

https://goo.gl/photos/vtNRHPZuyhqZoRPb8

I do have a wheels-falling-off problem, though. There's a 6mm D-shaft on the motors. I use 6mm reamed to 24mm hex adapters, with dual 3mm set screws. However, the wheels work themselves loose. (I have tried flooding the D/shaft cavity with LocTite 242, it doesn't really help):

https://goo.gl/photos/7QTcLaStqcGcFnEM7

Maybe I'll put a drill through the shaft to hit with the set screws; or perhaps just use regular screws at that point...

Brooks
08-13-2017, 06:46 AM
Quite impressive! If that was 2/20 then they need to work on the scoring!
Wheels falling off: Are you ramping the speed changes? I was wondering if the jolt of a hard speed change would loosen the wheels.
And you mentioned the speed of the motors - but it seems pretty "zippy" to me...

KurtEck
08-13-2017, 07:53 AM
Nice Rover! I like the stair climbing. I keep wanting to make something like that, which could wonder around my house, where the main floor is split up where the main living room/kitchen is at one level, then you drop down a couple of steps to my office and then a couple more to bedroom...

jwatte
08-13-2017, 11:42 AM
If that was 2/20 then they need to work on the scoring!


In what way? That training run would have placed perhaps 4 or 5.
The winner was a lightweight RC car that did as good a job as me on following the track, but at a faster speed.
Third place was an even lighter-weight RC car (a "donkeycar") running OpenCV on the OpenMV camera. Also following well, but slower because of physical hardware stability problems.


Wheels falling off: Are you ramping the speed changes?

I am ramping changes. It actually happens in lateral movement, not when going straight. Transport in the car, where there are sideways forces when I turn, is sometimes enough to make a wheel slide off. It's also happened when going slant-ways down a curb.


And you mentioned the speed of the motors - but it seems pretty "zippy" to me

1.5 m/s is no slouch, but the winner had turned down his speed to 2.5 m/s. The AVC winner monsters run at 10 m/s...


Nice Rover! I like the stair climbing.

Thanks!

To be honest, it's totally not worth all the time and money and skinned knuckles just to get that capability, for me, but it's worth it because I enjoy the travels at least as much as I enjoy the destination :-)


So, I could buy a solution to my speed problems... for $2000!

I'd need three of these:
https://roboteq.com/index.php/component/virtuemart/324/fbl2360-detail?Itemid=971
https://roboteq.com/images/stories/virtuemart/product/fbl2360-cover.jpg

and six of these:
https://hobbyking.com/en_us/dt6376-14p-sensored-motor-125kv.html
https://hobbyking.com/media/catalog/product/cache/1/image/565x414/9df78eab33525d08d6e5fb8d27136e95/1/2/125489.jpg

I might be better off just building my own brushless controller at MacroFab; should be much lower cost than the roboteqs.

Also, there are tons of similar-ish solutions. And perhaps using a much cheaper RC ESC and tapping into the Hall sensors for a separate microcontroller would be a more affordable way to get "good enough."I just wish the motor had more poles, because with direct wheel drive, I need torque, not speed.