Thanks for the explanations @strantor; I'm just not familiar at all with this stuff at the hardware level.
As for the software items I mentioned, I've already written them all. Everything basically works already but I still need to figure out how to put it all into together into a box that I can mount onto the handlebar. I'm sure there will be other issues to sort out once I try it outside.

I do have an Arduino but I felt like the Pi gave me more options. Maybe just because I'm more familiar with it.
BTW, it only takes 45 seconds for the Pi3 to bootup.

All I wanted to do originally was control the speed of the electric push cart via remote control instead of having to keep my hand on the handlebar all the time. Sounds like all I needed then is an Arduino, an IR receiver and then wire the Arduino directly to the existing potentiometer?

Please pardon me, I completely missed post #25. Not sure how; it has pretty pictures that should have jumped out at me. So I thought this thread was still at the stage of you asking questions rather than posting that you had it working.

The openCV aspect is pretty cool, and something you wouldn't have been able to do with arduino. Arduino would have been perfect for what you originally described though.

I am curious though about this 5-speed business. I see no reason why the motor controller would be limited in this way. I wager that it is capable of 0-100% stepless speed control, and the thing limiting you to 5 speeds is the handlebar twist grip. I originally called it a speed pot but maybe it isn't a pot at all, but rather a 5-position selector switch with 5 different resistor values. I would be curious what happens if you disconnect it and connect an actual Potentiometer in its place. Maybe that doesn't matter to you now with the solution you've devised, but if it does, and if you try it, I'd be interested in hearing the result.

 

I'd like to build a remote control to regulate the speed of my electric golf cart and I'm looking for suggestions.
I'm a hw novice (more a sw guy) but willing to learn attempt this if it's possible.

The golf cart speed is controlled via a rheostat and knob on the handle.
When I'm walking the course I'm constantly micro adjusting the knob to get the speed right
which means I have my hand on the handle all the time.

The rheostat only has 3 wires that go to the motor. I'm not sure how it actually controls the speed.
What I want is a remote device with 3 buttons, faster, slower, off which is all that the rheostat
seems to do. It tuns clockwise, counter-clockwise, and clicks at the end to stop the motor.

Thanks.
Alan

View attachment 267739

It sounds to me like all you need is a speed regulator.

A speed regulator will monitor the speed of the motor or wheels and make micro adjustments all automatically. All it takes is to find a sensor that measures the wheel rotation speed, and for super duper control the rotational acceleration.
It's a simple feedback circuit where the speed is measured and then the control circuit makes the adjustment based on a single adjustment you make with a potentiometer. If you adjust it for 1 foot per second for example then it goes 1 foot per second regardless whether or not you hit bumps or crevices. For feedback sometimes a small generator (or motor) is used and sometimes a digital device. If you do the whole thing in analog then an op amp does the control. If you do it in pure digital then a microcontroller is used.
This is a simple system yet was used and still used in may tape recorders/players to keep the capstan rotation constant.

In fact, there are some systems so simple they use two or three transistors to regulate the speed, and that was used on older tape recorders.
The way they did it, and can still be used today, is to measure the back emf of the drive motor so no sensor is even needed. It's nicer to use an op amp for control though and is more accurate.
If you are interested i can show you some diagrams that will help you understand these simple systems. For something like this i cant see resorting to something like Lidar that's such a huge overkill.

 

Since the intent of the control is to keep the cart at some certain distance from the person, that distance should be the feedback parameter, not the speed.
Controlling distance will also reduce the tendency to run away, and besides that, the visual effect will be incredibly cool. Not sure on exactly how it will need to sense, because it should not turn into a game of "keep away" , where it moves whenever the user reaches for it. Probably adding a start delay after it has been stopped for a bit will be adequate. That part will require a micro. Sensing some short distance well enough to control it will be a challenge to do simply. But it will reduce the need to steer it, as well as reducing the need for remote control. So it is worth thinking about. Quite possibly sensing an IR beacon on a belt clip would work. Some triangulation scheme could determine the distance. with enough accuracy for the application.

 

Thats pretty cool project, I think a follower robot is actually pretty useful, get it to carry things for you is a simple and useful thing to be able to do.
Watch out for computer vision itll snow you under, it gets complicated, but I think the solution you have now looks like the best option.

Best CV I can think of, is a 3d yolo type system (3d bounding cubes with identification class id each), beyond that I wouldnt bother, I think its a dead end, thatll pretty much handle a terminator Im pretty sure. Even getting complex Id just keep it simple as possible, or youll never get anything done with your life!!!

 

Since the intent of the control is to keep the cart at some certain distance from the person, that distance should be the feedback parameter, not the speed.
Controlling distance will also reduce the tendency to run away, and besides that, the visual effect will be incredibly cool. Not sure on exactly how it will need to sense, because it should not turn into a game of "keep away" , where it moves whenever the user reaches for it. Probably adding a start delay after it has been stopped for a bit will be adequate. That part will require a micro. Sensing some short distance well enough to control it will be a challenge to do simply. But it will reduce the need to steer it, as well as reducing the need for remote control. So it is worth thinking about. Quite possibly sensing an IR beacon on a belt clip would work. Some triangulation scheme could determine the distance. with enough accuracy for the application.

Hello,

Oh i must have missed that i read this thread quickly a little pressed for time.
Yeah that's a different story. Speed regulation could help if the movement of the person is constant but otherwise not.
The goal of the project seems to be kind of cute and would be cool to see in operation.

 

All I wanted to do originally was control the speed of the electric push cart via remote control instead of having to keep my hand on the handlebar all the time.

I guess I'm just to old. I still think your going to need to have your hand on the handle bar to steer the cart. Or add a second motor to power each wheel, and steer by differential type steering, like in a zero turn mower. The whole thing still seems like one of those ideas looking for a problem to solve, to me. I thought golf was to get exercise?

 

I guess I'm just to old. I still think your going to need to have your hand on the handle bar to steer the cart. Or add a second motor to power each wheel, and steer by differential type steering, like in a zero turn mower. The whole thing still seems like one of those ideas looking for a problem to solve, to me. I thought golf was to get exercise?

Hi,

From what i gathered it sounded like he wants the cart to follow him automatically, i guess that is what the Lidar was for. I could be wrong but that would be a cool idea and maybe a big seller. Put a lot of caddies out of work though as long as it can take orders to also hand him a certain wedge

 

The whole thing still seems like one of those ideas looking for a problem to solve, to me.

You might be right but so what? That's the starting point of plenty of technology.

I guess I'm just to old.

Old is a state of mind, not an age. I feel myself getting older than I'm comfortable with so I've been trying to dial it back lately.

I still think your going to need to have your hand on the handle bar to steer the cart. Or add a second motor to power each wheel, and steer by differential type steering, like in a zero turn mower.

Agreed. What has been disclosed thus far seems like it will only work in a straight line. Maybe that's "good enough" for a Rev0? I don't play golf so I'm not sure how much of the travel across the course is straight line travel. But I suspect that walking back to cart to correct it's trajectory for every little course change will be more annoying than making constant speed corrections. I think the zero turn evolution will need to happen sooner rather than later. There is a "pro" model of this cart which has dual motors. Maybe it would be a good idea to trade in before the modifications get too extensive.

 

Thats pretty cool project, I think a follower robot is actually pretty useful, get it to carry things for you is a simple and useful thing to be able to do.
Watch out for computer vision itll snow you under, it gets complicated, but I think the solution you have now looks like the best option.

Best CV I can think of, is a 3d yolo type system (3d bounding cubes with identification class id each), beyond that I wouldnt bother, I think its a dead end, thatll pretty much handle a terminator Im pretty sure. Even getting complex Id just keep it simple as possible, or youll never get anything done with your life!!!

I was thinking more like three IR sensors and using triangulation to determine distance. CV is a much more complex thing. Probably comparing signal strength to know the distance. Worst case would be time of flight, but the time difference between 5 feet and ten feet is not much. And maybe CV with an IR reflector scheme. CV and facial recognition would be cool but way more complex in my thinking.
And the really cool thing wouold be to have it going ahead , not following. The benefit is then that turning would slow the cart so that it could be manually turned. Same principle, just different polarities in the logic. Plus no way of it falling behind or wandering off.

 

I was thinking more like three IR sensors and using triangulation to determine distance. CV is a much more complex thing. Probably comparing signal strength to know the distance. Worst case would be time of flight, but the time difference between 5 feet and ten feet is not much. And maybe CV with an IR reflector scheme. CV and facial recognition would be cool but way more complex in my thinking.

Actually OpenCV is not that complex. Well, it can be, but it's pretty easy if you stay on the beaten path. The way he already did it is best IMO. OpenCV in Python is probably way easier than building a custom circuit especially for someone who has strong programming skills and no hardware skills (yet). Plus it opens up all kinds of possibilities like gesture recognition and as you said facial recognition. You could probably even program in some loyalty, a little pregame training exercise where it learns to recognize what clothes you're wearing and doesn't take off on the heels of your golfing buddies if one of them ducks behind a tree to recycle some beer. Might not work so well around dusk or dawn where the sun is shining in its face and all it sees is a a silhouette though.

 

Old is a state of mind, not an age.

Lately it's been more than that for me. It's more a state of body, my mind tells me I can do more than my body allows.

But I suspect that walking back to cart to correct it's trajectory for every little course change will be more annoying than making constant speed corrections

Instead of a twist grip like throttle, I was thinking more along the lines of some of the newer self propelled lawn mowers with variable speed. The ones where you choose the walking speed by how high the handle is lifted.

I don't play golf so I'm not sure how much of the travel across the course is straight line travel.

I don't either but can you imagine this cart coming to you while you move to hit the ball?

 

That is why my suggestion of the cart staying ahead of a retro-reflector clipped to the user's belt And in normal use it would stay about 5 feet ahead, up to possibly 8 or ten feet ahead. Loss of reflection wold initiate a stop. Not quite like a caddie, but easier to get along with. If the user turned then beam loss and stop and get re-aimed in the new direction. This control scheme would be simpler and much cheaper than a MV system. Best part is the MV could be added with very little having to be taken out.

 

Lately it's been more than that for me. It's more a state of body, my mind tells me I can do more than my body allows.

Sorry to hear that, i know it sucks. My dad is barely able to do anything nowdays and it's especially hard on him because he was always so active. I'm getting a taste of it myself, dealing with a broken tailbone and a jacked up middle back for the past 2 weeks after momentarily forgetting I'm not 18 anymore and doing a wheelie on a Honda Grom. I'm 20 years older and 80lbs heavier than I was last time I did a wheelie on a motorbike. Plus that thing is tiny. Went right over backwards.

Instead of a twist grip like throttle, I was thinking more along the lines of some of the newer self propelled lawn mowers with variable speed. The ones where you choose the walking speed by how high the handle is lifted.

First I've heard of this but it sounds smart.

can you imagine this cart coming to you while you move to hit the ball?

Yeah that's actually how I read it the first time around. Had a good chuckle picturing the thing accelerating the closer you get to it (the closer it gets to you).

 

Knowing both the position of the person, and the robot, would help, then the rest is a subordinate detection to those two xyz positions.

 

The most current concept is for the thing to keep a distance ahead of the user, and to stop whenever it loses contact. That avoids having to backtrack to reach it, and also assures that it will be in a position to be turned to go in the next direction. Actual steering would require a whole lot more computing power, and possibly a MV (machine-vision) add-on. That suddenly becomes more complex.

And about the wheelie on the Honda: HEARD ABOUT A NEIGHBOR who tried scrambling on a Honda 750 road bike. He fell somehow, and is quite dead as a result. Left a wife and young boy, aged 5. Some stunts are not wise choices.

 

The most current concept is for the thing to keep a distance ahead of the user,

Isn't it then likely to park itself by the hole while the golfer prepares to putt from the edge of the green?

 

Isn't it then likely to park itself by the hole while the golfer prepares to putt from the edge of the green?

NO, for the reason that the whole package is not automated except for staying ahead of the user while the user is walking along. when the ball location is approached the user will turn in a manner that will halt the IR reflection to the controller, which causes a stop. Thus the cart and bag will remain halted until the user presses the reset button, at which time the motion will resume. It requires a bit of inter action, but it is only automated to stay ahead of the user during travel.
This system is not to be a computerized beast that believes it is much smarter than the user. Thus no microsoft code will be used.

 

NO, for the reason that the whole package is not automated except for staying ahead of the user while the user is walking along.

If that is how the unmodified cart is used, it is different than any manual golf bag cart I've ever seen. Like a chain they are easier to drag than push. I don't golf but have friends that do, and their carts are pulled along not pushed. I'm like Jed Clampett, all that walking and getting agree just ot find golf eggs doesn't seem to be fun to me. From ~1:30

 

And about the wheelie on the Honda: HEARD ABOUT A NEIGHBOR who tried scrambling on a Honda 750 road bike. He fell somehow, and is quite dead as a result. Left a wife and young boy, aged 5. Some stunts are not wise choices.

I appreciate the concern and I don't ride, as a rule. I had my own life threatening incident on a motorcycle at freeway speed 16 years ago. Then 2 years ago my brother in law had a tragic motorcycle incident suffering a TBI he still hasn't recovered from. I have more than enough reinforced incentive to stay away. But a Honda Grom is to motorcycles what a go-cart is to cars. I didn't ride it on the road, just fooling around in the local high school's parking lot. It happened at no more than 20mph. I could just have easily done the same and suffered the same fate on a bicycle.

 

NO, for the reason that the whole package is not automated except for staying ahead of the user while the user is walking along. when the ball location is approached the user will turn in a manner that will halt the IR reflection to the controller, which causes a stop. Thus the cart and bag will remain halted until the user presses the reset button, at which time the motion will resume. It requires a bit of inter action, but it is only automated to stay ahead of the user during travel.
This system is not to be a computerized beast that believes it is much smarter than the user. Thus no microsoft code will be used.

Certainly non-powered golf bag carriers are puled, and certainly that is easier. No question about that. But the appearance of the cart in this discussion looks more toward pushed. And the effort to push a cart that stays ahead of the user o it's own power and control will not be much. The whole concept of the proposal is that the user would point the cart in a direction that they chose and then let go, and the cart will stay between 8 to 10 feet ahead. Thus no pushing needed most of the time, and never an instance of not keeping up and needing to backtrack and recover.