0
$\begingroup$

I asked a different question a while back that led me to realize the hardest part of my project, position sensing. I had an idea and wanted to know if I'm trying to reinvent the wheel.

Specifically, I was considering the idea of a wheel orientation sensor that essentially is a circular PCB with copper channels going out from the center like bicycle spokes. And depending on which channel(s) completes the circuit, I know at least within a small degree of error, which I can figure out with math/trig, what direction my robot is pointed. At that point I might use a fine tune algorithm to further point it, which should be easy. The idea is to use this to orient a laser. I looked into potentiometers and I think the precision I need would cost way more than I'm willing to pay, if it even exists.

Two questions, am I reinventing the wheel? Is there some sensor I can buy that's already essentially this, and has a good amount of precision? If I design it myself, what are some good ways to sense which channel(s) are completing the circuit? I can think of various ways, but none are elegant and simple. (Edit - one way I dreamed up is a digital bus, with one value per-channel that gets sent to the bus if the channel completes. Not exactly sure how to design this off the top of my head but I think it's possible. I can design my own board and stuff if I need to)

This could be a really good solution for me, I think, because what I can do is use math to reduce the gearing to get more precision out of my sensor. In other words, if the sensor wheel spins a lot faster than the orientation wheel, I can use whatever ratio I want to increase the precision of the robot. This comes at the cost of more time spinning the motor of course, which is fine with me.

Can I make this work?

Edit: The precision I need is: as much as possible :). It's for a buddy's pool table as a ball placement device. I need to point at various 1mm spots about 3-5 feet away. I want no more than 2mm error, preferably 1mm.

Also, I am going to store and load positions of the robot, so this circular "wire potentiometer" (I guess that's a thing?) should help with that as well, especially if I can figure out an elegant solution to the channel detection problem.

$\endgroup$
13
  • $\begingroup$ What is "precise" ? Any idea of the change in angle to move 1" at 100 yards? $\endgroup$
    – Solar Mike
    Commented Apr 1, 2023 at 7:32
  • $\begingroup$ @SolarMike Answered with an edit, thanks for that, I should have been more precise with my question in the first place $\endgroup$
    – neural
    Commented Apr 1, 2023 at 8:05
  • $\begingroup$ So, how does your 1mm at 5 feet translate into an angle? +/- 5 degrees? 1 degree? 30 seconds? 4 seconds? $\endgroup$
    – Solar Mike
    Commented Apr 1, 2023 at 8:17
  • $\begingroup$ OK, my math is a little rusty, but the answer I came up with is (degrees): 0.02221896082497133662762253845753 (and yes, that's the correct order of magnitude) $\endgroup$
    – neural
    Commented Apr 1, 2023 at 8:50
  • $\begingroup$ So how big is your circular pcb and how many channels? What resolution will you get? Close to your 0.02 degree? $\endgroup$
    – Solar Mike
    Commented Apr 1, 2023 at 8:52

2 Answers 2

0
$\begingroup$

@AJN answered my question. I think a wheel encoder and gear ratios can get me the precision I need.

I can calibrate it by pointing it at a spot manually, and then use wheel encoders on two axes to detect the number of turns. This plus gear ratios and math should give me the precision I need to automatically position the laser. I'm excited about this idea, it's simple and doable.

If @AJN would like to post an answer for credit I will mark it.

$\endgroup$
2
  • $\begingroup$ Be aware that gears exhibit backlash. $\endgroup$
    – Transistor
    Commented Apr 1, 2023 at 9:45
  • $\begingroup$ @Transistor - thanks, yeah, I need good gearing most likely. I'll figure out in the POC if it works or not, I can measure the output using string, etc. $\endgroup$
    – neural
    Commented Apr 2, 2023 at 2:08
0
$\begingroup$

Current drone technology appears to have this problem well sorted. It includes motion sensors and GPS. You should be able to find a package solution in that rabbit-hole. PS 1" @ 100yards is approximately 1 minute of angle.

$\endgroup$
1
  • $\begingroup$ Hmm, are there any packages you can recommend? I've looked around a bit but I probably don't know what to be looking for. I'm open to the idea for sure though, anything to save a lot of work of designing my own thing. $\endgroup$
    – neural
    Commented Apr 17, 2023 at 20:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.