-1

I am trying to figure out the possible solutions for detecting the position of the object in the container. the below image is to describe the scenario.

Object: spherical diameter: 5 - 7mm

Type: Ferrous or stainless steel or non ferrous (flexible)

container space to allow the object to move: 8 to 10 mm maximum.

Constraint: The solution has to be cheap. I cannot use camera both because of the cost and the power requirement. The object will be moving continuously but t a rate IR sensor would detect it.

This post is similar to my query but I need second opinion about possibility of the similar solution as i need to precise judge the X-Y position of the object (precise enough to judge whether the object is on the left or to the right of the wall). Also, considering stainless steel object.

enter image description here

I have found only this solution feasible so far.

  1. Reflective optical sensors in the wall (Example part)
  2. Separate Infrared transmitter and receiver pair (This will consume more power than the above one, i think)

It will be great help if you can suggest if i am on right track or should look at some other solutions.

Edit: if the constraint is only to find in which part the ball is, do I have any easy low cost solution? Considering the image is divided into 12 equal parts. The size of the wooden plank is roughly 15 cm x 15 cm. Thank you in Advance.

Community
  • 1
User323693
  • 9,151
  • 4
  • 21
  • 50
  • 1
    I'm not sure we have all the constraints here, because the obvious solution is a camera showing us the same view that we have in the picture. If that's not acceptable, why not? – Neil_UK Aug 29 '16 at 12:57
  • @Neil. Camera will be costly solution and power hungry too. I have updated the constraints. – User323693 Aug 29 '16 at 13:02
  • If you try to use the sensors you linked to, you will either need a metric crapload of them, or else you and I have very different ideas about what it means to precisely locate a ball bearing in a maze. – JRE Aug 29 '16 at 13:03
  • Are you trying to guide the ball into the pocket at the center of the maze? If so, a camera is the way to go. Video cameras don't have to be power hungry, and a single board computer (Raspberry Pi) should be up to the task of tracking the ball when using something like OpenCV. So, less than 1A at 5Volts should get you a workable system. Especially if the goal is to have a computer guide the ball into the pocket. – JRE Aug 29 '16 at 13:09
  • @JRE the design is meant to be very cheap. So, i am planing only to step up to Arduino Mega. Hence, a low cost solution as i am looking for – User323693 Aug 29 '16 at 13:11
  • By the time you've bought enough of those sensors to be able to track the ball accurately, you could have bought a Pi and a video camera. This assumes you are trying to guide the ball. You still haven't explained the purpose of tracking the ball. – JRE Aug 29 '16 at 13:18
  • @JRE in fact you are right. This is meant to be th3 cheap version and hence I am thinking also about relaxed constraint. I have done an edit in the question. Can you check, if there is a different kind of solution possible now? Thanks. I will also experiment with multiple wire loops placed below the board to detect the present of the ball..tracking the ball is to finally guide the user to put ball in the hole. – User323693 Aug 29 '16 at 13:47
  • If you use a ball that is magnetic, inductive sensing might work (not sure of just how inexpensive your requirement is) http://www.ti.com/lsds/ti/sensors/inductive-sensing-overview.page?keyMatch=inductive%20sensor&tisearch=Search-EN-Everything – Peter Smith Aug 29 '16 at 14:00
  • If you only need t know in which of the "hallways" the ball is in, then you might get away with using emitter/detector pairs aimed down each "hallway.". The problem being when the ball is in a corner. Then it will be detected in two " hallways.". I count 15 halls. – JRE Aug 29 '16 at 14:12
  • I'm wondering if a capacitive sensor for LCD screens stuck under the board (or maybe laminated within it) would work. Maybe someone would be kind enough to ridicule that idea or reinforce it. – Andrew Morton Aug 29 '16 at 14:24

3 Answers3

2

Option 1:

Constraint: The solution has to be cheap. I cannot use camera both because of the cost and the placement.

Consider a discarded Android phone running YOUR program calling OpenCV library methods to detect the position of the ball. It is likely you can do this even if you place the phone at an angle. As long as it can always see the ball. Of course an above view would be easier to program (ball is always round and (nearly) always the same size).

Option 2:

Weigh the board at 2 points and hing it at the opposite side. You may have to build this into 2 levels to get 2 completely free degrees of movement. Approximate the ball's position by reading the difference in the weight of the board as the ball moves around.

Option 3:

Consider painting the board black (non-IR-reflective color / maybe sun-screen) and keeping the ball highly reflective. Maybe even painting it white (IR-reflective-color). Now multiplex the IR transmitters position to shine down each path. Have a IR receiver above. The strongest reflection should only occur when the ball is illuminated.

st2000
  • 3,228
  • 9
  • 12
  • I can think of a couple of more in case you are interested. One is simply weighing the board at 2 points. But I'm not sure you can obtain the accuracy to manage more than to approximate the balls position. – st2000 Aug 29 '16 at 14:39
  • Thank you. Is it possible to share the image or link for more details for option 3? – User323693 Aug 30 '16 at 03:42
  • I don't know of any examples. I just thought that one up. I would experiment with ir leds pointed down a path or two just to see if it works. I would modulate the leds at about 38KHz to 40KHz and try using 38KHz to 40 KHz beam interrupt detectors. Not remote control receivers. Remote control receivers have an agc circuit and will eventually always find the signal. Ball or no ball. – st2000 Aug 30 '16 at 05:38
1

Wall mounted sensors require a continuous array of Optical interrupters for path blocking seems to be overly complex. The matrix would need logic to define the maze, direction and motion of the object.

It seems to be a job better suited for rats.

YOu could try IR LED emitters at one end of each path with an array of PD's mounted in the side wall for radial reflections. Using a pin hole apertures and a pin hole emitter off centre at the end to get 90 degree reflections. What you do with that info is not stated.

Given the insufficient wall thickness for 5mm devices, perhaps you can mount them vertically with a 90 deg horizontal reflector to the lane pinhole to detect the end positions reached and the lane blocked to simplify your puzzle of tracking a spherical steel ball.

The detection of 2 PD's reflected or blocked axial signals at an intersection could locate the corrordinates.

Tony Stewart EE75
  • 1
  • 3
  • 54
  • 182
  • thank you. I cannot afford camera due to cost and power reason. I have added the constraint in the question now. Any other solution is greatly appreciated. – User323693 Aug 29 '16 at 13:05
  • Thank you for your suggestion. What is the meaning of PDs. Sorry, i didn't get it. Also, can you give an example or a link, the idea you are suggestion about please? – User323693 Aug 30 '16 at 03:41
  • Photo Diode PD., no links just common use of transmission block of signal for short path between emitter and PD which is predictable unlike photo transistors with variable hFE – Tony Stewart EE75 Aug 30 '16 at 03:49
  • The logic is to install emitter in direction of travel and when next path is blocked, follow that vector Up down left or right as predetermined by relative location. – Tony Stewart EE75 Aug 30 '16 at 03:58
  • If a camera is too expensive and power hungry, dozens of photoelements probably are as well – Scott Seidman Aug 31 '16 at 11:04
  • agreed, hopeless idea. – Tony Stewart EE75 Aug 31 '16 at 13:25
1

The cheapest, fastest way I could think of (that isn't video based) would be to make the floor some sort of resistive touch screen.

Scott Seidman
  • 29,274
  • 4
  • 44
  • 109