Infrared array tuning for 1/4 inch line
I am having trouble getting the 8 infrared sensor array to tune well to a 1/4 inch black tape. It works well with 3/4 inch electrical tape. For a line maze competition, I need t use 1/4 inch though. I am using this tape https://www.michaels.com/westcott-vinyl-artist-tape-black-and-white/10148877.html#q=black+tape&start=5 . What is happening is that as I tune the sensors to see the 1/4 inch tape along the line of sensors, 1 is not sensitive enough while the others are just right. I can't find a good setting for all of them to see the line. Can someone confirm that this array wasn't intended for 1/4 or perhaps I need to be using a different kind of tape, or maybe I can balance the sensors a bit more some how. Thanks for your input.
Comments
-Phil
I'm sure you found the on board variable resistor already, that will help adjust sensitivity.
Also I found that shading the sensor from as much light as possible, gives more reliable results.
1/4" tape sounds very skinny, and if the tape doesn't reflect IR very well, then that will also lead to problems. Is it the same sensor on the array that doesn't work well every time?
-Tommy
But 1/4" is an unusually narrow line. For that, you'd have individual sensors spaced closer than 1/4" apart. Depending on your particular sensor's spacing, you might not get a continuous reading when the line is between two sensors. Your program may have at account for losing the line briefly and remembering the last sensor that saw the line, to continue that maneuver until the next sensor sees it.
When you get it going, you'll also find that max speed and reliability are also dependent on the fore & aft sensor location, that is, how far ahead of the wheels your line sensor is. Other variables include traction, weight, moment of inertia, distance between wheels, etc.
As I mentioned, the 3pi robot has been optimized for this. I am still amazed to watch it accelerate and slow down in the long straights on its second run.
Not line following, but amazing accel/decel and diagonals through the maze:
https://www.youtube.com/watch?v=A4hzCcFikm0
http://www.goldmine-elec-products.com/prodinfo.asp?number=G18740
It's not. Use 1/2" electrical tape. The sensors are spaced too far apart. Ideally two adjacent sensors should see the line at the same time. Otherwise you'll always have to "hunt" the line to know you're going straight.
Even though it's an array, pretend you're only dealing with two sensors. Think of it as binary number counting, 00, 01, 10, 11. With this arrangement the programming become much easier:
11 -- Squarely on the line. Go straight!
01 -- Veering off to the left
10 -- Veering off to the right
00 -- Egads, I've lost the line. Let's go in circles to find it again!
Obviously for this to work the sensors have to be packed close enough so that both will solidly straddle the line at any time.
For an 8-sensor array, you still work with binary counting, only the middle two digits denote centered travel. Values above or below this denote how far from center you are from the line.
Are you limited to 8 sensors, or can you use more?
You might get by nicely with only 5 sensors. Pololu's 3pi certainly does. Take a look at how they are positioned.
www.vishay.com/docs/83751/cny70.pdf`
Yes the array is http://www.parallax.com/product/28034
I have tried tuning the sensors with the resistor, but could not find a good setting, as 1 sensor in particular seems to be much more/less sensitive than the others. This does not seem to be a problem for a ¾ inch line.
There is no limit to the number of sensors I can use.
I am using the Propeller Activity Board, on the boe bot chassis.
The contest is part of the Seattle Robotics Societys Robothon September 20th. Their line follower contest is ¾ inch electrical tape, but the line maze are more difficult ¼ inch lines.
If the sensors were closer in sensitivity to each other, I think I could still use these. I am wondering if I can standardize the signals on my breadboard with voltage dividers. I may try that.
Thank you for the other ideas too. I checked out the sensors in other links. I have also been looking at those capacitive discharge arrays like
http://www.robotshop.com/en/pololu-qtr-8rc-infrared-sensor-array.html
and
http://www.pololu.com/product/1419/
Since the TSL1401 output is analog, it would work well with the Activity Board's built-in ADC. In any event, with the sensor's 128 pixels, the width of the tape would not be an issue.
-Phil
The Line Maze contest requires your bot to stay between the lines. yes?
This might require a different line finding strategy for the different events.
I think the Parallax array will work for both events, as long as it is not malfunctioning, I think one of the sensors on my array is weak also.
I will have to confirm, it has been a while since I played with it.
-Tommy
I guess the array could be attached at a 45 degree angle to the robot, that would tighten the space between the sensors. Sounds inconvenient.
Or, buy two arrays, and put one in front of the other, and offset from each other. Sounds expensive.
Wider Tape, or Narrower Sensors. Hrmm.
We used a shiny Silver Tape at the Parallax 'Giant 555 Circuit' line following contest. Wonder if the course is still there?.
Andy: That's not my bot, otherwise l would!