Line Follower - Need Help - Details Inside

Image of Line Follower


Link to video:
https://www.youtube.com/watch?v=VF5w7Ct5s8Q

Meat and bones:
Full disclosure, this is a school project. I have, however, never implemented a PID before. Also of note, is that I don't want to get lost any more than necessary into control theory. I just happened to miss this class first year, so I'm taking advantage of an opportunity to dig into the PID equation a bit, and implement it in C.

The four black squares are a sensor array. I'm going to use them to track the line. Adjustments will be made by changing the speed of the wheels, however the robot will only go forward.

I know I will need a calibration. Putting the robot on the line will enable me to get a set point pretty easily. What I'm confused on is how I will handle the drift left or right across the four sensors. I'm not really seeing how that will figure in, and I guess that is my first hurdle there. Any ideas on how to approach this from a more macroscopic level?

What kind of sensors are you using?

jaden said:

What kind of sensors are you using?

Click to expand...

https://katalog.we-online.com/led/datasheet/156120YS75300.pdf

They are analog sensors. So they will be able to measure the variance in either direction. I would think that I would need to set up some kind of span for each way, with right on top of the line being my zero point, though. I feel like I need to superimpose a coordinate axis on here.

tracking all depends in my experience with the conveyor
being at least 4 times longer than it is wide.


james

The catalog you have linked to shows LEDs, not an analogue sensor.

That is alright. The sensors (believe it or not) are completely irrelevant. They can be abstracted out. I can update a little bit with some more information:

• Wheels are the same size.
• Motors spin based on PWM signal and are controlled independently
• Sensors give an analog feedback value (exact value is irrelevant)

It seems that if I lay this on top of an x-y coordinate plane, I can calibrate the "zero" or origin of the 4 sensors. This will be my setpoint that I want to maintain. The total change in x (or dx) should be about the same between the sensors(dx). So I believe that I could use the average of dx between the four sensors as my error. Thoughts on this?

Next, I'm looking at the velocities of each wheel. Now, my process variable would be the two wheels. They will always be turning "forward", which is +y, but with different speeds so as to correct for dx. This is where I'm hung up at the moment in getting an overall look at how to implement this.

Also, just for completeness, here is the correct sensor link:
https://www.onsemi.com/pub/Collateral/QRE1113-D.PDF