This guide will help you get started with the line tracking algorithm on the Pixy2.

PixyMon will display what Pixy2 detects graphically and is a good way to familiarize yourself with what the line tracking algorithm is capable of, and it's a good way to make adjustments. Start by selecting the line tracking algorithm in the Program menu.

Draw a line on a piece of paper and put it in front of Pixy2. Pencil will work, but black marker is usually better. You should see something like this:

What Pixy2 has done is found the best line candidate to track and has assigned it to the Vector. The Vector is the line that Pixy2 (or your robot) is currently tracking/following. You can move the paper and see that Pixy2 can track the line fairly quickly. PixyMon slows down the tracking algorithm typically to 20-30 frames-per-second because it takes some time to send the raw frame data across USB, but if you change the view to “no background” (see pic below), you can get a feel for what the tracking algorithm looks like at the native 60 frames-per-second.

This (60 fps) is framerate your robot will see.

You can also see what happens when you turn on Pixy2's lamp.

Note, with the lamp on, Pixy2 will typically reduce the exposure (sampling) time of the image, which reduces motion blur. In general, motion blur happens because a camera is sampling a subject while the subject is moving. In this case, the subject is a line (or set of lines). If they are blurred due to motion (the motion of your robot), it can make the lines more difficult for Pixy2 to detect and track. Pixy2's lamp can help with this. The upshot is that the lamp make your robot go faster!

Go ahead and change back to the “All features” view if you switched views. Make a 4-way intersection by drawing another line through the original line and place it in front of Pixy2. Pixy2 will identify the intersection and quickly choose the default branch of the intersection, which is the “straightest” branch of the intersection.

You can change the “default turn angle” in the Expert configure tab. You'll see that the default turn angle is 0 (by default), which is straight. The angle is based on a straight-referenced angle coordinate scheme.

Angle coordinates from Pixy2's perspective.

Try changing the default turn angle to -90 (right-most turn/branch) and click on Apply.

Now when you place the intersection in front of Pixy2, it will choose the right-most branch of the intersection.

By default, Pixy2 will choose the branch automatically based on default turn angle or the most recent call to setNextTurn(). You can change this behavior by enabling “delayed turn” in the Expert configure tab.

Go ahead and enable delayed turn and click on Apply.

Now when you place the intersection in front of Pixy2, it won't choose the branch automatically. Instead Pixy2 will label the intersection because, it doesn't know what action to take.

In delayed turn mode, Pixy2 will wait for your controller to call setNextTurn() which will instruct Pixy2 to choose the branch of the intersection that matches the angle the closest.

We wanted Pixy2 to be able to receive simple commands or information while it is performing line tracking, so we added the barcode feature. The barcodes are intended to act as “roadsigns”. You can add graphics to the barcodes that are human-readable as long as you keep the barcode unobstructed. To test the barcode functionality, download the print the all_codes.pdf document.

Holding the document in front of Pixy2 while it is running the line tracking algorithm will yield something like this:

If you haven't hooked your Arduino to your Pixy2 yet, use the supplied Arduino cable to connect Pixy to your Arduino as shown below.

Note: if you're using Arduino Nano, the ribbon cable faces the interior of the Nano, it doesn't exit off the side like on the Uno (pictured).
If the cable is plugged in backwards, you won't be able to upload to the Nano, or have serial communication with it.

Next, download the latest Arduino library “arduino_pixy2-x.y.z.zip” here. Bring up the Arduino IDE and import the Pixy library by selecting Sketch➜Include Library➜Add .ZIP Library… (or if you're using an older version Sketch➜Import Library) in the Arduino IDE, and then browsing to the Arduino zip file that you just downloaded.

Next, load the “line_hello_world” example by selecting it in File➜Examples➜Pixy2. Upload it and bring up the Serial Monitor. You should see messages printed that look similar to this.

vector: (38 51) (39 9) index: 0 flags 4
vector: (37 51) (38 10) index: 0 flags 4
vector: (37 51) (40 12) index: 0 flags 4
vector: (37 51) (37 13) index: 0 flags 4
vector: (36 51) (40 15) index: 0 flags 4
vector: (36 16) (38 0) index: 2 flags 4
intersection: (36 16)
  0: index: 7 angle: -88
  1: index: 2 angle: -7
  2: index: 1 angle: 93
  3: index: 0 angle: 180
vector: (34 16) (37 0) index: 2 flags 4
vector: (36 19) (37 0) index: 2 flags 4
vector: (39 21) (37 0) index: 2 flags 4
vector: (35 22) (37 0) index: 2 flags 4
vector: (35 23) (37 0) index: 2 flags 4
vector: (35 24) (37 0) index: 2 flags 4
vector: (35 25) (37 0) index: 2 flags 4
...

In particular, when you hold a barcode, line or intersection, you can see the numbers that Pixy2 returns. Bear in mind that the numbers Pixy2 returns are based on the coordinate system below.

Line tracking image coordinates from Pixy2's perspective.

Also note that the intersections and barcodes are only reported once every time the intersection or barcode comes into view. This means you (the programmer) doesn't need to keep track of which intersections and barcodes have been seen before and which have not.

The line tracking API can be found here. It applies to Arduino (SPI interface) as well as the Raspberry Pi, or similar controllers (USB interface).