Pan-Tilt-Follow

This demo from the Edmonton Python Meetup shows something that a couple of people can put together in a couple of days: how Python programs can gather data about the world around them, process that data, and take actions accordingly.

Hardware setup

To build your own, you’ll need:

• A Raspberry Pi computer and camera. This can be as cheap as $70 with the minimum accessories, but could easily cost $150+ if you’re getting a comprehensive kit with the newest model.

  This is a tiny, cheap computer, based on smartphone parts, that has been made to run Linux for educational purposes. See raspberrypi.org for more info.

  The newer ones are faster but more expensive. A Pi 3 or Pi 4 should work fine.

  You can usually buy these in person at Memory Express. Online stores like BuyaPi have more selection and better prices, but you often have to pay for shipping and wait a bit for it to come.

  Don’t forget the following accessories.

  • A power supply: micro-USB for all Pis except the Pi 4 which uses USB-C

  • A micro-SD card for the OS

  • An HDMI adapter may be needed: mini-HDMI for the Pi Zero, and micro-HDMI for the Pi 4

  • For a Pi Zero, you will also need a USB-OTG adapter, and may need a longer camera cable

  • A case is nice but not strictly necessary

• USB Micro-B connector breakout, $2 at robotshop.com

• Pan-tilt kit, $9 including servos, or $5 if you already have servos to re-use.

• Breadboard and jumper wires, purchasable standalone or as part of a $35 introductory electronics kit. Ones that come with tutorials showing how to use all the included parts tend to be higher quality.

• Assorted wood, screws, and double-sided tape, available at any hardware store.

Circuit

A Fritzing schematic is in the schematic folder.

The circuit is fairly straightforward: the pi and the servos all need power, which is distributed using a USB breakout board plugged into a breadboard.

And the two servos get their signal data from separate GPIO pins on the pi.

Software setup

From a new install of raspbian:

• At a terminal, run

  sudo systemctl enable pigpiod
  sudo systemctl start pigpiod
  

In raspi-config:

• Enable camera

Recommended but not necessary:

• Enable SSH

Python

pip3 install --user inputs
sudo apt install -y python3-opencv opencv-data
sudo apt install -y libhdf5-dev libatlas-base-dev libjasper-dev libqtgui4 libqt4-test

Using it

To steer the camera with a gamepad, start

main.py

Note that this can mess up the OpenCV python code later, possibly requiring a reboot.

Then, to run the current face-track demo:

python3 follow.py