rpi-quadcopter

A quadcopter running a Raspberry Pi 4

This project is aimed at getting a quadcopter to run using a single Raspberry Pi 4 8GB board with a PiCam, an Adafruit BNO055 9DOF IMU and an Adafruit DPS310 Temp/altimeter. It will run on ROS2 Foxy on Ubuntu 20.04 Server 64bit. Initially I'll be focusing on software, reading the sensors, then sensor fusion and visual-inertial navigation using the PiCam. I hope to be able to do everything up to PID control on the Pi but it may be necessary to also use an Arduino to control the motors (if I can't get an ESC to do the job).

Setup

Download the Raspberry Pi Imager onto your host machine. When choosing the OS, select 'Other general purpose OS', then 'Ubuntu', then 'Ubuntu Server 20.04 64 bit'.

WiFi

Unless you have a wired ethernet connection, you will want to connect the Pi to WiFi. We do this as follows

$ls /sys/class/net

eth0 lo wlan0

edit the network configuration file to add the section on wifis, using your wifi name (e.g. wlan0). Be careful with the indentation!

sudo nano /etc/netplan/50-cloud-init.yaml

# This file is generated from information provided by the datasource.  Changes
# to it will not persist across an instance reboot.  To disable cloud-init's
# network configuration capabilities, write a file
# /etc/cloud/cloud.cfg.d/99-disable-network-config.cfg with the following:
# network: {config: disabled}
network:
    ethernets:
        eth0:
            dhcp4: true
            optional: true
    version: 2
    wifis:
        wlan0:
            optional: true
            access-points:
                "YourSSID":
                    password: "YourPassword"
            dhcp4: true

Run sudo netplan --debug try and then sudo netplan --debug apply to appy the changes.

Edit the cloud-init config to disable init

sudo nano /etc/cloud/cloud.cfg.d/99-disable-network-config.cfg

network: {config: disabled}

Desktop Environment

Ubuntu 20.04 does not have a desktop environment, which makes it difficult (but not impossible) to develop on the Pi. Since we want to run ROS2 and make use of tools like rqt_graph and rqt_plot we want to install ubuntu desktop. We do this as follows

$ sudo apt update
$ sudo apt upgrade
$ sudo apt install ubuntu-desktop
$ reboot

ROS2

Install ROS2 following the installation instructions

PiCam

Now that ROS2 is installed, we can visualise the camera data with cam2image from image_tools as follows ros2 run image_tools cam2image.

cam2image is quite limited as does not allow you to change the resolution or other settings. As described in this post you can use the Linux Video4Linux driver instead.

$ mkdir -p ~/ros2_ws/src && cd ~/ros2_ws/src
$ git clone --branch foxy https://gitlab.com/boldhearts/ros2_v4l2_camera.git
$ cd ..
$ colcon build
$ . install/local_setup.bash

Now we can run ros2 run v4l2_camera v4l2_camera_node and in another terminal ros2 run rqt_image_view rqt_image_view.

The ros2_v4l2_camera driver allows a lot of control over the camera, check it out.

GPIO

GPIO should be easy on the RPi and an initial search for GPIO on Raspberry Pi on Ubuntu finds this. Unfortunately this tutorial assumes 21.04 but we've installed 20.04 (for ROS2 LTS support).

Reading through many answers we find that we need to create a group that is permissioned to access /dev/gpiomem. This group already existing on Raspbian but not on Ubuntu 20.04. /dev/gpiomem is the part of memory where the GPIO bits map to the actual hardware. We do this as follows by creating a new gpio group, adding the current user to that group, giving the group ownership rights to /dev/gpiomem and then allowing group read/write access to /dev/gpiomem. A reboot is required for the new group and permissions to take effect.

$ sudo groupadd gpio
$ sudo usermod -a -G gpio $USER
$ sudo chown root:gpio /dev/gpiomem
$ sudo chmod g+rw /dev/gpiomem
$ reboot

we can test that we have access GPIO 18 without requiring sudo rights by running

$ echo 18 > /sys/class/gpio/export 
$ echo "out" > /sys/class/gpio/gpio18/direction
$ echo 1 > /sys/class/gpio/gpio18/value
$ echo 18 > /sys/class/gpio/unexport

I2C

Since our sensors support i2c we can test that they are working by running the following:

$ sudo apt-get install i2c-tools

Since we are using Stemma-QT (also known as Qwiic or JST 4-pin SH) cables to connect the Adafruit sensors, we can attach the 4 cables to the GPIO. Convention is that red is 3.3V, black is GND, blue is SDA (data) and yellow is SCL (clock). On the RPi 4, SDA is GPIO 2 and SCL is GPIO3.

Plug in the sensor and then run

$ sudo i2cdetect -y 1

We can find out what the I2C adapter number is (each I2C device gets a unique number starting from 0) by running

$ i2cdetect -l
i2c-1            i2c             bcm2835 (i2c@7e804000)                          I2C Adapter

We will need to install libi2c to communicate with the device

$ sudo apt-get install libi2c-dev