Tile detection using machine vision systems and the Unitree Go1 Edu robotic dog

This code repository contains the source code and test data for our school project. The code is only tested on Linux, (specifically KDE Neon, but Ubuntu should be fine) as we had problems connecting to the dog using other operating systems

Prerequisites

Installation requirements

This project uses the Linux command emulator Konsole to display data on the dog via the SSH connection. You may use any terminal emulator you want, but will then have to change the command strings using the emulator commands in robot_controls/server/ssh.py.

Konsole can be installed using

sudo apt install konsole

You will then have to install the python requirements by running

python3 install -r requirements.txt

Setup of SSH

The clients in the dog will be started automatically over ssh by the server. To make this possible, you must add some configuration to your /.ssh/config file

Host dog_pi
        HostName 192.168.12.1
        User pi

Host dog_head
        HostName 192.168.123.13
        User unitree
        ProxyCommand ssh -q -W %h:%p dog_pi

Then add your ssh key to the dog by running

ssh-copy-id dog_pi
ssh-copy-id dog_head

The password for both devices are 123

Usage

1. Put the dog on the ground, and position the rear legs in the same position as the front legs were in the box
2. Start the dog by first clicking the on/off button on the side once, and then holding it in until the lights start flashing
3. Wait until the robot starts and goes into standing position
4. Connect to the Unitree hotspot created by the dog

Manual control

Start the server_manual.py file

Automatic walking

Start the server_autonomous.py file

The autonomous server is also able to receive commands directly from the terminal. These include

Forward: Dog starts walking
Stop: Dog stops walking
Turn: Dog starts turning in the next determined direction Calibrate: Dog resets all sensors based on its current location

Testing on a video

It is also possible to test the detection system on a video by running the Start the video_loader.py file