trash_boat

#NOTE: CURRENTLY A WORK IN PROGRESS. What you see below is what our package looks like so far

What this package does

The purpose of this project is to operate a boat with the NViDIA Jetson Computer with a Mid40 Livox Lidar, Basler Camera, and Pixhawk (+ turbines) to autonomously clean up trash from bodies of water. In the future, we will update the exact specifics of our boat and sensors.

So far, when you run our package, it should take in video and identify objects of trash which the boat will go towards to pick up. At the moment, we are working on integrating movement and getting real-world tests in.

Installation

Install ROS Noetic (or another version that uses Python 3)

Link to Guide: http://wiki.ros.org/noetic/Installation/Ubuntu

Creating our Package

Create a folder (a workspace) and cd that folder

mkdir catkin_ws
cd catkin_ws

Create a src folder and open that folder

mkdir src && cd src

Create a package (with any name, ours is trash_boat) in the src folder. You can use the following command:

catkin_create_pkg --rosdistro noetic trash_boat std_msgs rospy roscpp

For more information about packages, look here: https://wiki.ros.org/ROS/Tutorials/CreatingPackage Note: Must include std_msgs and rospy in the command creation

Open the trash_boat folder (or your folder name) and then clone this github repository: https://github.com/EatSumRice/trash_boat

cd trash_boat
git clone https://github.com/EatSumRice/trash_boat.git

Extract all the items out of the trash_boat folder created within the trash_boat folder and then delete the old folder. (basically remove all the old items in the trash_boat folder and replace it with the new items cloned from the github repository)

Then install all the dependencies in the requirements.txt folder.

pip install -r requirements.txt 

Getting other Packages as Dependencies

Installing the Camera Package. Note: Might vary depending on your camera

Follow this guide to install the required packages for the pylon-ros-camera and install it in the same src folder: http://wiki.ros.org/pylon_camera Please note that you may need to install pylon (the deb or tar package): https://www.baslerweb.com/en/products/basler-pylon-camera-software-suite/ If you are running this on an NVIDIA Jetson and are having issues with building the package, this is probably the solution: basler/pylon-ros-camera#108

Installing the Lidar data retrieval package

Install the livox data retriever using this guide: https://github.com/Livox-SDK/livox_ros_driver

Quick Check:

Note that your file structure should look something like this image: https://drive.google.com/file/d/1iwxbmhFR0OyVdmJw_4klo4V31bHIjNKT/view?usp=share_link Robo_catkin is the name of my catkin_ws (but any name works really) Dragandbot, pylon-ros-camera, and ws_livox are both different packages that our package, trash_boat, uses trash_boat is our package

Installing our YOLOv5 algorithm for image object detection

Within the YOLOv5 Classifier folder in the trash_boat package, clone this repository: https://github.com/SophiaLee2023/YOLOv5-Trash-Classifier

git clone git@github.com:SophiaLee2023/YOLOv5-Trash-Classifier.git 

Building the entire catkin workspace

Now, you should catkin_make (which builds the packages). Do these commands within the catkin_ws folder:

source /opt/ros/noetic/setup.bash
catkin_make

If it compiles with no errors, great! Otherwise, there may have been some errors in the installation steps.

Installing some library requirements

Open CV

Install OpenCV onto the home location using this (if it works another way, great!): https://docs.opencv.org/4.x/d7/d9f/tutorial_linux_install.html

Pytorch

Install torch libraries with your decided operating system using pip3 https://pytorch.org/get-started/locally/ and choose the options stable, linux (or your operating system), python, and cuda 11.6 (leftmost cuda) Personally, I had difficulties using anaconda3 to install pytorch, so I wouldn't recommend using anaconda3 for this project. Once you get torch, make sure that the ros file uses python 3 with python --version. If it doesn’t, you might have to change the environment variable PATH (within your .bashrc folder)

MavROS

Install the rest of the dependences in the requirements.txt folder within the src folder in the trash_boat folder Install Mavros https://github.com/mavlink/mavros/blob/master/mavros/README.md#installation

How to Run our Program

First terminal:

roscore		

This is the base to run anything with ROS

Second terminal:

roslaunch pylon_camera pylon_camera_node.launch		

This starts the camera node to retrieve image data. For more information about the camera node, visit: https://github.com/basler/pylon-ros-camera

Third terminal:

roslaunch livox_ros_driver livox_lidar_rviz.launch bd_list:="0TFJBF001663J1"

This starts the connection to the lidar and starts getting data from it.

Fourth Term:

roslaunch trash_boat program.launch	

This prints out camera images and will run the YOLO v5 algorithm on the images (and print bounding boxes on the images).