Autonomous Docking and Underwater Treasure Search with BlueROV

A project for the control of the BlueROV by Blue Robotics in a 10m by 10m by 6m water tank in Jaume I University, Spain. This project is submitted during the Robotics Challenge at Jaume I University.
.
View Project Files »

Description

The BlueROV is a remotely operated underwater vehicle that has four vertical thrusters and four aximuthally-positioned horizontal thrusters. The robot has a couple of sensors integrated into it, including: depth sensor or pressure sensor, sonars and high-definition camera. The robot is usually conntected to the ground station using an optical fibre or any other suitable cable. This Robot was used in the Robotics Challenge in CIRTESU, Jaume I University.

The Robotics Challenge involves the localization and control of the BlueROV for autonomous docking, treasure search and static object inspection. However, in this repository only the first two tasks are implemented.

Image of (Heavy Configuration) BlueROV

Tasks Completed in the project include:

1. Generation of image data from video.
2. Training of Object Detection model.
3. Deploying Object Detection model in ROS Environment
4. Visual Servoing (Robot control using continuous image feed).
5. Testing in a controlled environment

Videos are otained from the treasures that are desired to the hunted. These videos are converted to images in the VideoToPic folder using the VideoToPic/vid_to_pic.py file. Thereafter, the extracted images are processed on RoboFlow - annotation, augmentation and bounding-box definitions, class definition, etc. Then, the model was trained on Yolov5 model using Google Colab. The codes for the remaining tasks are implemented in catkin_ws/src/vision_sim folder.

Tools Used

• Python Programming
• RoboFlow
• Robot Operating System
• PyTorch
• OpenCV

Usage

1. ROS Installation is required. The Unity Engine used is required (for simulation before testing in the pool).
2. Launch the Unity Simulation Engine which is connected to ROS using ROS-TCP-Endpoint
3. Execute the launch file using rosrun vision_sim aruco_detection_from_camera.py from a Linux terminal
4. You can stop the execution using Ctrl + C on the terminal.

Results

unity-docking.mp4

Docking of the BlueROV autonomously in the Unity Simulation Environment

unity-treasurehunt.mp4

Treasure Search with the BlueROV in the Unity Simulation Environment

pool-docking.mp4

Docking of the BlueROV autonomously in the Test Pool

Contact Contributors

• Farooq Olanrewaju - [email protected], [email protected]
• Md Sazidur
• AbdelHaleem Saad
• Maria Saleem

Supervisor: Prof. Pedro Sanz, Jaume I University; Prof V. Hugel, Director COSMER Lab, Université de Toulon.

(back to top)