“Coralbot” Coral Reef Mapping Robot

“Coralbot” Coral Reef Mapping Robot

In the universe, all life as we know it comes from the ocean. If the ocean dies, so do humans. The ocean is the main regulator for the climate, absorbing most of the sun’s heat and have produced more than 75% of the oxygen we breath. Coral reefs are the most important ecosystem on the planet as they provide a safe habitat for marine life to reproduce. It is projected that almost all coral will be dead by 2050. We must act very quickly to explore, map and protect the world’s coral. One of the biggest challenge we have today is that we do not have a reliable, low-cost solution to create coral reef map so that we can monitor the corals and take necessary action.

"Scoutbot" is a semi-autonomous boat that is targeted to map shallow coral reefs and monitor marine protected areas. It has both remote control (RC) capacity as well as route planning with GPS waypoints.

image

It is the combination of 4 major innovations that radically upgrades the state of the art techniques :

  • Diver -> Ocean Drone
  • Quadrat -> Laser Quadrat
  • Hand stitched 2D Map -> AI stitched 3D Map of Coral Reef
  • Computer Vision and AI to classify corals

Our aim is to be

10X more Accurate, 50X Faster, 200X Cheaper than current technology to explore, map, and protect coral reefs all around the world.

Collaborators: Cesar JUNG-HARADA, Ken and Joe CHEW, Eddie YUNG, Sidhant GUPTA, Rohak SINGHAL, Nitin PENDKAR, Kaveesh SIVARAJ, Shoyu KANEKO, Priyank SHARMA, Ethan BUI, Justin GRACE, Dooho HAN, Gökçe ÖZER, Katherine CHIN. Prof KS LUI, Edmund LAM, Chicky YADAV BHAVNANI, Kamakshi YADAV BHAVNANI. Special thanks to the villagers from Abra de Ilog, Ecotone Resilience, Professor Helen T. YAP and Prof Maricor SORIANO, University of the Philippines. Prof. K.S LUI and Prof. Edmund LAM for their support with this project. This project was possible due to contributions by the Gallant HO Experiential Learning Fund at the University of Hong Kong, Green Orange, Maker Family, Scoutbots and MakerBay.

Underwater crosshair lasers form a “Laser Quadrat” that replaces the conventional mechanical quadrat.

image

A high-resolution 2D map is produced.

image

3D map (cloud points) is produced.

image

image

Our AI automatically labelling input image

CoralBot Versions

Coralbot versions

Making of

Research paper

Commonly asked questions

What tools do you use?
  1. Five fake GoPros with single radio control. The water we are in and lighting conditions are very unpredictable, so we are in auto mode mostly. taking a picture every 0.5 seconds.
  2. Ardupilot for GPS waypoint: https://ardupilot.org/
  3. Agisoft Metashape for mapping: https://www.agisoft.com/
  4. OpenCV to adjust the image, and some retouching on photoshop: https://opencv.org/
  5. QGIS to tile the map: https://www.qgis.org/en/site/
  6. We do active labeling with: https://darwin.v7labs.com/
  7. We do the AI with https://learnopencv.com/custom-object-detection-training-using-yolov5/
  8. Our endpoint are on AWS with Sagemaker: https://aws.amazon.com/sagemaker/jumpstart

I want to build on top of your designs and technologies, who should I credit?

“Scoutbots Limited, MakerBay Limited. Cesar JUNG-HARADA, Ken CHEW, Eddie YUNG, Sidhant GUPTA, Rohak SINGHAL, Nitin PENDKAR, Kaveesh SIVARAJ, Shoyu KANEKO, Priyank SHARMA, Ethan BUI, JustinGRACE, Dooho HAN, Gökçe ÖZER, Katherine CHIN, Utkarsh GOEL. Michael O’BRIEN.”

I want to hire you to map my reef. Who can I get in touch with?

We are currently not offering “Mapping as a service” or “selling CoralBot robots” as consumer goods as the design and technology are in early stages. If you are interested to support more R&D, or if you have capital and you are comfortable us developing the technology as we do your mapping request, we should start a conversation. contact@cesarharada.com