ECOBOT

https://docs.google.com/presentation/d/1zbF-cug1iHLrZvG5N5WFX9I47XMYep60/edit#slide=id.p1

https://sites.google.com/view/ecobot-gui/home

Background:

After reviewing many resources of Nasa and other worldwide institutions implementation for monitoring, detecting, and collecting marine debris, Starting by studying Nasa applicable solutions to track every type of marine debris like microplastic debris and looking into their response to disaster debris Emergency response OR & Rs Marine debris we found a great effort for tracking and monitoring marine debris and less collecting, and the collecting efforts are missing new updates of nowadays strong smart AI/ML capable solutions; so we developed a solution to back up Nasa efforts not only for monitoring, tracking, and quantifying debris but also for collecting every type of the mentioned marine debris in Nasa space apps challenge and in resources provided to us.

Our Vision:

We developed a solution that is capable of quantifying and detecting and collecting marine debris. based on the SWARM Intelligence, which is the collective behavior of decentralized, self-organized systems, natural or artificial.

The kernel of our system is the AI/ML model we programmed to detect, monitor, quantify marine debris with an accuracy of 89 %. The model has another job which is guiding the Unmanned Surface Viechls (USV) and Unmanned Ariel Viechle (UAV) through their journey in the lake, river, or ocean.

Our design & Implementation:

For the purpose of removing the garbage batches mentioned in Nasa resources and every type of debris starting from "Microplastic debris", "Medium size Objects", "Large size objects", and Disaster debris; we redesigned an old design made by Razer Technologies to collect every type of the previously mentioned debris.

Fig. 1

Adding a group of sensors, cameras, and thermal camera alongside a powerful processing unit to help run the SWARMs & SWARM Units, the processing unit is a Cluster of (Rasberry pi 4) running a Debian Buster os & MPI library to handle parallel processing.

To redesign Razer technologies design we studied the environmental conditions that the design will operate in, by reviewing Nasa resource of Earth data. The USV will operate in swarms in every mission and our design have the adjustability to allow a single unit of USV to operate and go on a customed mission.

More explanations of the missions, modes, and ECOBOT body design were published on our website.ECOBOT

ECOBOT The SWARM:

Ecobot is based on SWARM Intelligence and is divided into two main parts also it operates in two cases:

1st case: is what we called the Loose case, where there is no special environmental condition for the Swarm to operate in.

2nd case: is the disaster Alarm case what we call a SWAT CALL, the SWAT is every Special Weapons And Tactics,

where we saw the Environment gathers its forces to show off we also showed off.

Finally, we added a special net design and ECOBOT Body for mico plastic collection and monitoring, more about it on our website.

Fig.2

More about how the swarm function on our ECOBOT website.

ECOBOT Workflow:

Fig.3

more about the Loose case in our website ECOBOT.

For every case previously mentioned in the workflow, we designed a ECOBOT Body and more about it on the website.

We reviewed the Emergency response of Nasa to Disaster debris and designed a backup for help called the SWAT CALL, in this case

there is a human factor observer looks at the ECOBOT Desktop application we designed to complete the dedicated system for monitoring, detecting, quantifying, and collecting depending on a diaster map we implemented relying on GDACs

which provides an updated map of the disasters in the world, we intended to use a world map for a disaster report for the sake of our target to allow every country to get rid of the marine debris and have a healthy environment for humans and creatures

Fig.4-a

Fig.4-b

more about the disaster map and the SWAT CALL process on our website.

Depending on this map a human can observe a disaster in the region they are in and after the END of the disaster, they send a SWAT CALL to do the mission and receive reports about the mission.

Mission Controller:

We aimed to create a dedicated whole system to accomplish our target to a cleaner earth. so added a dedicated developed Desktop application to configure the USV missions, sensors calibrations and also visualize the reports and control The SWARMS TEAM to control the swarm mission in Marine Debris removal.

Fig.5

Although our implementation is missing parts, yet it's capable of configuring the whole USV & UAV SWARMS, and more details about it on our website.

The previous Snapshot is a life map of our city sheiben elkom in El Monifia governorate, a waypoint mission is assigned to a USV

swarm, the values at the panel shows the number of USVs in service and out of service and the total number of them.

more about the GUI also on our website with more information about calibration tabs.

The Kernel of our Design is the AI/ML model that we had trained using Nasa provided resources, and random images a total dataset of 2035 images we classified them into 19 categories:

• Big objects
• Cloths
• Disaster debris
• Mass of plastic debris
• Microplastic
• Plastic and glass bottles
• Wood
• cans
• cigarettes
• coral reef debris
• fishing net
• glass
• metal
• paper
• plastic
• surface environmental debris
• suspended object
• undefined objects
• victims of debris

Classes we created to train AI model

we created these classed a manually labeled them and constructed our dataset, then created our model a complete code was uploaded on Github on this link https://github.com/MostafaBoshta/EcoBot . we reached an accuracy of 89% using MobileNetV2, a Convolutional Neural network.Also, we trained our model to detect microplastic and collect it using a special design more about it on our website.

Mission controller block Diagram:

e mentioned that we created a Cluster of Raspberry pi 4, on which the model is uploaded also it runs the system that controls the USV Unit. This is a block diagram of our design, More details about this black diagram are on our website.

Finally, after the mission completes the SWARM sends a report of the detected and collected debris to a visualization board, that is updated after completion of every mission. And we used Tableau with NASA datasets to visualize different debris monitored.

More details about each part of our project are on our website.

Advantages of our Solution over relaying on Remote sensing:

Remote sensing beyond limitations:

· Despite Using Remote Sensing, We Managed to Decrease the cost Used in this Module.

· And We Managed to Solve the Confusion and Misclassification of the Objects by Using Machine Learning and AI Models with High

Accuracy to Reduce the Errors.

· We Used Camera modules to Prevent Distortions in Images and Getting Clear Images.

And Provide an Organized Data Base to Classify the Data Coming from Different Sources as We have USV & UAV Modules.

.Applying our solution solves the problem of Cost and friendly using that faced remote sensing relying on satellites

To collect the images of our dataset to train our model we collected different images from Nasa Marine Depris Program Office and restoration:

https://marinedebris.noaa.gov/

Also, we trained our model on images underwater so we used old archive images of marine debris in coral reefs :

https://www.ncei.noaa.gov/access/metadata/landing-page/bin/iso?id=gov.noaa.nodc:0209247

We did not find microplastic debris dataset on NASA datasets but we visualized a microplastic debris quantity and locations using Nasa Dataset CSV format:

https://www.ncei.noaa.gov/metadata/geoportal/rest/metadata/item/gov.noaa.nodc%3A0211009/html

All of our code and are available on GitHub link:

https://github.com/MostafaBoshta/EcoBot

At the bigging of our journey, we gathered to read the summary of the challenge and choose our challenge, but this challenge captured us as humans then we knew as engineers it's our job to make the world a healthier place for humanity and living creatures. $Finally, it became clear for us to choose this challenge.

All of us made a great experience creating developing and searching, some of us learned a new programming language, other some found new ideas for graduation projects, others developed a new good habit :), we all collaborated to come out with this solution. At first, we wanted to create a practical prototype but due to the lack of funds, we didn't manage to.

Also, we lost a lot of time looking for a sponsor to help us implement our solution, but unfortunately, we didn't.

And we were left with only two weeks to finish our GUI and AI model also the Database visualization.

We really would like to thank our local lead team because of their support to us and fast response to any help we needed.

for the study of environmental conditions on our system we reviewed :

https://earthdata.nasa.gov/learn/discipline/ocean

For reviewing Nasa response to Emergencies of Disaster debris we reviewed:

https://marinedebris.noaa.gov/our-work/emergency-response

For studying the effect of microplastic debris we reviewed:

https://www.nasa.gov/feature/esnt2021/scientists-use-nasa-satellite-data-to-track-ocean-microplastics-from-space

We developed our model using Google collab:

https://colab.research.google.com/

To implement our disaster map we used:

https://gdacs.org/

To create our visualization we used:

https://www.tableau.com/

We created a poster to spread the awareness of the danger and effect of Marine debris using:

https://www.canva.com/

#MarineDebris #AI/ML #Hardware #sofware #desktopApplication #UAV&USV #unmanned_Viechles #Aqua_marine_creatures #Marine_life

This project has been submitted for consideration during the Judging process.