Team: 14
School: Los Alamos High
Area of Science: Engineering
Proposal:
Problem
Despite many efforts to clean up marine debris, 5.25 trillion pieces of plastic still remain in the ocean, killing countless marine species and damaging ecosystems. Understanding the concentrations of ocean trash and tracking the movements of debris is crucial in the efficient cleanup of the world's seas.
Currently, ocean plastic is mapped through satellite imagery, but the displacement of trash below the surface makes it difficult to identify. Before cleanup can be initiated, we must have better methods to gauge the impact of marine debris.
Methods
This project aims at creating an autonomous underwater vehicle (AUV) that utilizes cameras, sensors, a microcontroller, and GPS to categorize, identify, and map underwater trash. Detection of litter will be performed with a deep learning model - optimized to run on a microcontroller - allowing for cheap and power-efficient inference of underwater footage.
Geographic Constraints
Although our AUV cannot be directly tested in the ocean because of the geographic constraints of New Mexico, utilizing software-in-the-loop testing with data from past underwater vehicles while also running the AUV in other local bodies of water will allow us to evaluate our vehicle. In order to assess the accuracy of the neural network and form a comprehensive model, footage from past vehicles will be run through our model using the framework that was implemented in our AUV.
By utilizing a deep learning model with an AUV, the scope of marine trash can be evaluated with greater accuracy and lower costs than present-day methods.
Team Members:
Sponsoring Teacher: NA