Gulf of Mexico Trash Collection Simulation
A multi-agent simulation system for optimizing garbage collection drone operations in the Gulf of Mexico.
This project addresses the real-world problem of marine pollution in the Gulf of Mexico through computational simulation and optimization techniques. Built as part of APSC 200 at Queen's University.
Problem Statement
The Gulf of Mexico faces significant trash accumulation challenges. This project models optimal strategies for deploying garbage collection drones to efficiently clean the affected areas.
Technical Approach
Multi-Agent System
Developed a Multi-Agent System (MAS) that simulates multiple autonomous drones working together to collect garbage. Each agent makes independent decisions while coordinating with others for optimal coverage.
Gulf Stream Simulation
Used satellite imagery data to accurately simulate the Gulf Stream current patterns. This allows the model to predict how garbage flows and accumulates over time through a dynamic density map.
K-Means Clustering
Implemented K-Means clustering to identify optimal drone deployment locations and garbage collection hotspots. This approach led to a 20% efficiency improvement over existing methodology.
Results
The simulation demonstrated significant improvements in collection efficiency by:
- Predicting garbage accumulation patterns
- Optimizing drone patrol routes
- Coordinating multi-agent collection efforts
- Adapting to dynamic environmental conditions
Team
Led a team of 4 engineers through the design, implementation, and testing phases of this semester-long project.