Optimizing Search for Lost Objects in Water Bodies

Project Description:
"Optimizing Search for Lost Objects in Water Bodies" is a web-based tool designed to assist in locating objects such as airplanes and ships lost in vast water bodies. This project utilizes Bayesian Search Theory and the A* Search Algorithm to create probability-based search paths, enhancing the efficiency and success rate of search and rescue operations.

Introduction

This project is built to streamline and improve search operations for lost objects in oceans by calculating probability distributions around last-known locations and observed debris. The web app incorporates Bayesian updates and optimal path simulations to suggest effective search paths, integrating real-time environmental data such as ocean currents and wind direction.

Features

• Automated Data Collection: Uses Selenium to gather bathymetric data.
• Dynamic Probability Updates: Bayesian theory allows probabilities to adjust as new data is received.
• Optimal Search Path Simulation: Uses A* and probability distributions to calculate effective search paths.
• Downloadable Search Paths: Enables downloading optimal search routes in CSV format.
• User-Friendly Interface: Simplified UI for inputting data and visualizing search outcomes.

System Workflow

1. User Authentication: Search and rescue personnel log in to access the service.
2. Input Data: Flight data, last-known location, and debris recovery details are entered.
3. Data Processing: Bayesian and A* algorithms create probability distributions and optimal search paths.
4. Simulation: Users view the simulated search path in a graphical interface.
5. Data Download: Optimal paths and probabilities can be downloaded in CSV format for operational use.

Tech Stack

• Backend: Django, SQLite
• Frontend: HTML, CSS, JavaScript, Leaflet JS, Turf JS
• Python Libraries: NumPy, Pandas, Matplotlib, Scikit-Learn
• GUI: Pygame, Pygbag
• Automation: Selenium
• APIs Used:
  • Meteomatics (ocean current and wind data)
  • Geocoding (city location data)

Algorithms Used

• Bayesian Search Theory: Enables continuous probability updates for search areas.
• A Search Algorithm*: Calculates the most efficient search paths based on probability grids.
• Kernel Density Estimation: Applies unsupervised learning to generate reverse drift distributions based on debris data.

Usage

1. Log in to the application.
2. Input flight and debris recovery data.
3. Run the simulation to visualize the optimal search path.
4. Download the search path CSV for use in field operations.

Future Scope

• Expanded Search Area: Broaden the current search grid and allow users to input historical data.
• Collaborative Features: Enable data and strategy sharing among users.
• Advanced Analytics: Add predictive modeling and anomaly detection to enhance search accuracy.
• Commercialization: Deploy the platform as a service for government and private agencies.

Contributors

• Anushka Jadhav
• Atharva Jagtap
• Manan Kher

Acknowledgments

Special thanks to our guides, Prof. Anand Godbole and Mr. Raj Mehta, for their guidance and support.