AVS Academy Space Debris Project

Space debris, or "space junk," refers to human-made objects in Earth's orbit that are no longer useful, including derelict spacecraft, rocket stages, and even tiny paint flecks, which pose a serious threat to operational satellites and future space missions. This debris travels at extremely high speeds, increasing the risk of high-speed collisions that can create even more fragments. The proliferation of debris is exacerbated by the increasing number of satellites launched into orbit, raising concerns about the long-term sustainability of space access. 

Project Phases

AVS Academy’s flagship Space Debris Project is designed to progress through six distinct phases, ultimately culminating in the development of an artificial intelligence system capable of predicting spacecraft collisions. This AI will be integrated within a mobile application or website to provide accessible and real-time collision risk assessments.

Phase 1: Foundational Knowledge

In the initial phase, the team is focused on building a strong foundation by learning Python programming and mastering essential mathematical and physics concepts, such as vectors. This groundwork is critical for handling satellite data and performing subsequent calculations.

Phase 2: Satellite Data Visualization

The second phase centers on reading satellite Two-Line Element (TLE) data and visualizing satellite movements in both 2D and 3D. Libraries such as Plotly will be utilized to create dynamic and informative visual representations of satellite trajectories.

Phase 3: Collision Risk Logic and Dataset Creation

During Phase 3, the project will focus on developing logic to calculate the distances between satellites over time. The team will define a threshold for collision risk, which will serve as the basis for constructing a labeled dataset. This dataset is essential for training machine learning models in subsequent phases.

Phase 4: Machine Learning Model Development

In the fourth phase, a sample dataset will be created to train a simple machine learning model using tools like scikit-learn. The model will be designed to predict potential collisions by considering input features such as distance and velocity between satellites.

The final phase involves building a basic mobile application or website, utilizing platforms like Streamlit, to integrate the AI model. This will enable users to access collision predictions in a user-friendly format.

Project Progress and Timeline

Over the summer, the project team successfully learned the basics of Python and began developing an initial script to retrieve TLE data and plot satellites on a two-dimensional map. The team aims to complete this script by mid-September.

Following this, the next objective is to develop a comprehensive dataset containing satellite data and collision risk indicators, which will be used to train the machine learning model. The team targets finishing this dataset by mid-October.

Subsequently, the project will progress to training an actual machine learning model, such as Random Forest or XGBoost, using the curated dataset. The AI system will then be integrated into a website or mobile application that can be published to the public.

The projected timeline for the project’s completion is around Christmas or shortly thereafter, reflecting the team's commitment to delivering a functional and impactful solution.