This is our team's solution for OrionHack, a space themed hackathon. Awarded Second Place. 🥈
Submitted by Raymond Wong, Jimin Hong and Yuhao Li.
Over 27,000 pieces of mostly human made objects in the atmosphere. If collisions were to occur between objects, this could cascade into a chain reaction further generating more space debris, hence increases likelihood of further collisions. If such event occurs, this could cause catastrophic failure and limit areas of future space launches.
Project aims to focus on using machine learning to visualise and mitigate risks of collisions of two bodies. This can range from newly launched satellites to debris in orbit. LSTM machine learning model has been utilised to predict likelihood of such risks, trained on over 12,000 collision events with Kelvins - European Space Agency's dataset.
The code visualisation uses live data from CELESTRAK, consisting of 1700 debris and approximately 100 satellites (shown as the blue points). Risks of collisions shown through colour visualisations, such that white denotes lower risks, whereas red denotes higher associated risks.
 Raymond Wong |  Jimin Hong |  Yuhao Li |
All source code is licensed under the MIT license.