ColiPrev

Logo

Inspiration 🤔

Following the recent incident where an exploded Russian missile spread debris throughout the atmosphere, threatening critical satellites 🛰 and dependent technologies, we decided to create a solution to assist in both predicting and averting collision 🔥 of space debris with satellites.

"There are 34,000 debris objects of 10 centimeters or larger. The 10-centimeter size, which is roughly 4 inches, is important because according to scientists, a collision between an active satellite and an object of this size “will most likely result in catastrophic destruction of the satellite,” according to the U.S. government’s National Orbital Debris Research and Development Plan. On top of these clearly destructive objects, there are over 100 million debris objects at least 1 millimeter in size, which can also cause damage when orbiting at 17,500 miles per hour. The number of low-Earth orbit debris objects has increased by 50% in just the last five years (Yahoo Finance)."

What it does 💪

The program aims to use image recognition/machine learning to track the location of space debris as it orbits around the Earth. Once its orbit is determined, the software aims to determine if any satellites are at risk of collision with it 🔀.

How we built it 👷‍♀️

The software we built over the past 2 days consists of two parts:

An asteroid/space debris ☄️ image recognition 👀/machine learning program built with Python and related packages (torch, detectron2, pandas, numpy, pillow, matplotlib, cv2, tqdm). We used a pre-made image set to train the algorithm, which can also determine the x and y positions of the debris relative to the camera view of the satellite, which can consequently track the approximate position of the debris in the orbit.
A simulation software built with Python 🐍 and VPython for visual simulation to determine the risk of collision 🔥 between two objects when information about their orbit is inputted. The simulation runs the positions of the objects for the coming year and returns whether a significant risk was detected during the time period 📈.

Challenges we ran into 😥

We had difficulties finding images 📸 of space debris to train our algorithm. As a result, we were forced to use an image set of asteroids. Therefore, the image recognition serves more as a proof of concept rather than being specifically applicable to our use case.

Accomplishments that we're proud of 👏

Given that no members on our team have taken multi-variable calculus 🧮, we are incredibly proud that we were able to develop an accurate simulation of the 3-dimensional motion of the objects. This took significant time to both research and realize, but is something that we are undoubtedly proud of ⏰.

What's next for ColiPrev ⏭

If we were to continue developing this technology, we would work on increasing the accuracy of the simulation algorithm as well as work on procuring image sets of space debris taken from satellites in orbit 🛰.