Inspiration
Space has always caught our eye in hackathon competitions. The possibilities are endless in space so why would we want to ruin it. By collecting space junk floating around we can keep the unknown clean and safe for future endeavors.
What it does
The machine learning algorithm detects different kinds of materials floating around space such as plastic, metal, rock, and organic material. After capturing these materials the machine learning categorizes them and sorts them in various bins.
How we built it
The machine learning was built using the Juypter notebook through anaconda. The datasets were found online and imported to be learned and tested. After the machine had reached a percentage of 90% we uploaded that to Arduino. Using the Arduino device we were able to use servos that sorted the materials. We used a machine learning model and trained using a dataset gathered from Kaggle. We also quantum machine learning algorithm called the Quantum Support Vector Machine to determine the distances between two pieces of debris after identifying and classifying the debris. This information will be helpful with trying to optimize the path to take to gather the material. This also uses google cloud to store information and host the server for the real-time analysis of the parts.
Challenges we ran into
The biggest challenge we ran into was importing the datasets into Juypter notebook. After we were able to see all the data training and testing were a breeze. Another issue we ran into was implementing the python code into the Arduino. We also ran into issues with the storage system on Google Cloud.
Accomplishments that we're proud of
We are proud that we were able to train and test a working machine learning code. This was our second try using this software and after learning from our mistakes last time we were able to breeze through this process. This was also our first time playing with quantum algorithms.
What we learned
The biggest thing we learned was how to use Juypter notebook more efficiently. Along with that, we learned a lot about how datasets are made and imported. This was our first time using serial communication with python to send the servo control data about what material type the debris is. We learned how to create quantum algorithms with Qiskit and gained more experience in training optimized machine learning programs.
What's next for CosmicCleaner
The next step for CosmicCleaner would be to implement it into a real-life situation. Making durable hardware that can withstand various types of weather. As well as testing it on earth before throwing it into space.
Log in or sign up for Devpost to join the conversation.