Me and the team decided early on to gamble on an aggressive strategy that consisted in turning the cubes upside down, so that the server couldn't see them. To do this, we had to sacrifice the servo motor, which meant that we couldn't properly grab cubes for easier transportation. This unfortunately led to the 3rd place. Here is the official Instagram post.

Enjoy some auto html layout of these pictures :)

I never actually used this, but it was a fun project. It uses a pair of secret and public key to generate the password for any account. The secret key is the keyring password, while the public key is the name of the application of the account. The program spews out a different password for every key pair with the help of hashing.

While the Python version worked fine, it was very inconvenient to use. That's why I made a web version and later also the browser plugin version (I didn't bother putting it on the web store though).

Since there is not much explaining to do, here are some pictures:

Actually I still have to take and upload the pictures, so have a smiley face instead :D

The plan is to have a submarine controlled from my pc on land via an RF24 module for arduino. I made a Python script that captures keyboard or joystick events and sends them to an arduino nano via USB that in turn transmits via RF24 to the submarine antenna, where a second arduino translates the packets and controls the submarine.

Like any other RC submarine, I can't just rely on wireless communication, since 2.4GhZ waves barely penetrate water. My plan is to tether the sub to a floating antenna that will get all the signals, while allowing the submarine all the wanted mobility.

Besides the basic movement functions I also want to add an FPV camera, so I can see live where it's going on my computer screen.

For now the project is stuck on the ballast design, but I think that eventually it will be done.

It all started back in 2018, when I made a rudimentary Python script that connected to a broker's API and bought and sold crypto based on a questionable strategy found online in about 5 minutes. As expected, it did not perform well.

After some time I revisited the idea, but this time I wanted to implement it using AI. So like any good programmer would do (sarcasm), I just jumped into programming without even planning. I tried deep q-learning and KNN too, but soon I realised that it was a bit more complicated than I thought.

At this point the repo was (and still is) a total mess. This is why I deemed it beyond salvation and started a new one. This new repo was made from the documentation up. I first studied the concepts of KNN approaches and then started coding. I still ended up changing half of the documentation, but it good anyway I think.

I won't explain how it works here since It's all explained in the README file in GitHub.