PACTF is live! We bring yet another year of great CTF problems and prizes — over $20K this year. Sign up now at https://2018.pactf.com/register/ for a chance at winning a lifetime-free proton mail account that quite literally cannot be bought anywhere else!
0 Comments
I conducted a successful radio and IMU (Inertial Measurement Unit) test today! The drone, with IMU onboard, broadcasted a live stream of data over radio to a controller board which then streamed the data to my mac over serial. No cable connected the drone and computer! IMU (On drone) → Radio Broadcaster (On drone) → Radio receiver (On separate board) → Computer (Over serial) Since all parts were so low level, it had very low latency. Even though I enabled a gaussian filter over time and had built in delays to allow from interrupt signals. Such a low latency might make it possible to do off-board computation on live flights, which would make reinforcement learning (or other on-the-fly learning methods) possible! The drone project is revitalized! Andover has decided to approve and fund the project for this term. I've designed and made the new drone. Its goal was to be easily fixable upon crashing (which it'll doubtlessly do), so it is held together by common 5/40 screws. All parts, except motors and controllers, are either 3d printable or laser cuttable. Schematics and motor tests are on my GitHub (https://github.com/Reichenbachian/) here. Expect much more fun to come!
One of my articles, titled Artificial Intelligence is Applied Calculus, Not Magic was published to the subversionist! They are a pretty big publisher on Medium, and I'm super excited!
ADMIN(Alex, Darcy, Miles, Igor, Nicholas), my team for HackNEHS, won! Our project was called RAaQ(Research aggregation and quantification). It's currently hosted at raq.world. Keep in mind that it is still quite buggy and slow(coded in under 8 hours.)
RAaQ's goal is to automatically express idea correlation and public opinion of an arbitrary idea. We do this by scraping over 100 news sites and using NLP(Natural language processing) to extract meaning from it. This data is then put through a couple more processes including co-occurence matrix to extract meaning. The API can be accessed using raq.world/api?word={any word}. It is an entirely open source project and can be installed using ```pip install https://github.com/milesmcc/raq/archive/master.zip``` The project will continue, and hopefully turn into something a bit prettier and less buggy. Last week came in some pretty great results! I've been training an LRCN deep neural network to discriminate planetary light curves in partnership with the CFA. The new desks are coming along! I only need to add lights, and then I'll call it done! |
AuthorGotta make a thing or two. Archives
April 2018
Categories |