Hackathons

MakeMIT 2018:

Collaborated with Noah Moroze and Ashay Athalye

 

 

 

 

Dot Delivery: a candy cannon adventure

Cannon (completed and working):

  • Re-purposed Schrader valve interface to bike pump for pressurization
  • Large PVC pressure vessel for air storage, rated to 20 PSI (plenty for such a large air volume)
  • Two high-flow brass solenoid valves: one to a small auxiliary tank that allows for repeated firings, and the second to fire the cannon itself.
  • Interchangeable cannon barrels supporting Dots and Peanut M&Ms

Gimbal (completed and working, but not very well):

  • 2-axis stepper-motor aiming system supporting the cannon barrel
  • Laser cut acrylic design facilitating direct pan drive and linear tilt drive

Electronics and software (face tracking worked but the interface between the Nvidia Jetson and the Arduino was not completed – there were UART issues on the Jetson):

  • Nvidia Jetson running OpenCV for face tracking, distance reckoning and active estimation of positional error
  • Arduino with stepper drivers for interfacing with gimbal and solenoid valves

General hackathon notes:

This was by far the most fun I have had at a hackathon so far for a couple reasons:

  • The event was extremely well organized; there were multitudes of resources for all hackers, tons of tools (laser cutters, 3D printers, power and hand tools), working wireless internet, and helpful volunteers.
  • The event started at 8:00am and ended the same day at midnight, which meant that nobody was working under extreme fatigue, and the time pressure kept the project scope more bounded.

 

HackHarvard 2017:

Collaborated with Samantha Miller, Landon Buckland and Aileen Zeng

Bottom-up quantitative social modeling:

  • Model interactions between people and how they form groups.
  • Create probability distributions of human traits and develop insights into how people impact the groups that they are in and vice-versa.
  • Study how a population-density distribution changes human organization.

Future conceptual additions:

  • Add a top-down model with statistical inference to extrapolate human patterns from real-world socio-economic data.
  • Study the relationship between group-forming and governance.
  • Create an interactive data-visualization system.

This is mainly a concept, and is quite unfinished.

General hackathon notes:

Since it was a 36-hour hackathon, it felt extremely long – we biked back to MIT both nights to sleep in our own beds, and left a couple hours early to get rest.

HackMIT 2017:

Collaborated with Samantha Miller, Landon Buckland and Nicholas Freitas

Origami Paper Airplane Genetic Algorithm Optimization.

Summary: Given a paper size and desired relative throwing velocity, our algorithm will create several “species” of planes, and selectively “breed” them to encourage formation of characteristics that perform well (fly the farthest) in a physical model. The project will output a paper folding pattern for an optimal plane which can be intuitively constructed by the end user.

Theoretical description of components:

  • Encoding tool for defining sets of folds mathematically.
  • Interpreting tool to convert fold vectors to points and polygons.
  • Physical parameter modeling tool to calculate the relevant physical properties of the origami manifold for aerodynamic simulation.
  • Aerodynamic and kinematic simulation of origami models to determine flight distance, factoring in drag, vortex lift, standard lift, gravity, and initial velocity conditions. Conditions are checked from all angles to ensure that creative solutions are able to succeed.
  • Genetic algorithm to evaluate breed individual plane “species” with the best performance (determined by simulation), and create successive sets of airplanes with incrementally optimized performance.

Accomplished:

  • Completed genetic selection model for processing sets of plane species and selectively mutating those with proven favorable characteristics.
  • In-depth evaluation of computational modeling of folding patterns.
  • Computational model of 2-D representation of sequential folding of paper.
  • Partial aerodynamic and kinematic modeling of the bi-modal distribution of plane species with vortex and standard lift.

Needs future work/assessment:

  • User interface currently not yet implemented.
  • Kinematic simulation solves for predicted optimum ratios of forces instead of an actual distance.
  • Currently no translation tool between fold vectors and polygons.
  • Currently no interpreter to assess direct aerodynamically relevant data from species parameters.

General hackathon notes:

  • This was as 24 hour hackathon (and my first).
  • It was a fun project idea, but it was over-scoped.