Projects I may someday do

These are projects that I've been excited about at some point. As with everything on this website, they're more a menu of possibilities than a to-do list. I've provided a rough categorization below. This list is always incomplete, and never completed.

Primarily microcontroller- or similar-based

• An automated blinds opener device, so that my blinds will be opened automatically in the morning. I have a small Raspberry Pi board and a stepper motor, and I've built the circuit to connect them, and can control the motor.

  What remains to be done? I need to 3D model a container for all of it, including an arm that can attach to the blinds rod. I try to avoid internet connections in my home devices where possible, except for computers, so I won't have this connected to the internet. It will therefore have a problem with its internal timer drifting over time. My friend suggested powering it with a mechanical outlet timer that turns on at a certain time each day, and then it can count upward from 0, the time it turned on, until it's time to open the blinds. Later it will close the blinds, and eventually the mechanical timer will turn it off. The next day the cycle will reset. According to this scheme, the Pi board's internal time will drift just the negligible amount it does over the course of a day, rather than accumulating minutes of error over the course of a year. (I haven't looked into exactly how much error it'd accumulate over a year; it's likely somewhere between the order of seconds or the order of an hour.)

• Inspired by the space weather scientist Vincent Ledvina, I'd like to write scripts that can alert a user about when to look for auroras, based on substorm prediction using GOES magnetometer data. If that goes well, I may add coronal mass ejection impact detection as well using various other data sources described in Ledvina's guide to aurora chasing e-book.

  Why is this project in the microcontroller section of this list? Because I think it'd be really cool to use an AdaFruit Feather board, which has wi-fi connectivity and a screen, and a 3D printed shell that looks something like a Tamagotchi, to create a little handheld thing that can hold all of this!

• AI sky camera (possibly webcam) that can classify things it sees in the sky.

  My office is on a surprisingly high floor with surprisingly good windows, given that I'm a grad student—we normally rate cinderblock-enclosed basements that could double as a bomb shelter. Last summer, after I learned about crown flashes, I found myself looking at the tops of cumulus clouds without success, wishing I had a way to keep watch all the time and maybe even alert me when one was spotted. And why stop at crown flashes? I'd love to see rainbows, halos, and all sorts of other things—at this point my list includes lightning, aurora, meteor, bolide, undulatus clouds, and cloud shadows, among others.

  I have a Pi board and the Pi's AI camera, and am hoping to figure out how to train the AI camera to classify things it may never see from where I mount it. Is this a fool's errand? If it is, I'll at least learn a lot about automated image recognition. If I'm able to successfully train it, though, I'm hoping to also have it trigger more specialized cameras in different situations:

  • Wide-angle camera: triggered when a rainbow's spotted. Rainbows have a larger angular size than any of the Pi's other cameras can see in one field of view.
  • Either begin taking video or use an actual high-speed camera: triggered when a thunderstorm is seen in the distance. This is in the hopes of catching upper-atmospheric lightning, which happens very fast.

Primarily software-based

• Install HaloRay, GPU-accelerated ray tracing software for simulating halos that I struggled to install about a year ago for reasons I don't remember.
• Continue creating my sun position calculator scripts. Right now, it's giving results, but they're incorrect, so I need to debug it somewhere. There's plenty of software out there that calculates the sun's positions in the sky from anywhere in the world at any time (within a ~6000 year span, as most of these resources are based on a 1991 book by Jean Meeus called Astronomical Algorithms, whose algorithms are accurate in that time period). But I want to write software that does this for myself. I'll then ask it a variety of questions, including
  • What path does the sun take through the sky throughout the year at the north pole?—does it spiral around, sinking lower and lower as we get further from the summer solstice, until on the fall equinox it "rolls around" the horizon before later sinking completely below it?
  • Why do equatorial regions have two latest annual sunsets while regions further from the equator have just one? Is the equatorial biannual periodicity still present, but swamped, in the further-from-equator places, or is it an effect that actually vanishes further from the equator?
  • Is there anywhere in the world where the day length changes rapidly enough that the sun actually rises + sets + rises, or sets + rises + sets, three times in one 24-hour period? The further from the equator you are, the more rapidly the day's length changes in the vicinity of the equinoxes: where I live, it changes by nearly three minutes at its fastest; where my good friend lives in Europe, it changes by roughly thirty seconds faster. At McMurdo Station in Antarctica, it changes by up to an hour! but the daytime geometry there may not make this effect occur there. The sun takes roughly four minutes to fully set, if I remember correctly, so this triple daily sunrise/set may be possible. But I'd like to know for sure.
  • If I vary certain of Earth's orbital or rotational parameters, such as its orbital eccentricity or its axial tilt, how does that affect the sun's path through the sky? This should aid my understanding of how those parameters affect what I see of its path.
  • Animate this maximum solar elevation plot to show how it changes with latitude. I think I know how it works, I just think it'd be neat to see :)
  • Animate the solar analemma's shape and orientation for different latitudes, at a fixed time of day, and varying that time of day. Also animate at fixed latitudes and times for: varying Earth axial tilt; varying Earth orbital eccentricity. I know those are the two parameters that cause its shape, but I want to better understand why!
• I already have a pi-hole working on my home network. I'd like to get it working when I'm using a VPN, too. This means I get to learn some more about (computer) networking... which will be good for me, in a Calvin and Hobbes building character sort of a way
• Between December 2024 and the end of May 2025, it seemed every coronal mass ejection was launched outside the "earth strike zone," the central 30º of the sun's disk that maximizes the likelihood the material will impact the earth and create stronger auroras. I'd like to histogram solar flare + CME data from the Space Weather Prediction Center to see if that apparent effect was real.
• Learn CSS well enough to customize the look of this website into something I really like, and perhaps also to simplify some of the HTML coding I've been doing to make it

Primarily educational

• Make a website (here it is!)
• Record a handful of gopro-type videos of me biking around my city's streets, including annotations describing what I'm looking for or paying attention to as I go. I live in a city that's considered to have particularly dangerous drivers—though that hasn't really been my experience as a biker—and a lot of people I know tell me they're too afraid of biking here to do so other than in the limited (though growing!) portions with protected bike lanes. So I'd like to show what it's really like biking around, and provide an educational resource for what to pay attention to and how to navigate various kinds of situations a road biker will encounter, coming from someone very comfortable biking here and who is almost never in an uncomfortable situation.
  • I also have strong opinions on the right of way hierarchy between trucks, cars, mopeds/ e-bikes, bikes, pedestrians, and slow or disabled pedestrians, how to navigate that hierarchy with respect and consideration, and in which situations that hierarchy can shift. So at some point I'd like to make a compilation video showing different situations where each of these considerations comes into play.
  • There's no need to wait for the video to share that info, so a summary: in most situations in which traffic laws are being followed, trucks give right of way to cars, which give right of way to mopeds and e-bikes, then bikes, then pedestrians, then slow or disabled pedestrians.
  • In my personal opinion, there are locations where some can break traffic rules and retain right of way, as long as they don't abuse it. These locations are largely busy pedestrian areas in cities. In these locations—Newbury Street and Harvard Square and others in Boston, Times Square in NYC, etc—cars are the guests and frankly, the city should be designed in a way that keeps away all car traffic except for emergency vehicles, public transit, and transportation for the disabled. But until that city redesign, cars are the guests in that space, and pedestrians (to some extent) can jaywalk as long as they're not acting recklessly, such as stepping into the street without looking.
  • Bike lanes, even on sidewalks (as long as there's also a pedestrian sidewalk of sufficient size + condition), count as "the street." This means pedestrians walking into a bike lane without looking are acting as recklessly as if they walked into the street without looking, and pedestrians walking in a bike lane are acting as recklessly as if they were walking in the street. But benefit of the doubt should be given to pedestrians, since bike lane demarcation is relatively new in many parts of the US, so bike lane identification's a skill most people must learn through real-world experience!
  • Respecting right of way with respect and consideration means clear communication, and actions backing that up (slowing down a lot and at a distance that's comfortable for the slower or more easily damaged party in the situation), to make one's intentions clear. Extra deference must be shown when "invading" space meant for the slower/ more easily damaged party: that means if I'm forced to bike on a sidewalk, I'll either walk my bike or slow down to literal walking speed and not get too close to pedestrians (from their perspective, which is a larger distance than bikers usually recognize! I've noticed my perspective flip when I'm the pedestrian vs the biker in this situation), and settle into patience; if I pass someone and they don't know I'm there, I should (legally must; am still building the habit to) ring my bell (respectfully rather than aggressively), and pass them slowly. Respect + consideration also means giving benefit of the doubt when the slower/ more easily damaged party makes a confusing or dangerous decision. The greater share of responsibility is on the party that can more easily damage the other: this means bikers must give room for the pedestrian to make a mistake or act in an unexpected manner.
  • Finally: the hierarchy can shift if someone breaks a traffic rule. On my commute, I regulary go the wrong way down a particular one-way street that sees infrequent traffic. When I do so, I understand I'm sacrificing my right of way as a biker, and I stay alert to anyone entering the road. The moment I see one—even if they're still far away—I immediately pull over and partially or fully dismount, and make eye contact and signal they should go by. It's an uncomfortable situation for them, and they have right of way, so I make it clear I'm ceding that to them and they can proceed with minimal worry.
  • The rules of thumb are: in normal situations, precedence goes to the slower and more easily damaged actor; clear communication and consideration for the other's perspective is a must; right of way can be ceded but at the cost of taking on full responsibility for successful communication + consideration.
• Learn animation well enough to contribute to the library on the webpage for Five Hundred Seven Mechanical Movements, a classic reference book from 1908 listing and briefly describing 507 different mechanisms. I found it invaluable when trying to figure out how to design my automatic blinds opener device, which contains a rotating arm I'd like to be able to position along a circular path. (Someday, when I complete this project, I'll make a post describing it in more detail and with images!)

Primarily physics-related

• There's software–I know of HaloSim and HaloRay–that uses ray tracing to draw halos for various kinds of ice crystals in the upper atmosphere. This is essentially a sort of Monte Carlo solution of the analytic equations describing the halos emerging from a given ice crystal shape and orientation. With care and a sufficient background in geometry and ray optics, one can derive equations describing the halos created by those crystals, especially after deciding the path (the bounces) a light ray will take inside the crystal—certain very rare halos may involve as many as four internal bounces!

  But I wonder: could one derive equations that trace backward from the halo to the ice crystal(s) and orientation(s) and bounce path(s) light must take to create it? The answer is yes; with little thought, I could write down extremely abstract functional equations right now. A better question is: could I derive tractable equations to solve for? The answer may be no, but the harder I try, the better I'll understand halo formation from ice crystals.

  UPDATE: there's some literature on this!—specifically, analytic solutions for the halos ("caustics," here) generated by a given ice crystal shape and orientation. An early instance is Walter Tape's 1980 paper Analytic foundations of halo theory, and its references + clicking on "cited by" after searching for it on Google Scholar will yield much of the existing literature on ice crystal halo shapes.

  I would expect that the reverse problem would be difficult for a number of reasons. But I hope to try it out for myself sometime, at a minimum to better understand all of this.

Primarily outdoors

• Learn to throw a boomerang

Primarily physical objects

• Make a wall display for my Pokemon cards. I'd prefer it to be about as obtrusive as the art I already have on my walls—so, not very obtrustive—and to make it easy for me to switch out the cards on display. And to have minimal cost, of course; I am still a grad student :) I'm thinking something based on velvet. I'd like one type of display that will allow cards to partially overlap so just the art rather than also the text is visible, and another type of display that will show the entire card so I can display my full art cards. I think it'd be nice to be able to make a display with lots of cards with e.g. tropical waters to display in the winter, or with sunsets and reds maybe in the summer, or soft lines and pastels when that's more my mood—that sort of thing.

Crochet projects

• Various topological crochet projects (I particularly like Fig 1c here!)
• Protein crochet, such as is loosely described here