Building and Breaking

I've been tracking my training and ultimate frisbee games for some time now, and I recently decided to turn the tracked data into metrics to quantify my running performance. The initial motivation stems from watching a lot (a lot!) more Premier League soccer and starting to learn more about how high level soccer teams track their players to quantify statistics. Some metrics are out of my league, but some are much more attainable. For example, I can implement metrics such as: Total Distance covered, Sprint Distance, ...

Continue Reading

Last I left off, I had way more content that I wanted to embed and search relative to what my laptop could keep in memory, leaving me with no interactive tooling. Where to go from there? Let's start with sqlite-vec

Continue Reading

Building on the learnings from last time, I'm working on improving the download experience. The previous experience was somewhat limited when trying to incorporate new sources. Unfortunately, this efficient downloading has also turned the vector database into something too big to keep in memory on my laptop, so I need to seek out other options for continued use.

Continue Reading

Research notes for exploring celestial navigation

Continue Reading

Building on the learnings from last time, I'm working on speedups to significantly reduce the time to launch the interactive features to make them more, well, interactive. Previously, the script was taking 800 seconds! on startup and that has been improved significantly.

Continue Reading

Research notes for exploring celestial navigation

Continue Reading

I started with exploring celestial navigation as an initial use case for the tools for thought. My primary takeaways were that the database creation speed really limits the interactive nature of the tool and that the user interface could be improved.

Continue Reading

I've always been interested in increasing the number of research papers that I read and using that to speed up how quickly I can learn different knowledge areas. The problem I have is that I'm aware of large sources of papers (e.g. Mastodon, arXiv.org, and toward my specific interests, the robotics and software engineering topics within arXiv), but don't know what I don't know about content that I'm missing. Where I'd like to end up is that I can be working on writing a topic and in parallel tools can be working to help me surface references that can either be immediately cited or extend my learning on the topic in a tight loop with my writing.

Continue Reading

I'm focusing on starting 2025 by getting more actively connected to research in a variety of fields for the year.

Continue Reading

Design Driven Design builds feedback loops that allow for continuous improvements to open source projects during each pull request. The loops are: Design Writing informs Feature Testing which in turn informs the Design, Feature Testing informs Unit Testing which in turn informs the Feature Testing, Unit Testing informs Design Writing which in turn informs the Unit Testing. These loops form seams for collaboration across partners or for self-re-evaluation at mindful points in the process to maximize success.

Continue Reading

A new experiment for FormaK has landed: Superoptimization. Instead of individual peephole optimization, the experiment prototypes using search and a model of the CPU to find the optimal ordering of all operations.

Continue Reading

A new feature for FormaK has landed: Hyperparameter selection. Automatically select optimal parameters based on data from the modeled system.

Continue Reading

I have a sweatshirt that ripped a pocket. It's quite comfy, but I don't wear it much because I always worry about losing something in the pockets. Today I decided to sew up the pocket. I also considered patching it but the whole was small enough that it didn't need a patch.

Continue Reading

FormaK implements a new documentation builder to make documentation much easier to navigate and look better too.

Continue Reading

I have a small backpack that I love and have used for many years. Unfortunately, the backpack strap began to tear away at the shoulder, leaving me with an ever more tenuous hold on the contents. As a "new year, new me" project, I decided to fix the strap.

Continue Reading

Over multiple iterations of improving FormaK (reference IMU model rocket model, the original Python code generation), I've wanted to leverage the power of Sympy to provide efficient implementations of symbolic concepts before converting to Python or C++. The tool for this job is simplify. With one call, it can simplify polynomials, simplify trigonometry and other approaches. Combine this with Common Subexpression Elimination and we have a powerful pair of tools to write efficient code regardless of the model. There's just one problem: Sympy can be incredibly sluggish for some functions. Each call can take 10s of seconds. These 10s of seconds can stack up to minutes of time spent waiting and hoping for a result. For this experiment, I take some time to dive into what's going on and try to understand why it can be so darn slow sometimes.

Continue Reading

A new feature for FormaK has landed: the Strapdown IMU Reference model. The model is now available for inclusion into new models and use as a reference for implementing future models. This post covers some of the aspects of the design and development that didn't make it into the final design and feature.

Continue Reading

A new feature for FormaK has landed: the Strapdown IMU Reference model. The model is now available for inclusion into new models and use as a reference for implementing future models.

Continue Reading

A new feature for FormaK has landed: Innovation filtering. Automatically make Kalman Filters more robust.

Continue Reading

A new feature for FormaK has landed: The FormaK runtime. The first tool in the runtime is a Managed Filter that handles coordinating process and sensor updates.

Continue Reading

A new feature for FormaK has landed: improved model generation by leveraging a subset of the C++ Abstract Syntax Tree

Continue Reading

Launch day is here! Presenting Mast, an open source, privacy first recommender and discovery tool for Mastodon

Continue Reading

Launch day is here! Presenting Mast, an open source, privacy first recommender and discovery tool for Mastodon

Continue Reading

A new feature for FormaK has landed: generating models with calibrated sensors

Continue Reading

On the topic of broken parts, the 3D printed seat support that I'd made has also failed.

Continue Reading

While I had the bike in the shop (for too long!) I decided to acquire some vinyl sticker material to see what I could do about recreating the sticker graphics that have been falling off the bike.

Continue Reading

The second printed pedal finally broke. The two pieces still press fit together nicely, so it wasn't a catastrophic failure by any measure. I took some photos of the part after it happened with the intent to try and understand how it broke and how to make it stronger.

Continue Reading

The first day of the project started with a focus on getting TF-IDF set up and connecting the data to the format that scikit-learn expects. Also, a brief diversion to start with small data before going to the big data approach.

Continue Reading

I'll admit it. I'm a document hoarder. For years I've been collecting and tagging notes in Evernote as I'll read through them to save for later. Can I use that hoarding to learn better?

Continue Reading

It's been a little while since I last wrote about Formak ( FormaK Under The Hood: Optimization for scikit-learn integration , Sat 08 October 2022). Since then, I've been busy adding functionality to FormaK and I've also been improving the tooling for the project, inspired by Boring Python: code quality. The latest piece of functionality is the C++ code generation. The PR is in progress, but the initial experiments have proved out that the generation pipeline is feasible and able to be integrated into bazel.

Continue Reading

This post is a new episode in a miniseries focused on modeling the workings of a CPU to use for generating optimal code. This post focuses on taking a sequence of instructions and simulating their results while respecting the latency of each instruction execution and data dependencies between instructions.

Continue Reading

This post is a new episode in a miniseries focused on SIMD instructions. This second post focuses on fusing common single operations across multiple data

Continue Reading

This post is a new episode in a miniseries focused on SIMD instructions. This first post focuses on matching points in the compute graph where we can combine individual instructions to form SIMD instructions.

Continue Reading

For the new scikit-learn feature, let's dive into how the model gets fit to data.

Continue Reading

FormaK implements CI for new commits as of PR #4 via GitHub Actions

Continue Reading

There's a new feature coming to Formak: integration with sckit-learn. This enables quickly integrating new features such as model selection, pipelines and other data tooling.

Continue Reading

The second feature for FormaK landed: generating Python models

Continue Reading

The first feature for FormaK landed: the user interface!

Continue Reading

Using Python's type checking to check for mismatched physical quantities

Continue Reading

I love Bazel. It's one of those satisfying tools where computer science algorithms and data structures fuse with good software practice to create something that's satisfying on many levels. It knows how to minimally rebuild things. It knows how to hermetically build things in a repeatable way. I also hate Bazel.

Continue Reading

Starting live development of a new open source project

Continue Reading

Becoming a little more familiar with Python's Abstract Base Class functionality

Continue Reading

How do you test a Transformer and if scoping is working?

Continue Reading

Starting testing and immediately finding some bugs

Continue Reading

Progress on a better understanding of Python while generating random code

Continue Reading

Let's automatically construct programs for testing

Continue Reading

Cleaning and repairing an old Simplex Derailleur round 2 getting rid of some rust

Continue Reading

Cleaning and repairing an old Simplex Derailleur round 1: Disassembly and degreasing

Continue Reading

The pedal for my bike broke without warning. Let's take a look

Continue Reading

Sharing a satisfying effort to fix a hat and a glove with a little sewing

Continue Reading

How small can we get with a functional gear part on a FDM printer? What about packing in extra batteries?

Continue Reading

Designing and printing a replacement clip for a backpack

Continue Reading

How small can we get with a functional gear part on a FDM printer?

Continue Reading

Fixing a seat support that cracked (shattered?) in a crash

Continue Reading

Taking notes in Markdown (LaTeX?) and Python

Continue Reading

Blogs, Books and Podcasts (technically listening, I know)

Continue Reading

Two more iterations on the battery cover / side skirts

Continue Reading

Iterating on designs for a more "aero" replacement battery cover and the rear bumper

Continue Reading

Today was focused on quickly iterating to getting a functional rear bumper designed for extra airflow.

Continue Reading

Today was mostly focused on how the rear bumper fits on the body

Continue Reading

The first order of business for the day was printing the new aero front bumper. The end result was somewhat flimsy and didn't print well, but it's a great start. It's probably the least manufacturable of the part designs by traditional methods so far, making it perfect for 3D printing once the print is dialed in.

Continue Reading

The first round of chassis dyno printing is done. During initial testing I noticed that the car did not track straight, so I added guides to some of the rollers so that the car would be encouraged to remain centered on the dyno and not roll off into the sunset when I wasn't looking. It also makes higher speed testing more boring (in a good way) where the car has less of a chance (but still a chance) to jump off the rails and drag around higher voltage cables.

Continue Reading

The parts for the full chassis dyno arrived. After an initial print on the rollers and the bearing housing, it looks like the design is mostly ready.

Continue Reading

Today was mostly a breaking day. The printed half-shaft design melted itself apart after some high voltage testing for a couple of minutes, so it's going to need a redesign.

Continue Reading

Electronics time! Today I got the bench power supply hooked up to the fully running chassis that lives under the '05 Mustang.

Continue Reading

The suspension work continues! The rear design is now printed in a high-shell configuration. This leads to (mostly) nicely aligned filament.

Continue Reading

The suspension work continues! The rear design with short arms plastically deformed after the first flex, so I ended up changing to a more complicated design with longer arms. The new design should also allow for equal length and thickness arms on the front and rear designs.

Continue Reading

Today focused on simplifying the suspension with a printed design. There's nothing fundamentally wrong with the existing XMods suspension, but I do think there's room for improvement. For example, the existing suspension limits its own range of motion and has relatively poor ground clearance. Also, I've taken it upon myself to replace most of the front clip with printed parts now that I've gone off the deep end and printed the uprights.

Continue Reading

Replacing the skid plate for the steering rack

Continue Reading

Photos from Day 15: Printing a new front upright and new axles

Continue Reading

Photos from Day 14: A replacement half-shaft and a broken front upright

Continue Reading

Old Body Style, Next Generation Power and Suspension

Continue Reading

New wheel design! Inspired by the Alfa Romeo 8C

Continue Reading

Pictures from Day 11: Replacing the axles to enable using metal sleeve bearings

Continue Reading

The triumph of the day was proving out the concept for creating custom wheels. The center hub uses a D-shaped drive and is sized so the wheels snap-fit into place.

Continue Reading

Today's focus was on replacing the wheels for the XMods. This has two purposes: I've wanted to play around with printing my own wheel designs and I wanted to experiment with replacing the wheel axles to see if I could reduce slop in the wheels that comes from a loose fit between the axles and the hub.

Continue Reading

Images taken with a new macro lens

Continue Reading

One of the more complicated restorations that I'll need to complete is reviving the older shorter chassis for the Mitsubishi Lancer Evo and the classic Ford Mustang.

Continue Reading

One aspect of reviving the XMods parts that I have involves figuring out what I need to fix to get the shorter chassis running again.

Continue Reading

The wheel mounting design leads to some slop in the location of the wheels. In taking apart the mechanism, it appears that the slop comes from the plastic bushings not locating the driveshaft precisely.

Continue Reading

The latest design has almost all of the features that I'm looking for, with support for some more interesting customization to go beyond the style of the original bumper.

Continue Reading

Before and after pictures, the design is almost a good fit. There are some spacing issues at the wheel well but visually the layout and the headlights are close to the original style.

Continue Reading

The first attempt at curvature got close, but the dimensions didn't line up with the hood as expected.

Continue Reading

With the first prototype, I focused on getting something that would fit the 3 mounting points on the standard body.

Continue Reading

With the first prototype, I focused on getting something that would fit the 3 mounting points on the standard body.

Continue Reading

I recently dug up a box of parts for XMods, 1:28 scale RC cars with customizable chassis and bodies. Along the way I've started to document what I'm working on. The first project is going to be replacing the front bumper of the 2005 Ford Mustang.

Continue Reading

What are cards? How can I automate making a nicely rendered preview for the articles I write?

Continue Reading

In this blog post, I'm going to work on bringing a nicer experience to people who use screen readers by adding ARIA landmarks and roles.

Continue Reading

In reading online, I've always preferred simple, fast-loading websites that are easy to read and don't get in my way from reading their content. In this blog post, I'm going to work on bringing that experience to people who use screen readers.

Continue Reading

I think that data science is a fascinating topic, and I wanted to share some interesting things that I've found in the area.

Continue Reading

Back playing Firewatch. The game is more than just scenic views.

Continue Reading

I decided to go for the full-break method of investigating the electronics, and while it worked out well to start, it's looking iffy going forward.

Continue Reading

My background in electrical engineering involves the first 4 chapters of an electronics textbook I read on my phone when I had internet connection and power in town while hiking the Appalachian Trail. What can I figure out about how the circuit board works for the quadcopter?

Continue Reading

Mechanically, it's in shambles. What's left in the rubble? I dive into the electronics to figure it out.

Continue Reading

I've had the game Firewatch for a while, but I just started playing the game today. What makes the game particularly special to me now is that I spent last summer hiking the Appalachian Trail. Many aspects of the game bring back memories from that experience which I think is kind of amazing. This probably will spoil some of the game for you if you haven't already played. Day 1!

Continue Reading

This has been a back burner project for a while; I'm now finally taking the time to take apart a quadcopter I've had for a while to see what goes into making it tick (and see if I can't mess with it a little bit too)

Continue Reading

For a lot of my life, the robots I worked with were pretty close to what CS wants from them: a stream of sensor data in, nicely organized in a common format across all sensors. To make things happen, I stream some outputs in another selection of formats. The sensor …

Continue Reading