Shawn Hymel
@shawnhymel
Embedded Systems Educator & Course Developer | #IoT #EdgeAI
The demo balance bot is working! Want to chat about it? Come hang out this Thursday to see how I used #ReinforcementLearning to teach it to balance. 👉 Link to the webinar registration in replies #robotics #AI #embedded #Arduino @digikey.bsky.social
This old lav mic served me well for nearly 10 years. Farewell, friend. Your service will be remembered 🫡
Come hang out next Thursday (Aug 13). I'll be hosting a webinar on using #ReinforcementLearning for #robotics. We'll walk through a simple balance bot demonstration. See you there! Register here: event.on24.com/wcc/r/538413... #AI #MachineLearning #embedded #engineering @digikey.bsky.social
Join me next Thurs for a hands-on webinar on using #ReinforcementLearning for #robotics! I'll walk through the process of simulating a robot and training an #AI agent to balance it. Register here: event.on24.com/wcc/r/538413... #engineering #programming @digikey.bsky.social
More tests. Trying to figure out the real torque of the motors, as they’re being driven below the rated voltage given in the datasheet. #robotics
Doing some tests this morning to measure wheel speed and acceleration on the M5Stack BalaC bot. Since there are no encoders, we have to take a few measurements by hand… #robotics
I ultimately decided to go with #MeshCore and installed a repeater on my roof. There’s a good bit of MeshCore traffic in the Denver area! #LoRa @seeedstudio.com
The first episode of my video series is live! The goal is to build up to a full balance bot that's trained using #ReinforcementLearning. This episode is just about getting the #robot into the MuJoCo simulator. 👇👇👇 Link in replies #AI #robotics #engineering @digikey.bsky.social
Contact is amazing. It gets better every time I watch it. Sadly, that count is only 4 or 5 times, so it’s close :) But…here’s mine 😅
Curriculum learning helps #ReinforcementLearning agents to behave in ways you intend and avoid stumbling on random (unintended) actions that technically accomplish the goal. #AI #MachineLearning #robotics
I’m blown away at how well my new #Prusa handles PETG with no tweaking 😲 #3Dprinting
I had an amazing time at #OpenSauce this past weekend. Here are some of my favorite projects that I saw throughout. I’m thrilled to see the #maker world still going strong 🦾 #electronics #robotics #3Dprinting
#OpenSauce is this weekend! If you’re going, stop by the @digikey.bsky.social booth to chat with Zach, @beckystern.bsky.social, @oddjayy.bsky.social, and me. Ask us questions and show off your latest creations! 🤖 #maker #engineering
Made custom brackets for the Ikea Skadis pegboard to mount the parts bins 😁 #3Dprinting
The @seeedstudio.com SenseCAP P1 #Meshtastic repeater has a single Grove connector in the enclosure, which is perfect for something like a BME680 environment sensor. It’s great for broadcasting temperature, humidity, etc. in your area! #LoRa #electronics @digikey.bsky.social
The balance bot is officially alive 🤖 This project is all about taking #ReinforcementLearning out of simulation and seeing how it behaves in the real world. The sim2real gap is real and really tricky! #Robotics #AI #MachineLearning #Engineering #Education @digikey.bsky.social
In case anyone is wondering: yes, Prusa still ships gummy bears with their printers 🐻
The kind of toy I like finding on my doorstep 😍 #3Dprinter #Prusa @prusa3d.com
Progress! Domain randomization is a powerful technique for training robust #ReinforcementLearning agents. The bot now handles pushes like a champ 🦾 I was also able to get rid of a lot of post-processing (e.g. filters) in my #Arduino code. #robotics #engineering #education #AI #ML
First iteration of my #AI powered balance bot! It took some tweaking (sim-to-real is not trivial), but it shows that #ReinforcementLearning can work for controls. Next, I plan to use some domain randomization to make it steadier and more robust. #robotics #Arduino
Making progress! The RL agent is learning to balance the #robot pretty quickly. Next up: trying to get it to stay in one place. It's fun to watch the learning happen in real time 🤓 #robotics #AI #MachineLearning #ReinforcementLearning #embedded
