ABOUT
Accelerate her potenital: bolt into action!
As a part of Harvard's Master of Design Engineering program, I had the opportunity to take How to Make (Almost) Anything at MIT. This is a legendary course across Harvard & MIT, and Professor Neil Gershenfeld has taught the class every year since 2004 while directing the Media Lab’s Center for Bits and Atoms, and the class sparked an entire international “Fab lab” movement.
Each week we cover a different fabrication technique on our quest to make (almost) anything: it’s a semester-long crash course on making for those who have never made before.
​
For my final project, I was committed to create a wearable device designed by female athletes, for female athletes.
​
I designed Bolt: accelerometer designed for female athletes that sparks one's ability to crush performance goals, on and off the field.
Bolt is a comprehensive platform that combines intuitive and fun software with cutting-edge hardware. With Bolt, you can seamlessly visualize your acceleration data, gain personalized insights into your workouts, and turn your performance goals into achievements.​​
ROLE
PM
Mechanical Engineer
Software Engineer
Designer
DELIVERABLE
Bolt Accelerometer Sensor
Data Visualization Platform
Embroidered Bolt Top
TIMELINE
September - December 2024
1
Challenge
The Gender Gap in Sports Technology
In the world of sports technology, female athletes often find themselves underserved. As a former collegiate athlete and now marathon runner, I find myself wishing there was more tools designed around the female physiology and biomechanics.
Recognizing this gap, the Bolt Bra project aimed to create a wearable device specifically designed for women that could track performance metrics and provide actionable insights
​
I was committed to develop a comprehensive system that seamlessly integrated hardware and software, was comfortable to wear, and could effectively capture and analyze acceleration data for female athletes.
2024 New York City Marathon
Personal Record 3:32:41
2
Process
From Fabrication to Implenentation
I designed & built the Bolt Bra as a part of the MIT course, How to Make (Almost) Anything. This 16-week course immersed the class in fast-paced engineering modules each week.
For my final project, I designed a three-part system including the Bolt Sensor, the Embroidered Bolt Top, and the Bolt digital platform.
A preview of the design process at MIT
What materials & components were used?
Bolt's materials and components sourced from Adafruit, Amazon, Harvard REEF, Michael’s Crafts, and online platforms, with costs varying per item, such as the Adafruit Feather M0 ($19.95) and the BNO085 IMU ($24.95).
​
The 2D and 3D design processes included additive fabrication for a 3D-printed case with Harvard REEF PLA and subtractive techniques like embroidery for the Bolt Shirt, while software tools like Arduino, Python, and VS Code were used to program and integrate hardware components.
Additionally, the digital platform utilized OpenAI API for processing and Heroku for deployment, costing $5/month for basic usage.
​
Overall personal costs, given my access to Harvard & MIT resrouces:
-
Bolt sensor: $16.49
-
Bolt bra: $30.97
-
Bolt Digital platform: $5.00
-
Bolt box: Free! Harvard Makerspace
For more details, visit Week 15: Product Development page!
Sensor Fabrication and Production
Bolt Sensor ​
The heart and sole of this project! The Bolt sensor integrates the Adafruit Feather M0 Adalogger (SAMD21 microcontroller), BNO085 IMU, MAX30102 pulsometer, a 110 mAh LiPo battery, and a MicroSD card, all housed in a custom 3D-printed bolt case with magnetic connectors and finished with purple metallic paint.
Components were sourced from Adafruit, Amazon, and local resources, with a total cost of approximately $90. Using Arduino software for sensor control and Python code for data analysis, the project successfully collected and evaluated motion and heart rate data, answering questions on system accuracy, real-time data logging, and sensor integration, while identifying areas for improvement in battery life and connection stability.
Microcontroller Selection
As someone who is new to working with electronics, the Adafruit Feather M0 Adalogger was a great fit for my accelerometer. The Adafruit Feather M0 is ideal as an intro sensor and for the Bolt Bra design because of its compact size (2.0" x 0.9") and lightweight build (5.3g), making it perfect for wearable applications without adding bulk or discomfort.
Its SAMD21 microcontroller offers ample processing power, native USB support, and built-in LiPo charging, enabling seamless sensor integration, real-time data logging, and reliable performance—all critical for creating a user-friendly, functional wearable.
Arduino Code
I compiled & uploaded my acceleration code from Arduino. This sensor design was inspired by the Harvard course, the Physics of Sports. By configuring the sensor with ALGMR (acceleration, linear acceleration, angular velocity, magnometer, and rotation) we can deduct a variety of metrics such as acceleration, velocity, or cadence.
Designing the PCB (Printed Circuit Board)
Once I got the code working, it was time to design by custom PCB. WOW! This part of the project intimidated me. I had never even heard of a PCB, but after many YouTube videos, trial / error, and long hours... KiCad, BitRunner, Mods, Roland SRM-20 CNC Machine... I started to love this process.
Connecting the Sensors
All of my sensors used i2c connections, which was intuitive to find the SLA/SCL ports. From not really knowing what the difference between microcontrollers (ESP32, RP2040, SAMD21... I had no idea!) The electronics were challenging but once I understood the workflows, they felt familiar to the design software that I'm most familiar with.
After MANY variations, the Bolt sensor WORKS!
🔵 The blue blinks mean the sensor is being reset
🔴 The LED blinks red while a trial is being recorded
🟢 The LED blinks green when the user stops recording. The number of green blinks corresponds to the specific CSV file
🤓 Once your workout is complete, remove your microSD card, and upload your .csv files to your computer!
Bolt Case Iterations
There were 50 case iterations designed in Fusion 360 and 3D printed on an MP4 Prusa. The case design was one of the most challenging parts of the project.
Requirements:
-
The case needed to enclose the 1”x2” sensor with easy access to the on/off switch, SD port, and micro USB to allow sensor to charge.
-
In next steps, I want to embed a HR sensor in Bolt. To do so, you need skin contact, so the window was essential.
Design Phases:
​
-
Mirrored Cases: You must mirror the cases! Duh. But upon first printing, I just made 2 identical prints of the lightning bolt. Well, well you are placing them on top of each other the must fit!
-
Puzzle fit: I thought playing with one piece extruded +.2 inches and subtracted -.2 inches, that would be the winner. NOPE. I couldn’t make the whether it’s extruded or inverted, one of those components must be smaller than the other... because otherwise they won’t fit seamlessly.
-
Perimeter magnet: OK... Making progress. However, this magnet design took up way too much area within the sensor, so I had to increase the size of the total case.
-
Spot magnets: AHA! This is it. It extruded cylinders on the corners of the bolt case. The .125 inch circular magnetics fit beautifully. This secured bolt as you are on the move!
Embroidery & Laser Cut Packaging
The embroidered top was SO much fun to design. I learned how to use Brother PE900 embroidery machine during Wildcard week, and this is definitely a skill I will hold with me.
HTMAA introduced me to working with textiles, and I really enjoyed adding even a bit of personal flair to my design.
I knew that there should be a custom way to enclose the Bolt Sensor and Bolt Bra, and I thought the wood box would beautifully juxtabose the sensor technology and bra.
This box with hingest is 8x5x3". We use the Epilog Helix and Epilog Fusion Pro in the REEF. The box outlines are .001px.
Data Visualization: Putting your acceleration to the test
I developed a Flask App in VS Code with Python, HTML, and CSS to visualize the user's acceleration data. ​
Many data visualization tools are confusing, and don't explain data clearly. I deployed my Flask App with Heroku to make the data visualization process as simple as possible. Moving forward, I want to embed OpenAI insights into how to adjust your training baased on your data.
3
Solution
How to Make (Almost) Anything has been one of the most impactful expereinces of my academic career. Every Wednesday coming to the Media Lab, I felt challenged, inspired, and impowered. I am proud of this semester's learnings, and I cannot wait to see the incredible work my classmates do next!
