Your Smart Skateboard Companion









Learning to skateboard for the first time is hard. Really hard.

That's why two MIT engineers teamed up to build sk8mate, a smart skateboard built to help you learn how to skate better, mate!




sk8mate detects how badly you wobble, and adjusts the length of the skateboard that best suits your skill level: beginner, medium, advanced.

It's generally easier to balance and maneauver with a longer board. More advanced skaters can skate on smaller boards with comfort ease.

As you improve, sk8mate will challenge you to skate at a harder level when the time is right.



Features




sk8mate detects how badly you wobble, and adjusts the length of the skateboard that best suits your skill level: beginner, medium, advanced.

It's generally easier to balance and maneauver with a longer board. More advanced skaters can skate on smaller boards with comfort and ease.

As you improve, sk8mate will challenge you to skate at a harder level when the time is right.

This project explores an application of physical adaptive tools, tools that help adapt to a users' skill level to learn a motor skill.



Static Mode

You're in control of the board. Switch between easy, medium, and advanced modes with the click of a button, and skate at your own pace.

Adaptive Mode

Let sk8mate adapt to your skill level. With a click of a button, skate around and sk8mate will detect and adjust to your skill level. As you practice and improve, sk8mate will automatically adapt and adjust as well.


We built out custom components that add to the smarts of our adaptive skateboard, including a triple-axis gyroscope, custom-built sliding mechanism, and stepper motor to actuate the system.

Sensor: Gyroscope

A triple-axis gyroscope integrated at the front of the board senses balance and rotation in all 3 dimensions.

Learn more

Build: Sliding Mechanism

Our custom-engineered sliding mechanism allows the board to transform in length.

Learn more

Actuator: Stepper Motor

The stepper motor actuates the length of the board automatically on a rack and pinion setup, with a click of the button.

Learn more



3D Visualization



Implementation Challenges

Challenge Solution
Build a motorized mechanism for extending/retracting the board We cut two pieces of wood and bended acryllic to attach sliding rods to connect two sides of a skateboard. To actuate, we devised a rack and pinion mechanism through many iterations of trial and error. Additionally, we designed a housing to cover the gap between the two halves of the board.
Create skill level settings, and switching between them We decided to stick with 3 levels (beginner, medium, and advanced). Because the stepper motor doesn't keep track of position, we store the current state of the board and program differences in length, backward and forward, to get from any of the three levels to another. It took trial and error to get the right lengths. We then drilled holes and use a pin to lock the length in place.
Develop an algorithm to measuring how well the user is balancing We parse through sensor readings from the gyroscope and wrote an algorithm to detect sudden changes in balance, rotation, and wobbling. One big wobble only shows gradual change in orientation in each time step. Thus, we had to balance between recording enough data, storing a buffer, and measuring large changes in a given orientation over time.
Power the electronics and motor We use 3 different 9V batteries and wired in switches to turn them on and off easily.
With repeated user testing, the glue attaching our motor to the board broke off We tried several types of glue (i.e. epoxy, adhesive, and wood glues), but in the end the only adhesives strong enough were hot glue and gorilla duct tape.

User Study


We tested sk8mate with 3 MIT students with varying degrees of skateboard experience.

Participants attempted to skate 20 feet along a straight line with the static and adaptive setup.



In the static setup, participants skated 20 feet in a straight line 5 times, and adjusted the board length on their own as desired. In the adaptive setup, participants skated 20 feet in a straight line 5 times, starting at the easiest setting and then progressing to the hardest setting.

After each setup, participants skated the same distance at the advanced setting. At the end, participants were asked to fill out a post-study questionnaire to collect qualitative feedback, and compare the static and adaptive conditions.

  • “Adaptive learning feels more personalized and motivated me”

  • “Thought it was very cool”

  • “Falling off, having the board get longer, but then being able to skate on the longer board”


With the first generation of sk8mate, users preferred the static setup slightly over the adaptive one, raitng 3.66 to 3.33 out of 6 (most favorable).

Incorporating our user feedback, we went back to work and reengineered our wobble-detection algorithm and added in an LED to flash the number of wobbles sk8mate detected after a test run.


Source Code



Team


Kevin Shum

MIT '19 Course 6-3

Jessie Wang

MIT '19 Course 6-3