I worked as an undergraduate researcher at MIT Media Lab Tangible Media Group on the three shape display tables: TRANSFORM, COOPERFORM, and Ars inFORM. The shape displays consist of systems of hundreds of pins that can act individually or in groups to render 3D content physically, allowing guests to interact with digital content in a tangible way. These machines are legendary technology for MIT Media Lab with the inFORM traveling from MIT Media Lab to the Cooper Hewitt Design Museum in New York in 2014-2015 and to the Ars Electronica Festival in Linz Austria in 2016-2019!
I performed novel repair on the shape display tables, reverse-engineering how the tables worked and how to repair and maintain them after almost ten years of use, wear, and tear on each machine. My work restored the machines to their former glory and prepared them for exhibition at MIT Museum in 2023. I created extensive documentation on the mechanical repair and maintenance processes for future use and worked directly with MIT Museum and Tangible Media Group for knowledge transfer. I also worked with my research advisor, Ken Nakagaki, to renew the openFrameworks code to working, updated states. Additionally, we combined the code for each of the three shape displays into one program, OmniForm. Originally each of the three shape displays had different codes for the different hardwares. With this new OmniForm, one code can be used for any of the shape displays. Researchers would simply need to select which shape display they were using at the beginning of the code. Finally, I ideated, designed, and developed new ball juggling and Rubik's Cube Solving functions for the shape displays.
The inFORM shape display was first created around 2013 and has been traveling museums around the world since then. After the many years of continuous use and exhibition, the inFORM sustained significant wear, tear, and damage. The shape display was custom designed and produced with an external artist and had never been taken apart or repaired. I performed the first ever reverse-engineering of the shape display, completing the full disassembly, repair, and reassembly alone only with guidance from my advisor.
Testing the Hardware
I ran sample programs to visually assess performance and locate pins that were not moving, glitching, or otherwise not moving as expected. I also used a debugger mode to see which pins were not responding.
Debugger mode which shows pins that are not responding
A malfunctioning pin
I removed each of the 576 pin caps, disconnected every wire and ethernet cable, and removed each of the eight modules.
Shape display table with all modules removed
Me with some of my removed modules
All eight removed modules
Shape display table with all modules removed
The eight removed modules
I worked with my advisor to diagnose the issues and whether issues for each pin were as a result of a broken board or broken motors. I tested each board and actuating motor. I replaced broken boards. If boards were working but motors were broken, I kept the boards to reuse and save money but desoldered the broken motors and soldered on new motors.
Testing motor functionality and response to sensor input using debugging program
I also worked to refresh the appearance of the shape display table, soaking every pin in isopropyl alcohol. I replaced any bent tubes as any added friction impacted performance. I also replaced any broken pins.
Preparing to soak the pins in isopropyl alcohol
After weeks of repair, I finally reassembled the shape display and tested to be sure that the table was working as expected.
Fully reassembled COOPERFORM shape display table
I performed the first ever reverse-engineering of the shape display, completing the full disassembly, repair, and reassembly alone only with guidance from my advisor. The TRANSFORM hardware repair followed similar repair and maintenance processes to the inFORM hardware repair. I also sourced and installed cabinets for wire and ball storage for the TRANSFORM which matched existing aesthetics.
Removing broken modules
A removed module
Cleaning the shape display with vaccuum
Fully reassembled TRANSFORM shape display table
Documentation of Repair and Maintenance Process
I created this extensive documentation of all parts of my repair and maintenance process for the shape display tables for future knowledge and knowledge transfer. I presented this document in a meeting with Tangible Media Group and MIT Museum and walked them through this document.
MIT Mechanical Engineering Department Social Media
I had the unique opportunity to share my experience with my research on MIT Mechanical Engineering's official social media (@mitmeche on Instagram)!
Rubik's Cube Solving Mode
Inspired by my own personal interest in learning how to solve a Rubik's Cube, I designed and implemented my own Rubik's Cube solving mode with the Ars inFORM shape display table.
I programmed different functions in openFrameworks to hold the Rubik's Cube in place then flip and turn faces of the Rubik's Cube. When the different functions were played in sequence, the table could solve the Rubik's Cube. The future ambition of the project would be that the table would use computer vision to solve the Rubik's Cube entirely on its own without any human intervention.
The shape display table in the middle of turning one of the faces of the Rubik's Cube
Me shooting my video demonstration of my functioning Rubik's Cube solving mode
Full demonstration video of Ars INFORM Shape Display Rubik's Cube solving mode
Example of the openFrameworks code that I wrote for each Rubik's Cube solving function
One of the most iconic modes of the shape display tables is the Escher mode which transports balls across the table by pushing the ball along in circular patterns. Inspired by this mode and wanting to put a unique twist on this ball transportation as well as my own personal interest in learning how to juggle, I designed and implemented my own juggling mode with the COOPERFORM shape display table. This mode required much research and iteration including investigating different ball holding and launching shapes with the pins, different ball types, and adjusting PID controls to get different behaviors and results from the motors.
The future ambition of this project would be that the table would use computer vision to see where the ball is and automatically catch and launch the ball repeatedly in a juggling motion.
In this diagonal shape of the pins, I tried to control the direction that the ball would be launched. Ultimately, it was not really possible to predict or control the path of the ball well.
In this square shape of the pins, I tried to create squares that would hold the balls in place before they were to be launched
Testing mouse control with cupping to hold the ball and pressing mouse to launch the ball
Testing different ball types for best performance (heighest launch height, most predictable behavior, etc.). I tested plastic ball pit balls, heavy balls, and balloons. The ball pit balls (like the red ball in front) performed the best
Testing and adjusting PID control to get the highest launch height possible
The final openFrameworks code that I wrote for the juggling mode. This created a cup shape that would hold the ball. The user could move the cup shape around the table with the mouse. When the mouse was clicked, the ball would be launched into the air.
508-404-3168 | email@example.com | Pittsburgh, PA | https://www.linkedin.com/in/laurenelizabethplatt/