<aside> 🔴 Summary: Final Project for Intangible Interaction & ML for PCOMP, with Viola He.
</aside>
An image summarising our collaboration:
Context:
Viola found a collection of old medical reports from when they were hospitalized, which reminded us of topics we often discuss in relations to body and trauma. In a PET CT scan report, for example, my body, 19 year-old, female, is broken down into categories, described in abundance details with long jargons. Disease, hospitalization and treatment not only transformed and alienated me from my body, but also made me think about objectification of sick bodies as I was constantly displayed, examined, looked at and dehumanized.
Selected scans of Viola's medical reports
As a patient, the dehumanization is experienced through both unfamiliar procedures and unreadable reports, due to the barrier of languages and understanding of the systems at work. With these contexts in mind, we want to design a system to confront the absurdity and uncomfortable feelings of body alienation that comes from being examined in medical facilities.
Designing the system & interaction
The project will consist of a water tank, camera, one small screen (ipad?), pvc tubes and water pumps. By default, the tank is filled with clear liquid, with the screen half immersed in water. The screen will play a looped footage with caption and no sound. When a viewer approaches, they will be picked up by the camera, which triggers the pump to drop one single drop of ink into the water. Over time, the liquid will become significantly darker and blocks off half of the screen.
System Diagram
The tank is designed to look like an exquisite corpse, with all the running components exposed for the viewer to see. Ideally, the viewer will be intrigued by the intricate pvc tubes and screen. As they get closer to the tank, they will start spending more time looking at the content shown on screen, which will be looped with caption on and no sound. We hope this content is visually interesting to keep the viewers stay long enough to trigger a drop of ink.
How we envisioned the piece will be displayed in space
Water Pumps & Tubes – What pumps to use?
Playing with water pumps
POSENET – How to measure attention?
The question we encountered is how to measure and capture sustained attention? We initially wanted to use Time of Flight sensor but it's hard to differentiate between viewers passing by vs viewer staying and actively engaging. Viola worked with a range of different methods, including Face API and CV, but POSENET yielded the best result.
Working with POSENET, it is designed so that when the viewers' eyes are detected, the LED will lit up.
How to squeeze ink?
One of the critical moments of this experience is when the viewer's presence triggers a drop of ink. We initially experimented with robot arms and different squeezing bottle before 3D printing a motor mount and motor arm in order to have better control over the squeeze.