Back to works

Dans le blanc des yeux: telepresence binoculars for urban intimacy

A capstone project that reimagined coin-operated binoculars as a city-scale telepresence instrument. What looked like a distant-view machine became a live, face-to-face encounter between strangers in different parts of Montreal.

My RoleInteraction design and user journey mapping, programming, electronics integration, fabrication support
CourseCART 461, Concordia University, Fall 2024
FormatGroup capstone project, six people
Hardware Budget~CA$1,500 prototype build
StackRaspberry Pi 5, Arduino, Python, OSC, GStreamer, LTE/5G
DeploymentShort daytime demos across Concordia campus
Dans le blanc des yeux project poster

dans le blanc des yeux

The Premise

The project started from a contradiction: public spaces are full of people, yet many social encounters still feel distant. We wanted telepresence as a physical ritual, something that sparks curiosity from afar, then shifts into intimacy up close.

We designed two devices inspired by coin-operated lookout binoculars. Instead of optical lenses, each unit held a screen and sensors. In exploratory mode, one user could physically move their binoculars and control the remote unit elsewhere. Once a second person stepped onto the remote pressure plate, both units locked movement and transitioned into a private one-to-one mode focused on eye-level exchange.

The planned long-term vision was city deployment in contrasted social contexts: nightlife vs quiet neighborhoods, leisure zones vs study zones, and scenic outlooks vs everyday streets.

Close-up of one binocular prototype device

prototype closeup

Concept Development

Before committing to a build we rendered the device into real streets to test whether it read as public infrastructure. A viewfinder people already recognize invites use without a sign explaining it, so the silhouette mattered more than the finish.

Two enclosure directions came out of that. The first exposed the rotation column and its slip-ring stack, making the mechanism part of the object. The second wrapped the column in a fabric skirt, hiding the hardware and reading softer at street level. The exposed column is what we built.

Rendered binocular device with an exposed slip-ring column on a city street

exposed column

Rendered binocular device with a fabric skirt covering the column

fabric skirt

Interaction Logic

Mode 1: Remote Exploration

  1. User A approaches Unit 1 and starts moving it using the handles.
  2. Unit 1 sends orientation + pressure state to Unit 2 over OSC, and Unit 2 mirrors the motion.
  3. User A sees the remote location feed, while local speech can attract people near Unit 2 through outward-facing audio.

Mode 2: Eye-to-Eye Connection

  1. User B steps onto Unit 2 pressure plate.
  2. Movement is disabled to stabilize the encounter on both ends.
  3. The system switches camera/audio routing so both users enter a quiet, intimate two-way exchange inside the viewfinder.

Team

Alexander Terziyski

Server coding for video/audio feed

Lucien Cusson-Fradet

Physical design, electronics, floating build support

Maxime-Alexandre Gosselin

Physical design, electronics, floating build support

Michael Hemingway

Industrial design, fabrication, floating build support

Shayne Larocque

Programming, electronics wiring/setup, 3D printing support

Thomas Rompré

Server coding for video/audio feed

User Journey

I mapped the experience as an emotional arc rather than a feature list, because the whole project rests on a single transition: the moment a curious stranger stops looking at a place and starts looking at a person. Charting that told us where the hardware had to disappear and where it had to announce itself.

The arc runs from confusion at first sight through curiosity, surprise, engagement, and a deliberate exit. Splitting it into two mode-specific journeys exposed the handoff problem in the middle, which is the hardest part of the interaction to get right and the one we kept returning to in the build.

Full user journey map across discovery, initial interaction, connection, social interaction, and conclusion phases

the full arc

Watching mode journey table covering discovery, initial use, exploration, reaction, and watching phases

watching mode

One-on-one mode journey table covering transition, connection, interaction, departure, and reset phases

one-on-one mode

Mechanical Design

Each unit had to pan and tilt under motor control, carry a screen, two cameras, speakers, and a Pi 5, and survive being shoved around by strangers. The team worked the assembly out in CAD before cutting anything, which is what let us find the packaging conflicts on screen instead of at the bandsaw.

Continuous rotation drove the hardest decision. A cable running up a spinning column twists and eventually tears, so the column carries a slip-ring stack that passes power and signal across the joint while the head turns without limit.

Team artifact. The mechanical modelling was led by teammates credited above for industrial design and fabrication; my work on this side was electronics integration, wiring, and 3D-printing support.

Shaded CAD render of the binocular head mounted in its yoke with drive motor and base flange

full assembly

Front orthographic CAD view of the binocular assembly

front orthographic

Wireframe CAD view revealing the internal frame and mounting structure

wireframe through the shell

Wireframe CAD view showing the Raspberry Pi board and speaker packaged inside the head

packaging study

CAD view of the gimbal with bearing flange highlighted, stepper motor, and slip-ring column

gimbal detail

Electronics

The circuit splits across two Arduinos. One reads the pressure plate and the IMU; the other drives the steppers. An enable signal runs the length of the assembly between them, and the upper board relays state to the Pi over serial, which is what turns a footstep on a plate into a mode change on both devices.

Twelve volts feeds the steppers directly and drops through two regulators before it reaches the Pi and the screen, with a separate amplifier stage for the speaker. Splitting the rails that way keeps stepper current off the logic supply, which is what stops a motor kick from resetting the Pi mid-stream.

Team artifact. The circuit was designed collectively and I took part in that work; a teammate drew the schematic sheet.

KiCad schematic of the pressure and kinetic system showing dual Arduino UNO R3 boards, MPU-6050 IMU, stepper drivers, regulators, and speaker amplifier

pressure / kinetic system

System Architecture

The software stack was built around mode switching and resilient synchronization. We used Raspberry Pi 5 units with 5G connectivity, Python services for orchestration, Arduino for motor/sensor control, and UDP-based media pipelines for low-latency exchange.

Runtime stack

  • Python controller process coordinating subsystems on each Pi.
  • OSC over UDP for orientation, pressure, heartbeat, and connection state.
  • GStreamer pipelines for bidirectional video and audio with persistent pipeline switching.
  • Serial bridge between Pi and Arduino for motor + pressure control.

Hardware integration

  • Pi 5 + 5G hat + SIM card per unit for mobile deployment.
  • Internal IR camera for eye-level view, external camera for environment view.
  • Two audio channels: outward attractor + inward intimate playback.
  • Pressure plates around the pedestal to disable motor motion and enter private mode.
Eye tracking and visual processing test for internal camera

visual mode testing

Prototyping Journey

Concept Framing

Early Fall 2024

We framed the project around post-pandemic social disconnection and asked how telepresence could create a meaningful encounter in public space rather than a plain livestream.

Mechanical + Electronic Prototyping

Mid Fall 2024

We built two coin-operated-style binocular shells with motors, pressure plates, cameras, microphones, speakers, Pi 5s, and a 5G setup. Fabrication iterated quickly between wood structure, 3D-printed parts, and wiring revisions.

Realtime Sync Integration

Late Fall 2024

The key challenge was synchronizing motion, audio routing, and camera switching across two remote units while keeping latency low enough for a believable one-to-one interaction.

Exhibition Demo

End of semester

We presented the prototype in exhibition conditions on campus, validating the interaction design with dozens of users and documenting what held up and what still needed a production-grade v2.

Plywood test rig on a workbench holding a Raspberry Pi, LTE antenna, motor, and loose wiring

the plywood mule

Two installed prototype units during exhibition

exhibition prototype

Exhibition Learnings

What worked

  • The concept was immediately legible and surprising to new users.
  • Dozens of participants engaged during exhibitions and recognized the intended social tension between distance and intimacy.
  • Pressure-gated mode switching gave the installation a clear dramaturgy rather than a flat, always-on interaction.

What constrained us

  • Budget pressure as hardware scope expanded (~CA$1,500 total).
  • Power and thermal load on Pi 5 units under concurrent media + sensors.
  • Prototype-level physical robustness, not yet ready for unattended public city deployment.

Reflection and v2 Direction

This project is the clearest representation of how I like to work: concept to system, digital to physical, and art framing to technical execution. It was collaborative, messy, and ambitious in the way capstone work should be.

The next version would move from prototype materiality to durable urban hardware: metal enclosure, hardened mounting, environmental sealing, and maintenance-friendly electronics. The goal is a reliable inter-city public instrument for spontaneous encounters, sturdy enough to live outside a gallery.

If v1 proved the interaction can move people, v2 is about proving it can live in the wild.