Playful computation
Playful computation
Embedded Files
Encounters with ephemeral surfaces
A material and machine exploration through iterative design
with ABB-IRB120
This research explores robotics as a medium for play, curiosity, and material exploration. Through a series of iterative experiments using a robotic arm to generate and engage with soap bubbles and other minimal surfaces, the work investigates how computation can interact with ephemeral materials in embodied and exploratory ways.
Central to this research is the articulation of interaction modalities between human, robot, and material. Interaction here spans a spectrum: from scripted robotic motion and sensing, to human intervention, environmental influence, and moments of breakdown or recalibration. These modalities include physical interaction (robot– material contact), perceptual interaction (computer vision tracking and feedback), and procedural interaction (pausing, resetting, or reconfiguring the system).
Unexpected behaviors, emergent from misalignments, timing discrepancies, or material fragility, become central to the work. These moments of surprise are opportunities for discovery and engagement.
The thesis space
The thesis space
By selecting and recombining parameters: material behavior, robotic motion, sensing, and human proximity, the experiments explore how fragile phenomena can mediate interaction between humans and machines.
The thesis space explores a set of experimental scenarios shaped by multiple parameters. Rather than testing a single interaction, the work investigates how different combinations of variables afford or constrain interaction between humans, robots, and a fragile material.
Robotic systems, when paired with ephemeral materials such as soap films, can produce and respond bubble formations that serve as geometric probes, targets and interactional cues.
This pairing of a precise system and an unpredictable material can support exploratory, playful motion coordination between robots and humans, revealing robotic performance as a relational act.
MOTION MODES: path planned + externally guided motion (EGM)
path planned bubble making.
path planned bubble move (keepy uppy).
egm bubble poking.
end of the arm/ tools
end of the arm/ tools
A series of interchangeable 3D-printed tools. These include attachments for carrying a small fan to influence airflow and lift or move bubbles, as well as elongated forms to reach and puncture them. Designed in Rhino, each tool is modeled with a defined tool center point (TCP), allowing it to be accurately integrated into both simulation and physical execution.
version 1: joint 6 facing forward
This orientation framed the interaction as a form of attention, positioning the robot as engaged with the user across the workspace.
When deployed within the sensing scenario, this configuration introduced technical constraints. Movements approached kinematic singularities and triggered warnings related to EGM ramp timing.
version 2: joint 6 facing down
By reorienting the arm so that joint 6 faces downward, the robot could reach targets more comfortably within its workspace. The end effectors were redesigned accordingly, maintaining a forward-pointing interaction at the tip while aligning with a more stable arm configuration.
move
move
In this mode, the robot’s motion resembles a child playing “keepy-uppy” with a balloon, repeatedly nudging the bubbles upward and prolonging their time in the air. By introducing airflow into the interaction, the robot is able to extend the bubbles’ lifespan and create a more sustained playful exchange.
poke
poke
In this scenario, a bubble gun is used to create large bubbles, which tend to remain within the robot’s reachable workspace. Using the EGM pipeline, the robot continuously updates its target based on the detected position of the bubble source and attempts to approach and puncture the bubbles.
sensing and egm pipeline dev.
sensing and egm pipeline dev.
This experiment explores a vision-based interaction between human gesture and robotic movement.
A camera-based tracking system allows the robot to respond directly to the movement of a handheld bubble wand, creating an intuitive and playful interaction where motion becomes the interface.
Rather than following a fixed trajectory, the robot continuously adapts its behavior in real time, translating physical gestures into motion and making the relationship between sensing, computation, and robotic action visible to the participant.
initial proof of concept.
simulation with synthetic target for speed optimization.
robot studio simulation for debugging pipeline bottlenecks.
devs' view
The system operates through a real-time sensing and control pipeline. An Intel RealSense camera mounted to the robot cell captures synchronized color and depth data, while a tracking node detects the yellow bubble wand and reconstructs its position in 3D space. This position is transformed into the robot’s coordinate frame and published as a continuously updating target point.
poking motion
EGM ROS pipeline synthesis
Through ROS, MoveIt Servo, and ABB’s Externally Guided Motion (EGM), the robot receives low-latency motion commands and continuously adjusts its trajectory to follow the wand. Together, these components form a closed feedback loop in which sensing, transformation, and control are continuously linked.
The sensing layer was refined in parallel to improve the responsiveness and stability of the interaction.
These refinements transformed the system into a more coherent real-time exchange, enabling the robot to move at approximately 0.15 m/s while remaining responsive to human input. More importantly, this shift changed the character of the interaction itself: the robot’s movement became less predictable and more closely synchronized with the user’s gestures, creating a more immediate and dynamic sense of engagement.
the nature of the engagement between robot, material, and participant varies across scenarios, producing different forms of behavior, interaction, and observation.
SHORT_presentation.mp4
Across these experiments, interaction emerged from a balance of alignment between the robot’s motion, the ephemeral behavior of the material, and the participant’s ability to interpret and engage with the system. Play occurred in moments where these elements synchronized, and dissolved when they did not. In this sense, interaction is framed as contingent and situational.
By bringing together robotics, ephemeral materials, and playful exploration, this thesis contributes to an understanding of how technological systems can be experienced as participants in shared, time-based encounters. More broadly, it suggests that alignment and legibility may be critical considerations in the design of future interactive and personal robotic systems.
Date: May 2026 | Location: No siteType: Thesis Project Committee: Josh Bard, Daragh Byrne, Golan Levin, Jiaying Wei
Google Sites
Report abuse