LUMION
A purchase-prediction system that visualizes audience excitement through biometric sensing
Overview
Wristband-style light-emitting prototypeLUMION is a purchase-prediction system that embeds sensors into penlights and other live-event merchandise and visualizes, in real time, customers’ latent purchase intent and the movement of audience emotion from biometric signals such as pulse. It is the core product of Cocoreach Inc., and everything is developed in house, from the sensor devices to signal processing, wireless, and the cloud.
Founding and finding the problem
Having received so much support as an engineer in my junior and senior high school years, I had long felt that this time I wanted to be on the side that supports people who build. Meeting our current CEO through JETRO J-StarX @ Stanford / d.school became the turning point, and I co-founded Cocoreach Inc..
After founding the company, we interviewed people across live entertainment, in event operation, staging, and merchandise production, and a common problem emerged. Nobody could say in numbers how excited an audience had been, or how much a given initiative influenced customers’ willingness to buy; none of it had been quantified. In practice, venues rely on customer surveys and on organizers’ gut-based subjective judgment, and no objective metric exists for measuring customer response quantitatively. If emotion itself could be quantified, decisions about staging, merchandise, and promotion could all rest on data. LUMION is the product we built to solve this.
Concept for measuring emotion and engagement from biometric responses and applying it at in-person eventsThe system we developed
A LUMION prototype lighting upThe devices are wristband-type and merchandise-embedded biometric sensors, both designed in house. They capture heart rate and electrodermal activity; a proprietary filter removes motion artifacts (noise), and emotion is then estimated with the arousal-valence model. The aggregation side is built in the cloud, visualizing venue data in real time around Pub/Sub and BigQuery. We reconfigure the setup to fit each engagement.
Layout of the wristband prototype, with electrodes, a vibration motor, and a wireless microcontroller inside 
The underside, with an optical pulse-measurement board and skin-conductance electrodes
LUMION (front) 
LUMION (back). The part glowing green is the pulse sensorRunning actual field trials revealed a chokepoint in an unexpected place: wireless communication. In a venue where thousands of people are packed together, sending and receiving data from every device at once simply does not work in the 2.4GHz band, where it interferes with WiFi. We therefore designed our own communication scheme around LPWA in the 920MHz band, premised on operating at the scale of tens of thousands of units.
Captured data plotted onto the emotion circumplex 
Real-time monitor of pulse waves and emotion estimation 
A self-built ESP32 real-time monitor integrating posture, pulse waves, and LED controlField trials and expansion
The wristband device used in the trial at NanananasaiWe ran the first field trial at Nanananasai, the festival hosted by 100BANCH, which had accepted us into its program. In a venue of around 3,000 people, the data came through stably. At the same time, problems specific to real environments emerged, including shifted wearing positions, radio congestion, and the workload of operating the trial, and we have folded those lessons into the filter implementation and wireless improvements.
From July 2025, accepted into JETRO Beyond Japan Zero to X, we worked in Los Angeles to validate the North American live entertainment market and build relationships with local companies and investors. In 2026 we also presented the results and prototypes at a workshop at IEEE RoboSoft 2026, an international conference.
We plan to scale the trials up step by step, until visualizing excitement at live events is the standard.