Rethinking Occupancy Sensing for Office Lighting
​
​
Process
I began by observing and talking to employees to understand
how lighting fits into their day. What stood out to me was that
people rarely talked about lighting quality itself, but instead
described the effort required to manage it. For instance,
lights turning off unexpectedly while they were working and having to wave their arms to get the lights back on.
​
This led to a key insight: When people are already operating under high cognitive load, any need to monitor or correct lighting turns it from background infrastructure into a mental burden. People expect lighting to behave as a reliable environmental constant, not an object of attention.
​
The current system relying on PIR (passive infrared) sensing caused friction. It detected motion but often failed during seated work, causing 1-2 false shutoffs per hour in prototype testing​​​​​​​​
​
​
​
​
​
​
​
​
​
​
Needs
I mapped out user needs from what I learned.
Project Goal​
​
Design an office lighting system that adapts to human presence and context. As Arista Networks planned for a new office building, they reconsidered how occupancy-based lighting could work.
​I then moved to the business side. Arista needed an occupancy-based system that was energy efficient and simple to install. They wanted to explore alternative, forward-looking approaches.
Prototyping
I explored how to develop a system within these constraints that supported users’ needs for reliability. I prototyped by attaching sensors to toggle light switches: I mounted an ESP dev board next to an existing wall switch and connected it to a presence sensor and a servo with a mechanical arm. When the sensor detected a change in presence, the servo physically toggled the light switch.
​
Through my exploration, I discovered that mmWave radar sensors, which are successful in other industries but not commonly integrated in lighting solutions, offered a critical advantage. Unlike PIR, mmWave sensors can detect micro-movements such as breathing, allowing them to reliably register presence even when someone is seated and still. Radar sensors eliminated false shutoffs during testing.
​​
​
The system worked remarkably well. Even when I remained still for long periods, the lights stayed on, and they reliably turned off when I left the room.
Design Concept Proposal​​: Interaction logic and experience
​
​​
​
​
​
My proposed system combines daylight sensing, mmWave presence detection, and manual switches to automate lighting. Lights turn on only when someone is present and daylight is low, and turn off when either condition changes. Familiar wall switches always give users simple, reliable control over the system.
In prototype testing, PIR-based sensing produced on average 1–2 false-off events per hour during seated work while the radar-based system produced none. By making lighting reliable background infrastructure, the system can reduce cognitive load.​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
Although radar sensors are more expensive than PIR per module, installation dominates total cost, making precise presence detection a meaningful tradeoff given radar’s estimated >90% reduction in user interventions for false shutoffs during testing.
System Architecture
