Researched by BEAST Library Curator | Verified by Harper | Quality: 8.6/10
The Actuation Gap: In 2026, Sensor Fusion Finally Leaves the Dashboard and Enters the Control Loop
For three years, "sensor fusion" in commercial real estate has meant one thing: a prettier occupancy dashboard. Fuse thermal, PIR, badge, Wi-Fi and desk data, and you get a more trustworthy headcount for space planning and lease decisions. Useful — but it is measurement, not action. The number tells you a floor was 38% utilized last Tuesday. It does not turn down a single air handler.
The interesting shift in mid-2026 is that fused occupancy is moving from the reporting layer into the control loop — feeding real-time demand-controlled ventilation (DCV), the single largest same-year energy lever most office buildings have never pulled. Three things landing right now make that practical, and one discipline decides whether the savings are real. Here is what I would do about it if this were my building.
The lever hiding in your ventilation schedule
ASHRAE Standard 62.1 defines DCV as any method that reduces outdoor airflow to a zone based on occupancy signals. The logic is boring and powerful: you are heating, cooling, and dehumidifying every cubic foot of outdoor air you bring in, and most buildings bring in a fixed volume sized for full occupancy they almost never hit. Match the air to the actual people and you stop conditioning ghosts.
The published range on that lever is wide because building type dominates. Commercial DCV analyses put typical fused-sensor savings at roughly 25–35% of ventilation energy. A frequently cited CO₂-based DCV study found far larger swings by space type — on the order of 53%, 29%, and 40% in library, restaurant, and gymnasium settings respectively — because intermittency, not average occupancy, is what DCV monetizes. Translation for an office portfolio: the win is biggest in your spiky spaces (conference rooms, training floors, cafeterias, hybrid-schedule wings), not the evenly-staffed ones.
| Layer | What it does | 2026 status | FM watch-out |
|---|---|---|---|
| Sense (fused) | Thermal + PIR + desk + CO₂ + badge → one occupancy truth | Cross-vendor fusion now shipping (Butlr × Disruptive Technologies) | Don't buy more modalities than your control logic can consume |
| Bridge (interop) | Move the signal into the BMS | BACnet / ASHRAE 135 is the lingua franca; open APIs closing gaps | Confirm your sensor exposes a BACnet object or webhook, not just a portal |
| Actuate (DCV) | Modulate outdoor-air dampers / VAV per ASHRAE 62.1 | Where the 25–35% lives | Sequence must fail safe to code-minimum ventilation |
| Verify (M&V) | Prove the savings survive an audit | IPMVP / ASHRAE Guideline 14 baseline discipline | No baseline before go-live = no defensible claim |
Enabler 1: Fusion is now cross-vendor, not single-box
The March 12, 2026 partnership between Butlr (thermal occupancy) and Disruptive Technologies (ultra-low-power wireless desk, motion, proximity, tactile, and door/window sensors) is the tell. Their integration puts both vendors' sensors on a single floor plan in Butlr's Studio, each with its own icon — occupancy, temperature, proximity and door data in one operable layer. Butlr's July 2026 update pushed this further: its Connect platform (API tokens, webhooks, email alerts) went live for all customers on July 1, with a Portfolio Explorer and a Claude-powered "Ask Butlr" occupancy analyst arriving in August. Ricoh also took an investment stake.
Why this matters operationally: no single vendor owns every modality your DCV sequence wants. The 2026 model is fusion across vendor boundaries, which means your leverage is no longer "pick the best sensor" — it is "pick the platforms that expose clean, fusible data." Interoperability just became the moat.
Enabler 2: Fusion is moving to the edge — where control latency lives
DCV is a real-time control problem, and cloud round-trips add latency and a data-governance surface. The 2026 research direction is near-sensor fusion — running the fusion model on power-constrained devices at or beside the sensor rather than in a distant cloud. Recent work such as FusionSense (arXiv:2605.22868, tri-stage near-sensor learning for runtime-adaptive multimodal edge intelligence) and FusionBridge (2026 ACM/IEEE Embedded AI & Sensing Systems) targets exactly this: fuse multiple modalities on-device, under tight energy and latency budgets. For an FM, the payoff is twofold — faster, more stable control response, and less occupant data leaving the building (the privacy-first argument that made thermal and PIR fusion popular in the first place, now extended to the compute layer).
Enabler 3: BACnet is the bridge you probably already own
The unglamorous hero is ASHRAE Standard 135 (BACnet) — the interoperability protocol most BMS already speak. Fused occupancy is only a lever if it reaches the damper. Before you spec anything new, the question is whether your occupancy platform can publish a BACnet object (or a webhook your controls integrator can map) — not whether it has a nice dashboard. A dashboard-only sensor is a measurement tool; a BACnet-exposing sensor is a control input.
The catch: an unverified saving is a marketing number
Here is where FMs get burned. DCV savings are notoriously easy to claim and hard to defend, because you are comparing against a counterfactual — the air you would have moved. If you flip DCV on without a measured baseline, you have a vibe, not a verified saving. That is precisely the M&V discipline we covered in our recent piece on non-routine events and baseline integrity: establish an IPMVP/ASHRAE Guideline 14 baseline, meet the CV(RMSE) and NMBE thresholds, and log every non-routine change (a re-stack, a new tenant, a data-center load) that would otherwise let the baseline lie. Fused occupancy actually helps here — a trustworthy people-count is a cleaner independent variable for your DCV savings regression than a fixed schedule ever was.
The APAC angle: pollution makes this an IAQ story, not just an energy one
In APAC — the largest regional market, cited at roughly $5.4B in 2025 and ~42% of global share in one demand-ventilation market analysis — DCV is doubly loaded. High-rise offices across Shanghai, Bangkok, Ho Chi Minh City and Taipei run DCV not only to save energy but to manage indoor air quality against poor outdoor air, so the outdoor-air reduction has to be smart, not blunt. Facilities with advanced ventilation controls have been associated with materially fewer respiratory-illness episodes, which is a tenant-retention argument APAC landlords understand. And in Taiwan specifically, where Taipower's grid pressure and the PUE ≤1.3 hyperscale mandate are squeezing every kilowatt, ventilation energy that responds to real occupancy is exactly the kind of load-shaping the grid now rewards — the same forcing function driving coordinated AI-HVAC control. The broader demand-ventilation-with-multi-sensor-fusion market is projected to roughly double from ~$12.8B (2025) toward ~$26.4B by 2034 on these drivers.
What I'd do in the next 90 days
- Inventory, don't buy. List every occupancy-relevant sensor you already have (badge, Wi-Fi, PIR, desk, thermal, CO₂). You almost certainly own enough modalities to fuse. The 2026 mistake is adding sensors before you've operationalized the ones installed.
- Prove the bridge. Ask each platform one question: "Do you expose a BACnet object or a webhook my controls integrator can map to a DCV sequence?" If the answer is "we have a portal," that's a measurement tool, not a control input.
- Pick one spiky zone. Choose your most intermittent space — a training floor, a conference wing, a cafeteria — and pilot DCV there, not building-wide. Intermittency is where the 25–35% lives.
- Set the baseline first. Establish an IPMVP/Guideline 14 baseline before go-live. No baseline, no defensible number.
- Fail safe. Confirm the sequence reverts to code-minimum ventilation on sensor loss. DCV must never trade air quality for kilowatts.
The dashboard era of sensor fusion measured your building. The 2026 era operates it — but only if the fused signal reaches the damper and the savings survive an audit. Everything else is a very expensive occupancy chart.
Browse more practitioner intelligence in the AISB Library or the sensor-fusion tag.
Have a question about this topic? Ask our CRE AI Agent →