Occupancy Sensor Technologies Compared: PIR vs. Thermal vs. Radar vs. Optical AI

1. PIR (Passive Infrared) Sensors

PIR sensors are the simplest and most established occupancy technology. They detect changes in infrared radiation caused by movement — essentially, they sense when a warm body moves within their detection range (typically up to 6 meters from a standard ceiling height).

Strengths: Low cost, low energy consumption, easy installation, and works in all lighting conditions. PIR sensors are the workhorse behind most automated lighting systems, with compact wireless models offering battery life of up to 10 years and minimal IT infrastructure requirements.

Limitations: The fundamental constraint of PIR is that it detects motion, not presence. An employee sitting still at a desk for an extended period can trigger a false “vacant” signal — the familiar experience of waving your arms to turn the lights back on. PIR sensors cannot count occupants, distinguish between adjacent zones, or provide the granularity needed for space utilization analytics. For organizations that need real-time occupancy intelligence, PIR is insufficient.

Best suited for: Phone booths, lighting control, basic room occupancy status (occupied/vacant), and environments where budget is the primary constraint.

2. Thermal Sensors

Thermal sensors detect body heat signatures to identify occupants. Unlike PIR, thermal sensors can detect stationary people (a seated, motionless person still radiates heat) and can approximate headcounts in a defined area.

Strengths: Strong privacy positioning (no images, only heat patterns), light-independent operation, ability to detect stationary occupants, and suitability for privacy-sensitive environments like healthcare facilities. Wireless models offer multi-year battery life and on-device processing.

Limitations: Significantly lower spatial resolution than optical AI sensors. Accuracy degrades in warm environments or when people cluster closely. Cannot distinguish between a person sitting and a recently vacated warm chair. Limited ability to define granular zones-of-interest within a single sensor’s field of view. Cannot match the desk-level precision or multi-zone analytics that optical AI delivers.

Best suited for: Privacy-first deployments, healthcare environments, basic room occupancy counting, and organizations where camera-adjacent technology faces strong employee resistance.

3. Radar Sensors

Radar sensors emit radio waves and analyze the reflections to detect and track people’s movement in three-dimensional space. This technology counts people entering and exiting spaces, providing cumulative occupancy numbers without capturing any visual data.

Strengths: Complete privacy compliance (no images or visual data of any kind), unaffected by lighting conditions or sunlight, proven accuracy for entry/exit people counting, and cost-effective coverage for doorways and large spaces. Radar works through some obstructions and is unaffected by reflective surfaces.

Limitations: No visibility into how space is used once people are inside — radar excels at counting but cannot provide desk-level or zone-level utilization insights within an open area. Accuracy depends on cumulative in/out tracking at entry points, meaning small counting errors can compound over time. Not suitable for desk-level monitoring, meeting room behavior analytics, or understanding how space is actually being used — capabilities that optical AI sensors like PointGrab’s CogniPoint provide natively.

Best suited for: Building and floor entry points, real-time room availability, large spaces where traffic volume matters more than spatial detail, and deployments where complete visual privacy is non-negotiable.

4. Optical AI Sensors

The newest and most capable generation of occupancy sensors combines optical components (cameras or stereoscopic lenses) with on-device AI to deliver real-time, granular occupancy data. These sensors process visual data at the edge (on the sensor itself) and output only anonymous occupancy metadata. No identifiable images are stored or transmitted.

Strengths: Highest accuracy for real-time headcounts (95–99%), desk-level and area-level granularity from a single device, ability to define multiple zones-of-interest within one sensor’s coverage, true presence detection (not dependent on motion), and future-proof capabilities. Optical AI sensors combine people counting with spatial utilization analytics in a single device — something no other technology family can match. PointGrab’s CogniPoint is the leading example: a single edge AI sensor that delivers presence detection, accurate people counting, and deep occupancy behavior analytics simultaneously, with all processing done on-device for complete privacy compliance.

Limitations: Higher per-unit cost than PIR, though the total cost of ownership is often lower because fewer devices are needed to cover the same area with richer data. Requires professional installation for accurate zone calibration.

Best suited for: Enterprise deployments requiring desk monitoring, meeting room analytics, open area utilization, collaboration zone insights, people counting, and integration with booking/BMS/analytics platforms.

5. Wi-Fi & BLE Occupancy Sensors

Wi-Fi and Bluetooth Low Energy (BLE) sensors infer occupancy from the presence of connected devices — laptops, smartphones, and wireless peripherals that passively emit probe requests or advertisement packets. Rather than detecting bodies directly, they detect associated devices and use device counts as a proxy for people counts.

Strengths: No new hardware is required if the building already has a dense Wi-Fi network — the access points themselves can log device association events. BLE beacons are inexpensive to deploy. Useful for floor-level and zone-level footfall estimates, particularly in large campuses where people consistently carry smartphones.

Limitations: Significant accuracy gap versus body-detection technologies. Device counts ≠ people counts: one person may carry three devices, while a meeting room with five people sharing a single laptop registers as one device. “Device parked” scenarios — employees who leave laptops at desks — create persistent false-occupied readings. Wi-Fi sensing also raises GDPR/privacy concerns in several jurisdictions because MAC addresses can constitute personal data even when anonymized. Accuracy is typically ±20–30% at the room level.

Best for: Building-level or campus-level footfall estimates; complementary data source alongside more accurate sensors; organizations with existing high-density Wi-Fi infrastructure who want a no-hardware-cost starting point before investing in dedicated sensors.

6. CO₂-Based Occupancy Sensing

CO₂ sensors measure carbon dioxide concentration in the air. Because humans exhale CO₂, rising concentration levels in a space indicate occupant presence. When CO₂ drops toward ambient outdoor levels (approximately 400 ppm), the space is inferred to be vacant or lightly occupied.

Strengths: Dual-purpose devices — the same sensor serves both occupancy inference and air quality / ventilation optimization. Completely anonymous (no body, device, or image data). Low cost per sensor. Increasingly mandated by building codes in many markets for fresh-air compliance, making the occupancy benefit effectively free.

Limitations: High latency — CO₂ concentrations take several minutes to respond to changes in occupancy, making CO₂ useless for real-time applications like live desk booking or instant HVAC response. Cannot count people, only infer a broad “occupied / lightly occupied / vacant” gradient. Highly sensitive to ventilation rate; a well-ventilated room with 20 people can read the same as a poorly ventilated room with 4 people, producing misleading data. Not suitable for desk-level monitoring.

Best for: Ventilation and air quality management; rough occupancy indication for conference rooms and open areas where precise counts are not required; complementary layer in multi-sensor deployments focused on both people analytics and indoor air quality.

Side-by-Side Comparison

How to Choose: Matching Technology to Use Case

The right technology depends on what questions you need to answer:

• “Is the room occupied?” — PIR or thermal sensors handle this well at low cost.
• “How many people entered this floor today?” — Radar sensors at entry/exit points are the cost-effective choice.
• “Which desks are being used, and how are meeting rooms actually occupied?” — Optical AI sensors are the only technology that answers this with the required granularity.
• “We need data but employees are concerned about cameras.” — Thermal sensors, radar sensors, or optical AI sensors with edge processing (no images transmitted) all address this, with optical AI sensors offering the richest data.

Many organizations deploy a mix of technologies: PIR for simple spaces, radar at entry points, and optical AI sensors for high-value areas. However, this multi-vendor approach adds complexity in procurement, installation, integration, and ongoing management.

This is where PointGrab’s CogniPoint stands apart. As the only optical AI sensor that natively combines presence detection, accurate people counting, and deep occupancy behavior analytics in a single edge-processing device, CogniPoint eliminates the need to deploy and manage multiple sensor types. One device covers desks, meeting rooms, open areas, and collaboration zones — delivering the richest occupancy data available while maintaining complete privacy compliance through on-device AI processing. For enterprise organizations managing large portfolios, this single-device versatility translates directly into lower total cost of ownership, simpler IT integration, and faster time to actionable insights.

Comparing Occupancy Sensor Hardware for Office Buildings

When evaluating occupancy sensor hardware for an office deployment, four dimensions matter most: detection accuracy, coverage per unit, installation requirements, and total cost of ownership.

PIR sensor hardware is the lowest-cost option — small, low-power devices that mount easily on ceilings or walls. However, they require high sensor density (typically one per desk or small zone) to achieve reliable coverage across a whole floor, which drives up installation costs in large-footprint offices.

Thermal sensor hardware offers a middle-ground option — better accuracy than PIR for people-counting, with no visual data captured. Coverage per unit is limited compared to optical AI, and thermal sensors cannot deliver zone-level or desk-level granularity in open-plan environments.

Radar sensor hardware is best suited for entry/exit doorways and corridors. It delivers accurate directional people-counting with zero visual data, but cannot differentiate between zones within a large space.

Optical AI sensor hardware — like the PointGrab CogniPoint — covers the largest area per device (up to 3,000 sq ft) and delivers desk-level, zone-level, and room-level occupancy data simultaneously. Because fewer devices are required to cover a given area, the higher per-unit cost is often offset by reduced installation and cabling complexity. For enterprise offices prioritising granular data and software integration flexibility, optical AI sensor hardware delivers the best ROI at scale.

When comparing workplace sensors, the right choice depends on your primary use case: simple room-level presence detection, accurate people-counting, or granular desk-and-zone analytics. Most enterprise deployments use optical AI sensors as the primary hardware, supplemented by radar at building entrances for overall footfall.

Real-Time Occupancy Sensors: Which Technologies Deliver Live Data?

Not all occupancy sensors deliver real-time data. Understanding the difference between real-time and sampled/batch data is critical when choosing a sensor platform for dynamic office environments.

PIR sensors typically report presence events (motion detected / no motion) within seconds, making them effectively real-time for simple on/off occupancy — but they cannot update a headcount in real time because they do not count people.

Thermal sensors report occupancy estimates at configurable intervals — commonly every 30 seconds to 2 minutes. They are near-real-time for most use cases but lag slightly during rapid occupancy changes.

Radar sensors deliver real-time entry/exit counts as each person crosses a threshold, making them excellent for live people-counting at doorways. Within a room, however, radar cannot report zone-level changes in real time.

Optical AI sensors deliver the most complete real-time occupancy picture: desk-level status, room headcount, and zone analytics are all updated continuously as people move through the space. PointGrab’s CogniPoint processes data at the edge and pushes live updates via MQTT, enabling building management systems and workplace dashboards to reflect actual occupancy with latency measured in seconds — not minutes.

For use cases that demand real-time occupancy data — dynamic space booking, live wayfinding, responsive HVAC control, or safety compliance — optical AI sensors are the only technology that delivers the required speed and granularity simultaneously.

Privacy & GDPR Compliance by Technology

Privacy is frequently the deciding factor in occupancy sensor selection, particularly for organisations operating under GDPR (EU), PDPA (Singapore, Thailand), or equivalent data protection frameworks. The technologies sit on a spectrum from zero personal data risk to significant compliance overhead:

Note: “Optical AI” as used here refers specifically to systems that process imagery entirely on-device (edge AI) and transmit only anonymised occupancy counts — not raw video streams. Systems that transmit raw video to the cloud or perform facial identification carry substantially higher compliance risk.

Frequently Asked Questions

What are the main types of occupancy sensor technologies?

The four main technology families are passive infrared (PIR), thermal imaging, radar, and optical AI. Each uses a different detection method and offers different capabilities in terms of accuracy, granularity, privacy, and cost.

What is a passive infrared (PIR) sensor?

PIR sensors detect changes in infrared radiation caused by movement. They are cost-effective and energy-efficient for binary presence detection in smaller spaces like phone booths and individual desks, but cannot count people or detect stationary occupants.

How do radar occupancy sensors work?

Radar sensors emit radio waves and detect people by analyzing the reflections. They can accurately count people entering and exiting spaces without capturing any visual data, making them completely privacy-compliant. However, radar cannot provide the desk-level or zone-level insights that optical AI sensors deliver.

What is the difference between optical AI sensors and thermal sensors?

Optical AI sensors use cameras with on-device AI processing to deliver desk-level granularity, real-time headcounts, and zone-of-interest analytics. Thermal sensors detect body heat patterns for approximate occupancy counts. Optical AI offers richer data but at a higher price point, while thermal sensors provide strong privacy positioning at lower cost.

Which sensor type is best for large open offices?

Optical AI sensors are ideal for large open offices because they can define multiple zones-of-interest within a single sensor’s coverage area, providing desk-level and area-level utilization data simultaneously. For simpler entry/exit counting of open floors, radar sensors at doorways are a cost-effective alternative.

Are there privacy concerns with different sensor types?

All four technology families can be deployed in privacy-compliant ways. PIR and radar sensors capture no visual data at all. Thermal sensors detect only heat patterns. Optical AI sensors process visual data on-device and output only anonymous metadata — no identifiable images are stored or transmitted. Look for edge-processing capabilities and data privacy certifications when evaluating any solution.

What kind of sensor is used to monitor large spaces?

For large open spaces, optical AI sensors are the most capable option — a single unit can cover up to 3,000 sq ft and define multiple zones of interest, delivering desk-level and area-level data simultaneously. Radar sensors at room entrances are a cost-effective alternative for simple headcount in large rooms. For very large atriums or warehouses where exact positioning is not required, a combination of radar entry counters and PIR zone sensors is common.

Who makes the best office occupancy sensors?

The best office occupancy sensor depends on your specific requirements. For enterprise-grade accuracy with desk-level granularity and complete software integration flexibility, PointGrab’s CogniPoint optical AI sensors are a leading choice — delivering exact headcounts, sub-meter positioning, and a software-agnostic data layer that feeds any building management platform via MQTT or REST APIs. For simpler binary presence detection on a tight budget, PIR-based solutions are widely available. The right answer depends on what workplace questions you need to answer and what integrations your building management system requires.

What is the best occupancy sensor hardware for office buildings?

The best occupancy sensor hardware for office buildings depends on the level of granularity you need. For desk-level and zone-level analytics in open-plan offices, optical AI sensors such as the PointGrab CogniPoint are the leading choice — each unit covers up to 3,000 sq ft and delivers exact headcounts, sub-meter positioning, and real-time data without storing any images. For simpler room-level presence detection, PIR or thermal sensors offer a cost-effective alternative. When comparing office occupancy sensor hardware, evaluate coverage per unit, integration compatibility (MQTT/REST APIs), and total cost including installation and ongoing subscriptions.

What are real-time occupancy sensors and which technologies support them?

Real-time occupancy sensors update occupancy data continuously as people move through a space, with latency measured in seconds rather than minutes. Optical AI sensors are the most capable real-time occupancy sensors — they detect desk-level, zone-level, and room-level changes instantly and push live data via MQTT to connected platforms. Radar sensors provide real-time people-counting at doorways. PIR sensors report presence events in near-real-time but cannot provide headcounts. Thermal sensors typically update on a 30-second to 2-minute interval. For use cases requiring real-time data — live space booking, responsive HVAC, or safety compliance — optical AI sensors are the recommended choice.

What is the difference between PIR and radar occupancy sensors?

PIR sensors detect changes in infrared radiation caused by movement, meaning they require a person to move to register presence. Radar sensors emit radio waves and analyze reflections to detect people — including those who are stationary. The key practical difference: PIR cannot reliably detect a seated, motionless employee, while radar can. Radar is also significantly better at accurate people counting (±5–10%) compared to PIR’s binary presence signal. The trade-off is cost: radar sensors are typically 3–5× more expensive than PIR per unit.

Which occupancy sensor technology is most accurate?

For desk-level accuracy and multi-zone analytics, optical AI sensors are the most accurate technology available — capable of distinguishing individual workstations, counting room headcount to ±1–2 people, and tracking movement paths within a defined field of view. For entry/exit people counting, radar sensors typically achieve ±5% accuracy. For simple room-level presence, thermal sensors are reliable for detecting stationary occupants. PIR, Wi-Fi, and CO₂ are the least accurate options for precise occupancy data.

Are occupancy sensors GDPR compliant?

Most occupancy sensor technologies are GDPR compliant by design. PIR, radar, thermal, and CO₂ sensors collect no personal data — they measure physical signals (motion, heat, radio reflections, air chemistry) without capturing any information that could identify an individual. Optical AI sensors using on-device (edge) processing that transmit only anonymized counts are also broadly compliant. The exception is Wi-Fi and BLE occupancy sensing: MAC address data can constitute personal data under GDPR Article 4, and organizations using this approach should seek legal advice and may need to register the processing activity with their data protection authority.

Which occupancy sensor technology is best for desk-level monitoring?

Optical AI sensors are the only technology that delivers reliable desk-level occupancy data at scale. They can monitor dozens of individual workstations within a single sensor’s field of view using on-device AI to distinguish occupied versus vacant desks in real time. Radar sensors can detect presence in defined zones but typically lack the spatial resolution to distinguish adjacent desks reliably. PIR sensors require one sensor per desk to achieve desk-level data, making large-scale desk monitoring prohibitively expensive. Thermal sensors and CO₂ sensors cannot reliably differentiate individual desks.

Can I deploy more than one occupancy sensor technology in the same building?

Yes — and many organizations deliberately mix technologies to balance cost and capability. A common hybrid approach: PIR sensors for phone booths and small focus rooms (low cost, adequate for simple presence), radar sensors at floor entry/exit points (accurate people counting), and optical AI sensors in large open-plan areas and key meeting rooms (desk-level analytics and real-time booking integration). The critical requirement for a mixed deployment is a software-agnostic data layer that normalizes output from all sensor types into a unified analytics platform — without this, a multi-vendor deployment creates data silos rather than coherent occupancy intelligence.

Related Articles

• The Complete Guide to Occupancy Sensors for Offices
• Wired vs. Wireless Occupancy Sensors: Deployment Guide
• Thermal vs. Optical Sensors: Which Technology Fits Your Office?
• The Complete Guide to Workplace Sensing Infrastructure
• Wi-Fi Tracking vs Dedicated Sensors: What Workplace Teams Need to Know
• Meeting Room Utilization: How the WASH Index Exposes Hidden Space Waste
• Thread Mesh Networking for Commercial Buildings: Why Enterprise IoT Is Moving Beyond Wi-Fi and Zigbee