Measuring thermal radiation and quantifying the effect that it has on people is hard because it requires measuring the temperatures of all the surfaces in the environment and the geometry of the space. Our technology has only recently become feasible as the underlying components have improved. This has been compounded by a systemic underestimation of the importance of the radiative environment in both industry and academia. The universal adoption of air-conditioning helped to mask the impact of ignoring thermal radiation, however, it is a brute force approach that has led to excessive energy consumption and poor thermal comfort.
Smart thermostats are a significant improvement over basic setpoint thermostats and are easier to use than programmable thermostats, however, they still fail to measure half of thermal comfort. Additionally, basic motion detection is not sufficient to control HVAC systems. Whereas a conventional smart thermostat may measure a few data points per sensor, the SMART sensor takes tens of thousands of measurements across a variety of data types. This level of granularity allows much more sophisticated control algorithms to be used.
Human-in-the-loop controls integrate human feedback into HVAC control algorithms and help mitigate poor thermal comfort, however, they require frequent input from the user. They are fundamentally limited because occupants only make adjustments when they notice that they feel uncomfortable. Therefore, these controls cannot fully address thermal comfort complaints. Additionally, they fail to account for the spatial variation of thermal conditions leading to frequent disagreements over setpoints.
The SMART sensor can leapfrog human-in-the-loop controls by using human-as-the-loop algorithms. Rather than requiring human input, the SMART sensor directly measures occupants’ thermal comfort by tracking the surface temperature of their skin and adjusts the environmental conditions before they become uncomfortable. The SMART sensor is also able to find the optimal environmental conditions for a group of individuals by accounting for both personal and spatial variations.
Conventional occupancy detection sensors all rely on proxies for occupancy. PIR and ultrasonic sensors use motion as the proxy, leading to frustrating situations like your lights turning off because you’re not moving enough to trigger the sensor. Other sensors use CO2 as the proxy for occupancy. Unfortunately CO2 sensors are notoriously difficult to use because their accuracy is poor, they drift over time, require manual calibration and have dependencies on other environmental factors such as air velocity, humidity, air temperature and the temperature of the sensor itself. The advantage of CO2 systems is that they can theoretically estimate the number of occupants rather than giving binary result, however, this leads to problematic feedback loops. For example, a sensor detects a high CO2 level suggesting a large number of occupants, it tells the HVAC system to increase the ventilation levels. This lowers the CO2 levels without changing the number of occupants.
The net result is that existing occupancy detection sensors cannot be used to control HVAC systems because they’re not accurate enough and, according to ARPA-E, “thermal inertia would result in extended discomfort. More seriously, such failures could set ventilation to an inappropriately low setting and result in CO2 or other volatile organic compound (VOC) increases that could impact productivity, comfort and potentially hearth – all invisibly without notice to the occupant.”
The SMART sensor is the first method of directly sensing people rather than relying on proxies. The combination of geometric, thermal and temporal data allows for drastically more accurate occupant detection, counting and tracking.
The energy consumption statistics in the “Why Does It Matter?” section are from the U.S. Energy Information Administration (EIA) and Intergovernmental Panel on Climate Change (IPCC). The SMART sensor’s energy savings are based on numbers from the Advanced Research Projects Agency-Energy (ARPA-E) and our models.