THE PROBLEM
Many control systems are good at following rules.
Turn on below this temperature. Turn off above that temperature. Run the fan for this many minutes. These rules can be dependable, but the real world contains more variables than a handful of fixed settings can capture.
Building controls research is moving toward systems that can respond to occupant needs, equipment condition and energy conditions. The U.S. Department of Energy describes advanced controls as a way to respond to occupant and grid needs, while NIST is investigating AI approaches for control decisions, fault diagnosis and optimisation.
Intelligence begins with sensing.
Temperature alone is rarely the whole story. Depending on the application, a useful control system might need information about humidity, CO₂, occupancy, weather, energy prices, equipment status or thermal storage.
Once the system can see more of what is happening, software can move from simply executing commands toward choosing between better and worse operating options.