Summary of SYNTHETIC SENSORS, All-In-One Smart Home Sensor
Synthetic Sensors are low-cost, wall-powered general-purpose sensors that monitor entire rooms by tracking ambient environmental data and using machine learning to recognize events. Developed by Gierad Laput at Carnegie Mellon, they virtualize raw sensor inputs into actionable feeds, enable scalable second-order sensing for counts and notifications (like restocking), and can trigger other IoT applications without instrumenting individual appliances.
Parts used in the Synthetic Sensors Project:
- Wall-powered general-purpose sensor device
- Ambient environmental sensors (unspecified types)
- Machine learning models
- Information architecture/software to virtualize raw sensor data
- Second-order sensing components for counts and notifications
- Connectivity components for IoT integration
- Deployment fixtures for room installation
In the era of Internet of Things, we wanted most of our home appliances to become smart. But currently, smart devices may cost much more than their offline counterparts and they often do not communicate with each other. Trying to overcome these limitations, A Ph.D student invented a way to turn entire rooms into smart with a single low-cost device called “Synthetic Sensors“.
Gierad Laput, is a Ph.D. student of computer-human interaction at Carnegie Mellon University. His research program explores novel sensing technologies for mobile and wearable computing, smart environments, and the Internet of Things.
Synthetic Sensor is a general purpose sensor that is powered directly from a wall socket and tracks ambient environmental data to monitor an entire room. It removes the need to attach additional hardware to each of home appliances.
We explore the notion of general-purpose sensing, wherein a single, highly capable sensor can indirectly monitor a large context, without direct instrumentation of objects. Further, through what we call Synthetic Sensors, we can virtualize raw sensor data into actionable feeds, whilst simultaneously mitigating immediate privacy issues. We use a series of structured, formative studies to inform the development of new sensor hardware and accompanying information architecture. We deployed our system across many months and environments, the results of which show the versatility, accuracy and potential of this approach.
The device uses machine learning to recognize the events that happen in the room, like recognizing a particular sound pattern as taking a paper towel, but it cannot monitor when the roll may need to be changed. However, by using a “second order” sensors, the devices can capture counts and send notifications of the need to replenish. This capability can be scaled to an unlimited degree giving consumers highly specific and applicable feedback.
Developers can use the recognized events as triggers for other IoT applications. For example, one could use “left faucet on” to activate a room’s left paper towel dispenser and automatically schedule a restock when its supply runs low.
Read more: SYNTHETIC SENSORS, All-In-One Smart Home Sensor
- What is a Synthetic Sensor?
A general-purpose, wall-powered sensor that tracks ambient environmental data to monitor an entire room. - Who developed Synthetic Sensors?
They were developed by Gierad Laput, a Ph.D. student at Carnegie Mellon University. - How do Synthetic Sensors recognize events?
They use machine learning to recognize event patterns from ambient sensor data. - Can Synthetic Sensors monitor individual appliance states directly?
No, they monitor room-level events indirectly without attaching hardware to each appliance. - What are second-order sensors in this context?
They are sensors that capture counts and send notifications, such as indicating when supplies need replenishing. - Can developers use Synthetic Sensor events with other IoT apps?
Yes, recognized events can be used as triggers for other IoT applications. - Does Synthetic Sensors address privacy concerns?
Yes, the approach virtualizes raw sensor data into actionable feeds while mitigating immediate privacy issues. - Was the system tested in real environments?
Yes, it was deployed across many months and environments to evaluate versatility and accuracy.
