Millions of smart devices around the world are built on AWS IoT Greengrass. From doorbell cameras to washing machines to industrial HVAC units, Greengrass is deployed at a massive scale – the quiet hero of IoT.

Let’s elaborate a bit on two types of embedded device solutions based on AWS IoT Greengrass:

Wearable devices for first responders: First responders, relief workers, firefighters, and the like are often deployed en masse to dangerous environments. In many cases, the workers themselves are dispersed and not aware of the larger picture. Wearable sensors, cameras, and panic buttons reduce the time needed to assist a worker in distress, as well as preserve evidence and direct resources to where they can do the most good. Running inexpensive single-board computers dramatically reduces the time to market for such devices.

The base station they report into can be rapidly developed using AWS IoT Core for LoRaWAN – an AWS-managed service based on Semtech’s LoRa Basics Station and network servers that run inside AWS IoT Core. Customers can build gateways using commodity hardware (see this tutorial for doing so on a Raspberry Pi). In addition, the AWS Partner Device Catalog has dozens of listings for prebuilt and certified hardware based on this service.

Home monitors for energy efficiency: Energy providers want to instrument customer’s energy usage patterns and educate them on ways they could lower their consumption. To do this, an inexpensive single-board computer hosting the AWS IoT Greengrass agent is attached to the breaker panel in the customer’s home. It has small external clamps on each circuit, and the circuit names (kitchen, laundry, living room, and so on), along with appliance model information, are posted to the central management portal by the technician doing the installation. Current draw readings are aggregated with any incoming streams from IoT-capable appliances and sent back in daily batches over an inexpensive cellular NB-IoT connection. All of this data is used to train an ML model in AWS SageMaker in the cloud, and over time, the model develops an ability to make recommendations directly to users in their homes via Alexa-capable speakers or the Alexa mobile application. For instance, it could alert the user that they have left the garage light on past 9 P.M., something that normally does not occur. Or, based on the model’s understanding of a vast array of appliance models, it could suggest a specific model of refrigerator based on how much money it will save the customer on their bill versus the cost of replacement.