Mix and match ESP SDKs
Cryptic title? Let me explain.
Espressif, as of 18th Feb 2019, has 5 SDKs for the ESP32 SoC on GitHub
| Name | Description | Link |
|---|---|---|
| ESP-IDF | Foundational layer | Link |
| ESP-MDF | Wi-Fi Mesh | Link |
| ESP-ADF | Audio applications | Link |
| ESP-WHO | Face detection and recognition framework | Link |
| ESP-VA-SDK | Alexa, Google Voice Assistant, Google DialogFlow | Link |
(There are some more interesting projects – cloud integrations, DSP functions – but they are not specific to a particular domain. They are applicable to all)
"The most interesting applications happen at the intersection of domains"
Integrating face recognition software with a conversation voice interface was one such application that my colleagues at Espressif did for one of our CES demos this year.
In the above video, ESP32 detects and recognises a face (everything including the training is done offline). After which it starts a conversation, which is driven by Google's Dialogflow in the cloud.
This demo requires the best of the ESP-WHO development framework, and the ESP-VA SDK.
Similarly, we could mix the Mesh development framework, with the ESP-WHO framework to create security cameras for an office complex. Or use the ESP-Mesh framework with the Voice Assistant SDK to create smart home assistants that are truly one with the house by communicating locally (maybe not all your lights need to be connected to the Internet ¯\_(ツ)_/¯).
The possiblities are endless.