Arvind Padmanabhan

13-Jan-2018

Flavours of Android

Android

Things

for IoT &

embedded devices

What Google Has to Offer

Android Things certified SoMs

Board Support Package (BSP)

SDK

Android Things Console

Android Framework & Tools

Things Support Library API

Google Services

Architecture

Android vs Android Things

  • Flexible access to hardware
  • No system apps (such as Contacts, Calendar, Telephony, Settings ...)
  • Only one app exposed to user: there can be background apps
  • App started automatically
  • Full-screen display: no status bar or navigation buttons
  • Display is optional
  • Peripheral I/O API: GPIO, PWM, I2C, SPI, UART
  • User Driver API: extend framework, inject HW events
  • Cloud IoT Core: easy integration to Google Cloud

Google Play Services

Google Play Store

Firebase, Fit, Location, Maps ...

API

LoWPAN Support

  • Support for Thread protocol
    • Needs IEEE 15.4 radio
    • Direct device-to-device comms
    • IPv6 mesh networking
  • APIs
    • Scan for and join a network
    • Create a new network
  • Tools
    • OpenThread on nRF52840-PDK
    • CLI for dev and debugging: lowpanctl

Supported HW

NXP Pico i.MX7D Block Diagram

System-on-module (SoM)

System-on-chip (SoC)

Hardware Specs

PICO-IMX7-AT $64

PICO-PI-IMX7 $79

Starter kit $169

@ technexion.com

SoC NXP i.MX7D Block Diagram

Does my toaster need  quad-core-GHz processing

with 512 MB RAM and multitasking?

Potential Applications

  1. Where image/video/audio are involved
  2. Where local processing is needed, possibly with AI/ML
  • Home appliances: washing machine, refrigerator, dishwasher ...
  • Smart metering
  • Point-of-Sale (POS) reader
  • Vending machines
  • Interactive ads
  • Access control, camera monitoring, security systems
  • Gateways, routers
  • Asset tracking, predictive maintenance

CES 2018, Las Vegas, 9-12 Jan

Lenovo, LG, JBL, Qualcomm, MediaTek, NXP, Rockchip, iHome, Goertek, and Tymphany

  • Google Assistant: smart speakers
  • Google Cast: smart displays

Demo #1

1. Take picture

2. Store in DB

3. Send image to Vision API

4. Respond with labels

5. Store labels

Resources