This deck is presented during my speaking in Microsoft's //DevCon / Digital Economy Summit, Jakarta, Feb 27, 2020, which was one of a kind event since it was attended by Mr. Satya Nadella (CEO of Microsoft) and Mr. Joko Widodo (President of Indonesia). I shared about how Azure can power SMARTernak - a livestock-farming assistance platform - to help farmers.
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Â
Microsoft Azure-powered IoT & AI Solution To Help Farmer
1.
2. Microsoft Azure-powered
IoT & AI Solution
To Help Farmer
Use Case & Technical Perspective, AI at the Edge / TinyML
Microsoft //DevCon/ Digital Economy Summit 2020
The Ritz Carlton Jakarta
Feb 27, 2020
Andri Yadi
CEO, DycodeX
Microsoft MVP, Azure
3. Andri Yadi
Co-founder & CEO of DycodeX
Vice Chairman, Indonesia IoT Association
Microsoft MVP, Azure
a (at) dycodex.com | andriyadi.com | github.com/andriyadi
Physicist, Developer, Maker, Community Guy, Entrepreneur
About Me
MicrosoQ Most Valuable Professional (MVP) for 12 years
Code for food & passion for 20 years
Break & make electronic stuUs for 22 years
Trying to change the world through entrepreneurship, 15 years now
4. ââI have to visit hundreds of cows every day,
just to check if they still move or notâ
A caretaker in Cattle Breeding Centre - Ministry of Agriculture
Manages 1400+ cows
âHERD TRACKING
âWe often lose counting when a herd of cows
moves from one to another grazing areaâ
A big-size farm, Indonesia, manages 8000+ cows
BEHAVIOUR TRACKING
âWe have no reliable means to track cowâs feeding
behaviour. Non-eating cows are often found dead
after 2-3 daysâ
A medium-size farm, Subang - Indonesia
Manages 400+ cows
5. BASE STATION
5-10 km coverage,
to connect devices
with Cloud.
* Can be NB-IoT, Cellular, or
LoRa/LoRaWAN
WEARABLE DEVICE
Device worn by cattle
contains plethora of sensors
BARN SENSORS
Collection of sensors to
monitor environment,
water & food supply,
and more
FARM MANAGER /
OWNER / INVESTOR
Monitor the farm and manage
operation comfortably from
anywhere in the world
DRONE
Provide surveillance
and help herding.
VIRTUAL FENCE
Contain and move
cattle without physical
posts and wires.
CLOUD
Where the heavy-
lifting happens.
CARETAKER
One farmer/
caretaker can easily
handle hundreds of
cattle, and vast
grazing area
SMART CAMERA
Monitor behaviour of 5
- 10 cattle, count herd,
and more using
AI-powered Machine
Vision.
The Solution
6. * âTernakâ in SMARTernak means âcattleâ in Bahasa Indonesia
AIoT-powered Livestock-Farming Assistance
Platform
6
Thatâs exactlyâŠ
7. On-farm SmaE Camera:
Monitor activity of 5 - 10 cattle -
focus on group activity
Count herd & movement
CaGle-wearable device:
Precisely monitor each
cacle for:
Location: latitude, longitude,
movement speed, direction
Body temperature
Ambient temperature & humidity
Ambient light & sound
Movement: linear acceleration,
angular velocity, direction
Device removal status
Battery voltage & capacity
Monitoring & Insights
Barn Sensors:
Monitor ambient temperature,
humidity, and air quality in barn
Monitor food & water supply
Monitor caGle where-about & well-being
Provide insights - powered by AI - to help
increasing yield and reducing loss
8. Low Power
Wide Area
Network
(LPWA)
High Level Architecture
Gateways /
Base Stations
Things
Internet
Apps
Ingestion
(Network Server)
Infrastructure
Logics
API
Cloud User
Internet
Drone*
SmaE Cam
Wearable
Water & Food
Monitoring
Ambient
Environment
* In development
9. Cattle-wearable Device
Dual IoT Connectivity
Short Range: Bluetooth v5, BLE
Long Range: LoRa/LoRaWAN or NB-IoT,
for 5-10km coverage area
Powerhouse
Ultra low power Microcontroller
with on-device Machine Learning capability
Smart Energy
Battery Gauge
Solar Energy Harvesting
Smart Switching Between
Main & Backup Battery
Fast Charging from USB Type-C
Firmware-optimised
power consumption
Packed with Sensors
GPS / GNSS
Accelerometer, Gyroscope, Compass
Body Temperature
Ambient Temperature & Humidity
Barometric Pressure
Ambient Light
MEMS Microphone
Device Removal Detector
Actuators
Audio Alert via Buzzer
Color LED
Enclosure
Waterproof &
submersible (IP67)
Hardware is 100% designed in-house, mass-
produced in Indonesia, by 100% IndonesianBut SMARTernak is about platform.
Itâs open to 3rd party devices.
11. Letâs dive into a feature: On-device Activity Prediction
Demo Video: http://bit.ly/smrtrnk-ai-3
12. Built-in
IMU Sensor
Possible to transmit
reliably over
LPWA
Predicted activity
(standing, lying-down, feeding, and more)
App
Raw data
5D 1B 16
56 01 41 âŠ
(1080 Bytes)
Processed data
0A 0B 01
(3 Bytes)
{
âcattle_idâ: 2,
âactâ, âstandingâ,âš
âŠ
}
LPWA = Low Power Wide Area Network
Payload length: ~200 - 1600 bytes
Learn & predict cattle activities based on sensor data with
the help of on-device Machine Learning
AI at The Edge
Deep Neural Network
(Run on Microcontroller in device)
Standing
13. Why the hassle?
walking
walking
walking
Type Komodo
Animal activity recognition
No need to visually observe the animal activity.
Focus the acention to troubled animals
Activity stats:
Activity count and duration statistic during cejain
timespan
Activity duration and switching. Can be used for
determining symptoms of possible health issues.
AleE when animal has stopped moving, not feeding,
not enough activity, and more
Itâs all about tailoring Activity Insights & Recommendation
14. Achieving AI at The Edge
And how Azure helps
Azure Machine Learning Pipeline
TensorFlow model Zle
Compress
TFLite
Flat
Bu[er
ConveE
C Byte
ArrayIntegrate
into MCU
Firmware
TinyML Pipeline
Custom
Logics
ML Model
Microcontroller (MCU) Firmware
TensorFlow Lite
for Microcontroller
15. Machine Learning
ClassiZcation
Whatâs happening right now?
Anomaly Detection
Is it ânormalâ?
Forecasting
What will happen in the future?
Some ML problems
For those problems,
ML Inference is possible on
ultra-low power, low cost,
tiny Microcontroller
17. Microso] MVP:
Find MVP or Join the elite group: hcps://mvp.microsoQ.com/
Azure:
Azure Machine Learning: hcps://azure.microsoQ.com/en-in/services/
machine-learning-service/
DycodeX:
SMARTernak: hcps://smajernak.com
ESPectro32 dev board: hcps://shop.makestro.com/product/espectro32-v2/
Other IoT + AI products & solutions: hcps://dycodex.com
TinyML:
TensorFlow Lite for Microcontrollers: hcps://www.tensomlow.org/lite/
microcontrollers
Contact me:
andri@dycodex.com
hcp://github.com/andriyadi | hGps://slideshare.net/andri_yadi/
Call to Action
18. MVP Booth
NOTES:
- Size : 300 x 200 cm
- Power Electricity : 6 ampere / 1300 watt
- TV 43 Inch : 2 units with HDMI connectors
- Power port : 2 Spots @ 4 ports
- Internet : 2mbps provided per booth
- ID Card : 4 pcs
- Others : Storage, Table, 4 chairs
- Collaterals : Flyers, swags, any small
materials are okay. Stand-up
banners are not allowed
Check out some
cool demos at
our booth!
19. PT. DycodeX Teknologi Nusantara
Today, we're pioneering and leading in developing end-to-end home-
grown A"i$cial Intelligence (AI) & Internet of Things (IoT)-based
products & solutions in Indonesia.
Only a few of staEups in the country capable of doing both AI & IoT,
in-house!
As seen on
My Company
Our vision is to solve big problems with technology.
AI + IoT enabler
20. Real World Products & Solutions
Home-grown AI & IoT-powered devices & platforms
EnvironmentAgricultureAsset tracking Energy Design House
AITrash
Smart Trash Management
PowtraX
Electricity metering
Smarterbike:
Bike Fleet Management
(e.g: bike sharing)
DytraX:
Asset Tracking Platform
IoT Prototyping
Platform for
Education, Makers
Custom hardware
design:
already helped 15
companies/startups
End-to-end AI + IoT
development:
firmware, cloud, AI,
web & mobile app
Industrial IoT
Machinery Health
Sensor
Alora
Industrial Environmental
Sensing
SMARTernak
Livestock-farming
assistance
21. DycodeX IoT Prototyping Platform - for Education & Makers
IoT Prototyping Pladorm
Optimized for learning STEM, coding
& electronics, also for fast
prototyping
Hardware:
ESPectro: A low cost ESP8266-based
development board having capable
Microcontroller, built-in WiFi connectivity
ESPectro32: A ESP32-based development
board having highly capable, low-power
dual-core Microcontroller, with built-in
WiFi & Bluetooth connectivity
Extensible with Backpacks: NB-IoT/
Cellular, LoRa, Sensor Kits, Motor Driver,
and more
SoQware:
Arduino SoQware-compatible, enriched
by easy-to-use libraries
Learn to code with:
Visual drag-and-drop blocks*
IoT Cloud: to easily connect IoT devices
NB-IoT + GPS/
GNSS Kit
LoRa/LoRaWAN
+ GNSS Kit
Sensors & Actuators
Kit (20+ Peripherals)
3G + GNSS +
Sensors
* in development
ESPectro32
All electronic boards are 100% designed & assembled in-house
Backpacks
ESPectro
22. AI + IoT enabler
Keep in touch
hi (at) dycodex.com | https://dycodex.com
Bandung, Indonesia