smart cozy homekit

17
Making Machines More Human Smart Cozy Homekit

Upload: jyun-yan-cheng

Post on 21-Jan-2017

252 views

Category:

Engineering


0 download

TRANSCRIPT

Page 1: Smart cozy homekit

Making Machines More Human

Smart Cozy Homekit

Page 2: Smart cozy homekit

What is IoT?

We enable machine integrate data from internal to external sources and make a wise decision.

Page 3: Smart cozy homekit

Motivations

● We all feel differently under the same weather. Weather index (i.e., apparent temperature) won’t be able to tell us what temparatue/cloth suit you the most.

● In this project, we use Exosite IoT platform and machine learning algorithm to create an intelligent system, which provides suggestions of how to wear clothes and adjust indoor temperature based on your history data(data insights).

Page 4: Smart cozy homekit

System Features

● Our system can open/close winodws based on either weather index or history data.

● Provide suggestions base on user’s history data and display on IHD(In Home Display).

● Our system learn from user’s history data and provide suggestions on provide how to wear clothes

● self-learning (combine open data, sensor data, and machine learning)

Page 5: Smart cozy homekit

Requirement Devices

● Raspberry Pi 2 Model B● Touchscreen for RPi● WiFi USB dongle (optional)● Bluetooth USB dongle● Bluetooth Motor module● THL(temperature, Humidiy, Light) sensor

○ TI SensorTag

Page 6: Smart cozy homekit

Software Architecture● Device Controller ● Learning Engine● Open Data Extractor● Information Provider(UI)● Distribute software architecture

○ Each module instance can execute on different machines, and they can exchange data via Exosite cloud platform --- oneP.

Page 7: Smart cozy homekit

Software Architecture

Learning Engine

Open Data Extractor

Device Controller

Information Provider

(UI)

OnePtraining data

Page 8: Smart cozy homekit

Learning Engine● Input: Temperature(indoor & outdoor), Relative Humidity(indoor & outdoor),

and Wind speed(indoor & outdoor)● Prediction Output: Indoor (for window control) & Outdoor Comfortable index

(for clothes suggestion)(0 ~ 5, cold to hot) ● Prediction period: 1/min● Learing Period: 1/10mins● Use Naive Bayes classifier

Page 9: Smart cozy homekit

Device Controller● Deliver indoor data to the cloud(temperature, relative humidity, etc)● Control the window open/close according to indoor comfortable index and

pm2.5 index.

Page 10: Smart cozy homekit

User Interface

Control window

Automation mode control

Weather forecast

Clothe suggestion

Air Pollution (PM2.5) warning

Page 11: Smart cozy homekit

No More Schedule

Page 12: Smart cozy homekit

Simplify Interface

Page 13: Smart cozy homekit

WeatherPM2.5 http://taqm.epa.gov.tw/taqm/tw/Pm25Index.aspx

API http://opendata.epa.gov.tw/Data/Details/AQX/

Page 14: Smart cozy homekit

Demonstration

Page 15: Smart cozy homekit

Future works● Our system can

○ extend to control more appliances, e.g., air condition.○ can collect more sensor data to learn more personal patterns.○ can also combine the AR technology to sell in-season clothes○ and more...

Page 16: Smart cozy homekit

Finally...Nuan Nan is 暖男 in Chinese = A considerate guy in English = Our system

Page 17: Smart cozy homekit

Thanks for your time!@Gene Cheng

@Calvin Zheng

@Vincent Huang

@Naima Hsu

@Ivy Liao

@Denker Huang