Maker Approach to Product Innovation BRINGING TO LIFE WEARABLE / IoT - - PowerPoint PPT Presentation
Maker Approach to Product Innovation BRINGING TO LIFE WEARABLE / IoT IDEAS With RAPID PROTOTYPING using Open HW and SW MOE T MOE TANABIAN ANABIAN VP of Engineering | Head of IoT Innovation Lab Samsung Electronics ABOUT ME MOE TANABIAN Vice
BRINGING TO LIFE WEARABLE / IoT IDEAS
With RAPID PROTOTYPING using Open HW and SW
MOE T MOE TANABIAN ANABIAN
VP of Engineering | Head of IoT Innovation Lab Samsung Electronics
MOE TANABIAN Vice President of Engineering, Head of Smart Things IoT Innovation Lab Samsung Electronics, San Jose, CA 16 years of industry experience in building and launching CE, Mobility and Wireless products in companies such as Samsung, Amazon, Nortel
2
3 ¡
BACKUP ¡SLIDES ¡ ¡
Silicon Valley way of Innovation by Tinkering, Hacking and Making
4
5
THEN HOW ARE GREAT NEW PRODUCTS BORN? Lets look at How Silicon Valley (and our lab) Innovate…
7
It all starts with a burning desire, a missing piece –
8
And–is this a material addressable market with growth potential? & can it be built and can Samsung ship it and make it a business?
9
If all YES, Then we get together start working on it
10
Quick guideline for where to look for potentials and pain points
Samsung ¡can ¡fulfill ¡
¡ Combine ¡Design ¡and ¡Technology ¡address ¡the ¡ Experience ¡Gap: ¡ ¡
ID ¡
11
MulA-‑Device ¡ Experiences ¡ Best ¡UI ¡is ¡ ¡ No ¡UI ¡ Superb ¡SIMPLE ¡ Experiences ¡
Projected U.S. Biofuel Source: Biomass as Feedstock for a Bioenergy and Bioproducts Industry:Seamless D2D experiences across Samsung devices Drastically Simple UX – Attractive, Natural design Context, Machine Learning – to enable minimal to No UI UX
Quick guideline for where to look for potentials and pain points
12
Then we do the Initial coarse designs
13
AND THEN WE MAKE STUFF
14
And we write code…
15
Make more stuff…
16
Write more code…
17
Test again the design for UX, and Technical…
Make ¡ ¡Test ¡ ¡Validate ¡
18
We do this cycle a few (~6-15) times (each taking 2-3 weeks)
19
And fjnally….
20
Then we get some sleep and rest (Once in a while!)
21
It’s a collaborative process with Samsung headquarters
22
And celebrate when we have completed the project
23
Mix of Design & Technology Rapid Iterations
TENETS OF SUCCESSFUL MAKING How do we do this?
Duality of skills in x- Functional Small Teams
25
26
27
Also Worth reading
Tenet ¡3: ¡x-‑FuncAonal ¡Team, ¡x-‑Skill ¡Learning ¡
28
29 ¡
With RAPID PROTOTYPING using Open HW and SW
OUTLINE
30
WEARABLES: Design Success Factors
Connectivity & Sensors Energy & Battery Consumption Reference Design for Wearable Experiments
31
CHARACTERISTICS OF WEARABLES
UX: COST VS BENEFITS
32
SOCIAL WEIGHT
CL → cognitive load PP → physical presence SC → social convention
33
Source:A. ¡Toney, ¡B. ¡Mulley, ¡B. ¡H. ¡Thomas, ¡and ¡W. ¡Piekarski, ¡“Social ¡weight: ¡designing ¡to ¡minimise ¡the ¡social ¡consequences ¡arising ¡from ¡technology ¡use ¡by ¡the ¡mobile ¡professional,” ¡Personal ¡and ¡Ubiquitous ¡ Compu0ng, ¡vol. ¡7, ¡no. ¡5, ¡pp. ¡309–320, ¡2003.
CL: Cognitive Load -- PP: Physical Presence -- SC: Social Convention
CL: • PP: • SC: • CL: ••• PP: •• SC: • CL: •••••• PP: ••• SC: ••
34
SOCIAL WEIGHT
CL: ••• PP: ••• SC: ••• CL: ••• PP: •••• SC: ••••• CL: ? PP: •••••••••• SC: ••••••••••••••••••••
35
CL: Cognitive Load -- PP: Physical Presence -- SC: Social Convention SOCIAL WEIGHT
Integration with other platforms & devices Discoverability of functionality How many wearables/person? Turning it off and showing that state: e.g. Glass vs Autographer Cost
36
OTHER UX CHALLENGES IN WEARABLES
OUTLINE
37
WEARABLES: Design Success Factors
CONNECTIVITY & SENSORS
Energy & Battery Consumption Reference Design for Wearable Experiments
Voice ¡ Data ¡ Audio ¡ Video ¡ State ¡ Bluetooth ¡
Y ¡ Y ¡ Y ¡ N ¡ N ¡
BLE ¡
N ¡ N ¡ N ¡ N ¡ Y ¡
Wi-‑Fi ¡
Y ¡ Y ¡ Y ¡ Y ¡ N ¡
Wi-‑Fi ¡Direct ¡
Y ¡ Y ¡ Y ¡ N ¡ N ¡
ZigBee ¡
N ¡ N ¡ N ¡ N ¡ Y ¡
ANT ¡
N ¡ N ¡ N ¡ N ¡ Y ¡
State: ¡ ¡
Low ¡bandwidth, ¡ ¡ Low ¡Latency, ¡ ¡ Low ¡Power ¡Data ¡
Source: IEEE
38
CONNECTIVITY AND SENSORS IN WEARABLES – SHORT RANGE CONNECTIVITY OPTIONS
It’s good at small discrete Data transfers. It has new Radio, new Protocol stack and new Profjle architecture
Latency fast transactions (~3ms from start to fjnish)
NW ¡Available ¡ 73.0F ¡ Pause ¡|| ¡ 49.6 ¡M/h ¡ 11:24AM ¡
¡ BLE is optimized for linking things that have “Data” and “Web Services” that want this “Data” ¡
Source: IEEE
39
CONNECTIVITY AND SENSORS IN WEARABLES – CONNECTIVITY - BLE
3 Advertising Channels and 37 Data Channels (2MHz each)
Source: IEEE
40
CONNECTIVITY AND SENSORS IN WEARABLES – CONNECTIVITY - BLE
Time (us) Master Tx Radio Active (us) Slave Tx
0 ¡
176 ¡ ADV_DIRECT_IND ¡
326 ¡
CONNECT_REQ ¡ 352 ¡
1928 ¡
Empty ¡Packet ¡ 80 ¡
2158 ¡
144 ¡ Aeribute ¡Protocol ¡ Handle ¡Valid ¡Indica0on ¡
2452 ¡
Empty ¡Packet ¡ACK ¡ 80 ¡
2682 ¡
96 ¡ LL_TERMINATE_IND ¡
2928 ¡
Empty ¡Packet ¡ACK ¡ 80 ¡
Source: IEEE
41
CONNECTIVITY AND SENSORS IN WEARABLES – CONNECTIVITY - BLE
How does BLE achieve Low Energy?
Source: IEEE
42
CONNECTIVITY AND SENSORS IN WEARABLES – CONNECTIVITY - BLE
43
There are other short range wireless technologies that can potentially be built in to Wearable devices:
complex use cases. The drawbacks are:
BLE is seems to be winning the battle of the Wearables – especially the devices that are often used in pair with a Smartphone
CONNECTIVITY AND SENSORS IN WEARABLES – CONNECTIVITY – OTHER (WI-FI, ZIGBEE, ETC)
44
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – ACCELOROMETER (MEMS)
45
3pl Axis Accelerometer – ADXL362
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – ACCELOROMETER PROTOTYPING EXAMPLE
46
3pl Axis Accelerometer – ADXL362 CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – ACCELOROMETER PROTOTYPING EXAMPLE
47
MEMS based Gyros are devices that detects rotation by turning it to some electrical signal change. They are based on Corolis force.
in resistance and some to capacitance
interface (e.g. I2C)
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – GYRO PROTOTYPING EXAMPLE
48
3pl Axis Gyro – ITG-3200
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – GYRO PROTOTYPING EXAMPLE
49
3pl Axis Gyro – ITG-3200 CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – GYRO PROTOTYPING EXAMPLE
50
Magnetometer is a fjeld magnetic sensing with a digital interface for applications such as compassing and magnetometry.
ADC to output digital signal
interface (e.g. I2C)
applications
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – MAGNETOMETER
51
3pl Axis Magnetometer – HMC5883L
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – MAGNETOMETER PROTOTYPING EXAMPLE
52
3pl Axis Magnetometer – MAG3110 CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – MAGNETOMETER PROTOTYPING EXAMPLE
53
Pressure and Altitude sensor is often a MEMS based sensor that can measure that change in pressure of – liquid or gas
Piezoelectric, Piezoresistive, etc
due to altitude change of a few centimeters
interface (e.g. I2C)
movements
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – PRESSURE AND ALTITUDE SENSOR
54
Pressure Altitude Sensor – MPL31152 ¡
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS –ALTITUDE SENSOR PROTOTYPING EXAMPLE
55
Pressure Altitude Sensor – MPL31152 CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – ALTITUDE SENSOR PROTOTYPING EXAMPLE
56
If packaging allows, it is often more efficient to use a single chip combining several sensors, or Sensor Fusion chips. The actual sensor can be Capacitive, Piezoelectric, Piezoresistive, etc
which also includes fjrmware for more complicated functions such as gesture detection
estate and power are at premium and sensors are included only if they are essential
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS – SENSOR FUSION
57
SENSOR FUSION– MPU-9150
SOURCE CODE EXAMPLE:
heps://github.com/sparkfun/MPU-‑9150_Breakout/blob/master/ firmware/MPU6050/Examples/MPU6050_DMP6/MPU6050_DMP6.ino ¡
CONNECTIVITY AND SENSORS IN WEARABLES – SENSORS –SENSOR FUSION PROTOTYPING EXAMPLE
OUTLINE
58
WEARABLES: Design Success Factors Connectivity & Sensors ENERGY & BATTERY CONSUMPTION Reference Designs for Wearable Experiments
59
ARGUABLY BATTERY LIFE IS THE MOST CHALLENGING PROBLEM IN DESIGING WEARABLE DEVICES
60
BATTERY HARDWARE SOFTWARE THE 3 INFLUENCING ELEMENTS ON BATTERY LIFE – BATTERY, HARDWARE, AND SOFTWARE
Lithium Ion (Li-ion): Strength:
§ The newest and fastest growing battery technology. § Li-ion batteries are smaller and lighter (higher energy density) Weakness: § Charging time too long, § Still low energy density § They are more expensive Application: the main type of battery used in mobile devices and handsets today Li-ion with steady current within its dominated C can provide 500-700 charge
61
BATTERIES IN MOBILE DEVICES - BATTERY CHEMISTRY
Source: Batteries in a portable world by Isidor Buchmann
62
BATTERIES IN MOBILE DEVICES - DISCHARGE CHARACTERISTICS OF LI-ION
63
BATTERIES IN MOBILE DEVICES - RECENT ADVANCES IN BATTERY TECHNOLOGY t
Source: Batteries in a portable world by Isidor Buchmann
Smart battery pack should be able to report:
They come in different capabilities:
I2C data communication
64
BATTERIES IN MOBILE DEVICES - SMART BATTERY PACKS
Radios, Screen and CPU are the top power consuming components of a typical wearable device
Ap Application plication Pr Proces
sor
BLE BLE Transceiv ansceiver er Wi-Fi Wi-Fi Transceiv ansceiver er
Audio AMP udio AMP & Codec & Codec Ac Acceler celerometer
Magnetic sensor Magnetic sensor
GP GPS
Receiv eceiver er Gyr Gyrosc
Gr Graphics aphics
Pr Proces
sor (GPU) (GPU)
Display Display
1-1.5” L 5” LCD, OLED CD, OLED
Very ery High P High Power er High P High Power er Moder Moderate P ate Power er
To conserve power, Power Manager turns off components that are not being used
MEMOR MEMORY eMMC eMMC
Power er Manager Manager IO IO (U (USB) SB)
Camer Camera
CCD/ CD/CMOS CMOS (sensor) (sensor) AFE AFE
65
HARDWARE - MAJOR POWER CONSUMING COMPONENTS
High High Component P Component Power er State State Lo Low w TIME TIME INA INACTIVE CTIVE LOW P W POWER WER HIGH P HIGH POWER WER (A (ACTIVE) CTIVE) User User activity activity PM PM Shutdo Shutdown wn Timer Timer 66
SOFTWARE In general, a power manager saves battery by shutting down components that are idle for a specifjc period of time
Areas to optimize for power There are many areas that SW developer can control, and optimize the power that the app draws, often without compromising user experience: n CPU n Radio n Display n Disk / Flash Access n Alarms n Sensors n Graphics / UI
67
SOFTWARE Power optimization in sw: the most influence over battery life
CPU optimization n Any optimization that reduces demand on CPU, will make the app more power efficient:
Ø Algorithm optimization (generally better O() == less power) Ø Caching data, intermediate results, UI assets Ø Batching CPU intensive jobs to allow the CPU to shut down Ø And fjnally: Do not execute code if it doesn’t do anything! (Disable background services when not needed)
Disk / Flash n Caching n Batching R/W access
68
SOFTWARE – POWER OPTIMIZATION IN MOBILE SW – CPU, DISK FLASH
69
n Display is often (among) the top power consumers on a mobile device n Display power optimization can and should be beyond simple use of use locks, etc n Intelligent handling of screen brightness can reduce the overall power consumption of device signifjcantly DISPLAY
69
SOFTWARE – POWER OPTIMIZATION IN APPS – UI AND DISPLAY POWER OPTIMIZATION
70
n Power is measured directly from the aggregate point of battery terminal n The physical device is needed to conduct this method n Each component of interest is separately stress loaded to determine its power consumption n During measurement for each component, the component power coefficients in relation to relevant system variables are determined Direct Method Direct Method n Based on a model of the device’s power consumption (often Linear Regression) n The model is trained using a series of Direct measurements n The model can then be used to predict power consumption based on the values of systems variables – collected from logs Indirect Method Indirect Method
70
POWER MEASUREMENT – DIRECT AND INDIRECT
Steps for Direct measurement method:
1. Identify major components to be included in the model e.g. Display
corresponding to each component e.g. screen brightness intensity 3. In a series of measurements, isolate each major component and measure the power consumption factor directly
Cur Current sensing ent sensing resistor esistor Stabilizing Stabilizing Capacitor Capacitor
Bat Battery tery / / Power er sup supply ply
Source: Google, U. of Michigan
71
POWER MEASUREMENT – DIRECT METHOD
Steps for building and training the power model:
model as a function of all the
consumption of the different use cases and tasks in your applications
Source: Google, U. of Michigan
72
POWER MEASUREMENT – INDIRECT METHOD
73
Produced by referencing : Into the wild, Studying Real User Activity patterns to guild power optimization, by Alex Shye et al
Direct (Measured) vs Indirect (Predicted) power consumption POWER MEASUREMENT – COMPARISON OF DIRECT AND INDIRECT METHODS
PREDICTED POWER
74
n Instrument to sample voltage, current and power at high sampling frequency rates (1000 sample per second and above) n System to host simulated cloud services n Equipment to measure ambient light n Equipment to measure ambient noise
LAB FUNCTIONS
POWER MEASUREMENT – HOW TO SETUP AND A POWER MEASUREMENT LAB
OUTLINE
75
Wearables: What & Why? Form Factor & Ergonomics Wearable I/O UX Challenges in Wearables Connectivity & Sensors Energy & Battery Consumption REFERENCE DESIGNS FOR WEARABLE EXPERIMENTS
Component selection for a Wearable heavily depends on its primary purpose
range CPU
76
REFERENCE DESIGN FOR WEARABLES – HARDWARE
Sensor Sensor Fusion usion Module Module Power er Management Management Cur Cursor sor Libr Library ary Motion Motion Output Output And And USB SB/SPI /SPI Interf Interface ace Calibr Calibration ation Activity Activity Clas Classi sifj fjcation cation De Device vice Sup Support port 3- 3-Axis Axis Linear Linear Ac Acceler celerometer
3- 3-Axis Axis Gyr Gyrosc
77
REFERENCE DESIGN FOR WEARABLES – FITNESS TRACKING – HILLCREST LABS’s FSM Series
78
MCU U / / Application n Pr Proces
sor
Power er Management Management Stor Storage age eMMC eMMC UI UI But Button ton
Sensor Sensors
Wir Wireles eless Bat Battery tery I/ I/O uU uUSB SB Display Display 1-1.5” OLED 5” OLED
REFERENCE DESIGN FOR WEARABLES – HARDWARE - SMARTWATCH – BLOCK DIAGRAM
Ap Application plication Pr Proces
sor
BLE BLE Transceiv ansceiver er Wi-Fi Wi-Fi Transceiv ansceiver er
Audio AMP udio AMP & Codec & Codec Ac Acceler celerometer
Magnetic sensor Magnetic sensor
GP GPS
Receiv eceiver er Gyr Gyrosc
Gr Graphics aphics
Pr Proces
sor (GPU) (GPU)
Display Display
1-1.5” L 5” LCD, OLED CD, OLED
MEMOR MEMORY eMMC eMMC
Power er Manager Manager IO IO (U (USB) SB)
Camer Camera
CCD/ CD/CMOS CMOS (sensor) (sensor) AFE AFE
79
REFERENCE DESIGN FOR WEARABLES – HARDWARE - WEARABLE CAMERA or GLASS
3P APPS?
MULTIPROCESSING? ¡
You probably need an OS:
You can probably get away WITHOUT an OS Y Y N N
80
REFERENCE DESIGN FOR WEARABLES – SOFTWARE – TO OS OR NOT TO OS
Social Weight = Cognitive Load + Physical Presence + Social Convention.
validation of your device
81
82
(Uno, Mega, etc)
(9DOF Sensor stick, etc) RAPID PROTOTYPE OF A HEALTH TRACKER
83
RAPID PROTOTYPE OF A HEALTH TRACKER
84 ¡
BACKUP ¡SLIDES ¡ ¡
MOE TANABIAN @montanabian www.linkedin.com/in/mtanabian/
85
Moe Moe Tanabian anabian
@motanabian http://www.linkedin.com/in/mtanabian mometan@gmail.com
85