topics in designing an android sensor subsystem_ pitfalls and considerations presentation

8/12/2019 Topics in Designing an Android Sensor Subsystem_ Pitfalls and Considerations Presentation

1/35

Designing An Android

Sensor Subsystem

Pitfalls and Considerations

Jen Costillo

[email protected]


2/35

Simple Choices

Userexperience

Batteryperformance

7/15/2012 2Costillo- OSCON 2012


3/35

Established

or Innovative Product?

Established

Will I be making another

new product in 6

months?

Is the reference design

considered good enough

for the application?

Innovation-Driven

Do I have new sensors

types?

Are features more

important than release

date?

Are money and

resources noproblem?



4/35

Forsaking Reference Designs



5/35

Going On Your Own

If you make your own,

Youre on your own

Integration pains

Test time Gesture testing

becomes a challenge

Calibration blues

Larger mechanicalfootprint

But

power

Control code size

Control mechanicalfootprint

In-house expertise

7/15/2012 Costillo- OSCON 2012 5


6/35

Android Universe

7/15/2012 Costillo- OSCON 2012 6

Sensor Hub/Coprocessor Sensors

Interface Kernel

Driver Sensor Driver

Sensor HAL

Sensor Service Sensor Manager

Sensor JNI

SensorManager

Android Application

Hardware

Linux Kernel

Libraries

Frameworks

Application


7/35

HARDWAREHardware


Linux Kernel

Libraries

Frameworks

Application


8/35

Hardware Architecture



9/35

Sampling Rates:

The 3 RatesUnder-sampling

Inaccurate, sluggishresponse

Slight power savings

Sampling

Rate

Over-sampling

Accurate, smoothresponse

Power-hungry



10/35

Wake up events and power

considerationsApplication

Processor only

Internal

Coprocessor

External

Processor

D

Reference supported

Most power hungry

Reference supported

Most work done foryou

More processor selection

More outcome controlMost customized

Footprint impact



11/35

Hardware Summary

PowerConsumption= sensorsn+any dedicated

processor

Latency =Max(sensorsn)

+ dedicatedprocessing

time

SensorSolution

Use tie-breaker criteria

7/15/2012 Costillo- OSCON 2012 11


12/35


13/35

Sensor

Kernel Driver


CoprocessorMicrocontroller

Application Processor

PeripheralInterface

SharedMemory


14/35

LIBRARIES ANDSERVICES

Libraries


Hardware

Frameworks

Application

Linux Kernel


15/35

Sensor HAL and Services

HALdevice///libsensors

Serviceframeworks/base/services/sensorservice

Managerframeworks/base/libs/gui



16/35

Sensor Fusion

7/15/2012 Costillo- OSCON 2012 16

Sensors

Linux Kernel

Libraries

Sensor Hub

http://en.wikipedia.org/wiki/Sensor_fusion

https://www.llnl.gov/news/newsreleases/2010/NR-10-01-06.html
http://en.wikipedia.org/wiki/Sensor_fusionhttps://www.llnl.gov/news/newsreleases/2010/NR-10-01-06.htmlhttps://www.llnl.gov/news/newsreleases/2010/NR-10-01-06.htmlhttps://www.llnl.gov/news/newsreleases/2010/NR-10-01-06.htmlhttps://www.llnl.gov/news/newsreleases/2010/NR-10-01-06.htmlhttps://www.llnl.gov/news/newsreleases/2010/NR-10-01-06.htmlhttps://www.llnl.gov/news/newsreleases/2010/NR-10-01-06.htmlhttps://www.llnl.gov/news/newsreleases/2010/NR-10-01-06.htmlhttps://www.llnl.gov/news/newsreleases/2010/NR-10-01-06.htmlhttp://en.wikipedia.org/wiki/Sensor_fusionhttp://en.wikipedia.org/wiki/Sensor_fusion


17/35


18/35

Gesture Detection Comparison

Make Buy

Completesolution

Already Tested&tuned

MinimalSchedule Impact

Compact code

In-houseexpertise

Off-load AppPro

Application

Powerful processor

Sensor hub

Off-load to cheaperpower

Wake up EventHandling

7/15/2012 Costillo- OSCON 2012 18


19/35

Calibration


Use of Calibration Gesture in the

Compass App By Catch.com


20/35

FRAMEWORKS


Hardware

Application

Linux Kernel

Libraries

Frameworks


21/35

Virtual Sensors

Leverages 1+ physical or other virtualsensors

Multiple Options

Google (version 3)

Reference Vendor

Sensor Vendors

7/15/2012 Costillo- OSCON 2012 21


22/35


23/35

Android Virtual Sensors

Gyro Accel Orientation

Compass Accel Rotation

7/15/2012 Costillo- OSCON 2012 23


24/35

Virtual Sensors Challenges

Garbage In- Garbage Out

Latency

Non-Synchronizedsamples

Implementation

Dependencies Multi-vendor problems,

verify Vendor ID

http://www.invensense.com/midc/presentations/James%20Lim.pdf7/15/2012 24Costillo- OSCON 2012

Sensor1

Sensor

2Virtual

Output

Sensor

1

Sensor

2

Virtual

1

Virtual

2


25/35

APPLICATIONS


Hardware

Linux Kernel

Libraries

Application

Frameworks


26/35

Using Sensors

7/15/2012 Costillo- OSCON 2012 26

mSensorManager =

getSystemService(Context.SENSOR_SERVICE);

mSensorManager.registerListener(

mSensorListener,

mSensorManager.getDefaultSensor(

Sensor.TYPE_ACCELEROMETER),

SensorManager.SENSOR_DELAY_NORMAL);

void onSensorChanged(SensorEvent event) {

// get sensor data

float x =

event.values[SensorManager.DATA_X];

}


27/35


28/35

Types of Sensor Problems

Bias

Drift

Settling Time Jitter/Noise

Environmental

Interference

7/15/2012 Costillo- OSCON 2012 28


29/35

Bias

Problem: Data is offby a constant value.

Sources:

static calibrationfailure

Solutions:

Calculate linear

offset at start ofapplication

Recalibrate locally

7/15/2012 Costillo- OSCON 2012 29


30/35

Drift

Problem: Shift of

data without cause

Sources:

Magneticinterference

Poor HW calibration

Solutions:

Increase smoothing

techniques

7/15/2012 Costillo- OSCON 2012 30


31/35

Settling Time Problem: Extended

time before

finalized steady

data.

Sources: Latency

Sensitivity

Solutions:

Limit additional

processing

7/15/2012 Costillo- OSCON 2012 31


32/35

Noise

Problem: Data jumpsaround constantly

Sources:

Sensor

Calibration

Poor filtering

Solutions:

High pass filter

Linear averaging

FFT

7/15/2012 Costillo- OSCON 2012 32


33/35

Environmental Interference

Problem: Inconsistentresults

Sources:

Magnetometer

EMI

Solutions:

Reference Device

Calibration gesture

7/15/2012 Costillo- OSCON 2012 33


34/35

Best Practices in

Application Development

Select the right sensor for the job.

Use the Correct Data Rate.

UI or GAMING are the most common.

Use Sensor In Context

Customize for your hardware and system

capabilities

Magnetometer-based sensors are the mosttouchy.

Keep the Gesture UI simple.



35/35

[email protected]

Additional resources

http://processors.wiki.ti.com/index.php/Android_Sensor_PortingGuidehttp://www.kandroid.org/online-pdk/guide/sensors.html

http://invensense.com/midc/

http://developer.android.com/reference/android/hardware/SensorEvent.html

http://processors.wiki.ti.com/index.php/Android_Sensor_PortingGuidehttp://www.kandroid.org/online-pdk/guide/sensors.htmlhttp://invensense.com/midc/http://developer.android.com/reference/android/hardware/SensorEvent.htmlhttp://developer.android.com/reference/android/hardware/SensorEvent.htmlhttp://invensense.com/midc/http://www.kandroid.org/online-pdk/guide/sensors.htmlhttp://www.kandroid.org/online-pdk/guide/sensors.htmlhttp://www.kandroid.org/online-pdk/guide/sensors.htmlhttp://processors.wiki.ti.com/index.php/Android_Sensor_PortingGuide

topics in designing an android sensor subsystem_ pitfalls and considerations presentation

Documents