[Dev] Sensor Fusion solution for Sensor Framework - Gravity and Linear Acceleration

Ramasamy Kannan ram.kannan at samsung.com
Tue Apr 29 08:48:13 GMT 2014


Dear All,

In continuation of the mail below regarding the sensor fusion solution being developed for sensor framework, please
find the proposal for determination of Gravity and Linear acceleration. The document is a continuation of the 
orientation estimation - sensor fusion solution for sensor framework that was already shared as per the below mails.
Both Gravity and Linear acceleration are determined using the accurate sensor orientation estimated using sensor
fusion.

Please provide your comments, feedback and suggestions.

Regards,
Ram

------- Original Message -------
Sender : Ramasamy Kannan<ram.kannan at samsung.com>  Chief Engineer/SRI-Bangalore-TZN/Samsung Electronics
Date   : Apr 03, 2014 09:15 (GMT+05:30)
Title  : Re: [Dev] Sensor Fusion solution for Sensor Framework - Orientation
 Estimation

Hi,
Here are the answers below, based on current plan and visibility.

How do you intend to integrate with the location stack?
This will not go into the location stack. The code to obtain new sensor data obtained through fusion such as orientation, linear acceleration, gravity, altitude, pedometer... would be available as individual virtual sensor plugin. The sensor plugins (for both hardware and virtual sensors) will be part of the sensor framework. Applications can register for specific sensor events exposed by the sensor framework API to get access to the virtual sensor data.

Is this for phones/tablets/cars/fridges/etc?  
It could be used for any of these devices provided they include the sensor framework and also the device should support underlying hardware sensors/drivers needed for implementation of the virtual sensors.

Do you plan to use Kalman filtering as the general case and PDR for the special case or only Kalman?
Currently the plan is to use Kalman filter as a common code or plugin, for all virtual sensor plugins (to be supported in sensor framework) that may need its use for generating new sensor data. PDR or other navigation systems would be part of the application layer, can register and use the orientation, pedometer... data which are readily available from the sensor framework. There are different Kalman filter implementations available (simple ones which work well for floating point processing, more complex ones that may be needed for stability in case of fixed point processing). The kalman filter implementation would be decided as implementation progresses.

Regards,
Ram

------- Original Message -------
Sender : Leibowitz, Michael<michael.leibowitz at intel.com> 
Date   : Apr 03, 2014 03:22 (GMT+05:30)
Title  : Re: [Dev] Sensor Fusion solution for Sensor Framework - Orientation
 Estimation

On Tue, Apr 1, 2014 at 10:13 PM, Ramasamy Kannan <ram.kannan at samsung.com> wrote:
> Hi All,
>
> Please find attached the proposal document for Sensor Fusion solution for Tizen Sensor Framework. The first phase of development is to estimate device orientation by fusing data from accelerometer, gyroscope and magnetometer sensors. The proposed method uses sensor fusion algorithm based on Kalman filtering.
>
> Please provide your comments, feedback and suggestions.

Neat!  How do you intend to integrate with the location stack?  is
this for phones/tablets/cars/fridges/etc?  Do you plan to use Kalman
filtering as the general case and PDR for the special case or only
Kalman?

Cheers



-- 
Michael Leibowitz
_______________________________________________
Dev mailing list
Dev at lists.tizen.org
https://lists.tizen.org/listinfo/dev
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Sensor_Fusion_Gravity_and_Linear_Acceleration.pptx
Type: application/octet-stream
Size: 59085 bytes
Desc: not available
URL: <http://lists.tizen.org/pipermail/dev/attachments/20140429/cd94ee92/attachment-0001.obj>


More information about the Dev mailing list