we are trying to get fall detection using LMS6DSOX in sensor tile box? 1. How to calculate or get fall detection? 2. Is free-fall same as fall detection?If not how to come up with the detection 3. Kindly share pdf for starters if any

[RShiv.1]

Kindly throw us some light as we need to detect fall detection using STMR9 with LSM6DSOX

waiting for your response

[RShiv.1]

Hi,

any update on how to use lsm6dsox to detect fall detecion..kindly throw some light.

regards

Ravi chandran

[Eleon BORLINI · Accepted solution]

Hi Ravi @RShiv.1​ ,

Currently, the free fall detection (a bit different from a man-down application, but "yes" to question 2) is implemented at firmware level, not exploiting the embedded feature of the LSM6DSOX.

>> How to calculate or get fall detection?

In this video, it is described how to implement the free fall detection via the STBLE sensor smartphone application.

You can just set the threshold above which the free-fall is defined (in the example, 200 milli-g) and you'll receive a notification every times this threshold is crossed.

If you want to implement the free-fall at firmware level, you can start from the FP-SNS-STBOX1 examples.

If otherwise you want to exploit the embedded feature of the LSM6DSOX for the free fall detection as explained in the app note AN5272, p.42, you can exploit the INT1_DSR and INT2_DSR pins from the schematic:

 

If my reply answered your question, please click on Select as Best at the bottom of this post. This will help other users with the same issue to find the answer faster. 

-Eleon

[RShiv.1]

Hi Eleon,

thanks for your reply.

1.So now if I want to use this for fall detection in case of wearable which uses this chip,then how do I use this?

2.what needs to be configured? I am planning to use sensorTile Box for initial testing.

3.should I use the machine learning core which is there inside this?

4.If machine learning is used how can I capture the inputs and generate ,and give this to decision tree?Algorithm if any.

kindly let me know how to proceed to get the fall detection

regards

Ravi chandran

[RShiv.1]

Hi eleon,

we are looking at mna down application..kindly let us know how to detect this using lsm6dsox .we are trying to us sensorTile box as start point.

--Ravi

[Eleon BORLINI]

Hi Ravi @RShiv.1​ ,

If you want to use the Sensortile.box as a free fall detection tool using the the LSM6DSOX, I suggest you to start from an existing FW, configure the LSM6DSOX for the free fall detection and then enable a LED (for example the green one).

You can do it as described in the previously linked video.

You don't need to run the MLC, in principle.

However, if you are interested in a more complex pattern recognition (such as a particular man-down), you might need to set up a more complex configuration.

In this case, on the BLEMLC section (p.17 of the user manual UM2626), and from slide 280 of this presentation you can find more information for a starting point.

-Eleon

[RShiv.1]

Hi Eleon,

Thanks for your reply.As said earlier we wanted Man down application only.Thanks for sharing the BLEMLC user manual and presentation so that we have a starting point to this.Now going deeper is this application already implemented in lsm6DSOX??,...any paper leads so that we can refer for this...kindly let us know

--Ravi

[Eleon BORLINI]

Hi --Ravi @RShiv.1​ ,

unfortunately there is no demo application available for the man-down event detection, mostly because it should work 100% of the times, since it is a medical application, with related responsibilities.

You could however make some basic tests exploiting the free fall, and then improve your tests by training a proper MLC. You might for example use the 6D position recognition, that should be able to distinguish between stand position and flat position. But you should carry out some test to test whether it can be reliable enough for your purpose.

There is also a broader literature on the topic, that you can find online and that could drive you to a better implementation of the algorithm.

-Eleon

[RShiv.1]

Hi eleon,

Thanks for your valuable reply and suggestion.we will take care of what all you said during our algorithem implementation.will come back to you if I have some query.

Thanks and regards

Ravi chandran