Only Y axis filtered from Accelerometer

rcaboni

I need to check the sensor inclination.If the inclination is over 45° i need to do samethink.
I read the accelerometer with :

    accelModule.routeData()
    .fromAxes()
    .stream("accelSub")/*.log("accelLogger")*/
    .commit()
    .onComplete(new CompletionHandler<RouteManager>() {
        @Override
        public void success(RouteManager result) {
            result.subscribe("accelSub", new RouteManager.MessageHandler() {
                @Override
                public void process(Message msg) {
                    final CartesianShort axisData = msg.getData(CartesianShort.class);
                    int inclination = (int) (axisData.y()/11.41);
                    String val = String.format("%s %d°", axisData.toString(),inclination);
                    Log.i("test", "Stream: "+val);
                }
            });

            .......
        }

    });
Ok, work fine. But i would like to filter the data under 45° because i think this limit battery usage.
I try to read acceletometer only for the Y axies,(because i read the Comparison is not possibile on all axies), with fromYAxis but the value is very different.
                   accelModule.routeData()
                           .fromYAxis()
                           .process(new Comparison(Comparison.Operation.GTE, 500))
                           .stream("accelSub")
                           /*.log("accelLogger")*/
                           .commit()
                           .onComplete(new CompletionHandler<RouteManager>() {
                               @Override
                               public void success(RouteManager result) {
                                   result.subscribe("accelSub", new RouteManager.MessageHandler() {
                                       @Override
                                       public void process(Message msg) {

                                           Bmi160SingleAxisMessage bmi160Msg= (Bmi160SingleAxisMessage) msg;
                                           Log.i("test","SingleAxisMessage Y = "+msg.getData(Float.class));

                                       }
                                   });

                                   .....
                               }
                           });
If i read CartesianShort with fromAxes the Y is beetween 0 - 1030
but if i read Bmi160SingleAxisMessage with fromYAxis the interval is 500 (correct because i Comparison it) - 49000
Same one have an explanation or same suggestions

Eric

It appears that the filter is returning both Y and Z axis data as opposed to only Y axis.  You can compute the Y axis value yourself from the byte data.

result.subscribe("accelSub", new RouteManager.MessageHandler() {
    @Override
    public void process(Message msg) {
        Bmi160SingleAxisMessage bmi160Msg= (Bmi160SingleAxisMessage) msg;
        short unscaled= (short) (bmi160Msg.getData()[0] | (bmi160Msg.getData()[1] << 8));
        Log.i("test", "Y-Axis= " + (unscaled / bmi160Msg.getScale()));
    }
});

rcaboni

Thanks Eric for your answer.

I understand there is a little bug and the API return Y and Z value instead only Y.

But in this way the (unscaled / bmi160Msg.getScale()) is often a low value.
If i don't move the sensor, Y value are same time correct, and same time very low (under 0).
In fromAxes this not occur. The read is more accurate.

With fromYAxis() i cannot you the
process(new Comparison(Comparison.Operation.GTE, 500))
because, i imagine, the comparison is with YZ value and not only on Y axies.
Correct ?
But i want to limit only Y.

I start to think it's not possible, to actual firmware version, to use fromYAxis for limit the information transmitted on BT.

Eric

Hrm, yes I believe your hypothesis is correct about comparing a YZ value.  What you can do to work around this is in the meantime is to attach a Maths processor that adds 0 before you compare it.

accelCtrllr.routeData().fromYAxis()
    .process(new Maths(Maths.Operation.ADD, 0))
    .process(new Comparison(Comparison.Operation.LTE, -0.8))

Also, the acceleration units are g, not mg; your value should be 0.5 not 500. Lastly, set the axis sampling range before you construct the data processing chain.

rcaboni

Thanks Eric, but i solution is not very good.

I don't know if i understand well "set the axis sampling range before you construct the data processing chain".

My source is :



accelModule.routeData()
.fromYAxis()
.process(new Maths(Maths.Operation.ADD, 0))
.process(new Comparison(Comparison.Operation.LTE, 0.5))
.stream("accelSub")
.commit()
.onComplete(new CompletionHandler() {
@Override
public void success(RouteManager result) {

result.subscribe("accelSub", new RouteManager.MessageHandler() {
@Override
public void process(Message msg) {
try {
Bmi160SingleAxisMessage bmi160Msg = (Bmi160SingleAxisMessage) msg;
short unscaled = (short) (bmi160Msg.getData()[0] | (bmi160Msg.getData()[1] << 8));

final Float yAxisData = (unscaled / bmi160Msg.getScale()) * 1000;
int inclination = (int) (yAxisData / 11.41);
String val = String.format("%f %d°", yAxisData, inclination);
Log.i("test", "Stream: " + val);
((TextView) findViewById(R.id.textView3)).setText(val);

}catch (Exception ex) {
Log.e("test","Error in CompletionHandler process : ",ex);
ex.printStackTrace();
}
}
});

accelModule.setOutputDataRate(50.f);

accelSetup = true;
loggingModule.startLogging();

accelModule.enableAxisSampling();
accelModule.start();
}

@Override
public void failure(Throwable error) {
Log.e("test", "Error committing route", error);
accelSetup = false;
}
});

In this way i received often many Y-axis zero, this not occur if i use fromAxes.

Have you a forecast or the new firmware upgrade with the corrected fromYAxis() ?

Eric

• What is not good about my solution?  Of course it's not ideal but this is what you have to do to temporarily work around the issue
• There is a function called setAxisSamplingRange for the Accelerometer class.
• You don't need the byte manipulation code using the workaround nor do you need to cast the msg object.  You can revert your code to getting the data as a float.
• You cannot use a comparator with the entire XYZ acceleration data (fromAxes), only with the individual axis (fromXAxis).
• It has been filed in our bug tracker but there is no timetable for addressing this issue.

rcaboni

Thank you very much, Erik,
   i remove the bit manipulation and cast the data to a float, and now all work very well.
I update the code if same one need it in the future :

accelModule.routeData()
        .fromYAxis()
        .process(new Maths(Maths.Operation.ADD, 0))
        .process(new Comparison(Comparison.Operation.LTE, 0.5))
        .stream("accelSub")
        .commit()
        .onComplete(new CompletionHandler<RouteManager>() {
            @Override
            public void success(RouteManager result) {

                result.subscribe("accelSub", new RouteManager.MessageHandler() {
                    @Override
                    public void process(Message msg) {
                        try {
                            Float yAxisData = msg.getData(Float.class) *1000 -40;
                            int inclination = (int) (yAxisData / 10.86);
                            String val = String.format("%f %d°", yAxisData, inclination);
                            Log.i("test", "Stream: " + val);
                            ((TextView) findViewById(R.id.textView3)).setText(val);

                        }catch (Exception ex) {
                            Log.e("test","Error in CompletionHandler process : ",ex);
                            ex.printStackTrace();
                        }
                    }
                });

                accelModule.setOutputDataRate(50.f);

                accelSetup = true;
                loggingModule.startLogging();

                accelModule.enableAxisSampling();
                accelModule.start();
            }

            @Override
            public void failure(Throwable error) {
                Log.e("test", "Error committing route", error);
                accelSetup = false;
            }
        });

Thanks for your support again.