Hi,
I have acceleration data from Arduino with DXL335 accelerometer and I want to get the displacement plot.
How can I do that?
Kind Regards
Integrate over time and get velocity.
Integrate again and get displacement.
Mind systematic errors if you're going to use this in realtime application .... since they will cumulate over time. If it's only off-line analysis, then probably it's ok.
You didn't provide enough information about your dataset.
You can double integrate over time but each time you integrate you are inevitably introducing significant errors in the velocity and displacement estimates if your acceleration recording is prone to baseline drift and noise. All approximations from accelerometers will suffer from such errors and the displacement estimates will most likely be inadequate for most applications. If you attempt such approximations, make sure that you also know consider (an initial) condition(s) where the velocity takes zero value, since double integration will return the change in the velocity over each d(t), irrespective of the initial velocity.
The problem is that accelerometers are sensitive to any kind of acceleration, i.e.: due to translation, rotation (both tangential and centripetal/fugal accelerations) and gravity. That is, each accelerometer measures the sum of all the acceleration components acting along the accelerometer axis, and there is no means to separate them from one another. Of course, translational and rotational components both depend on the trajectory that is expected to be reconstructed, whereas the gravity component depends on the orientation of the accelerometer with respect to the earth vertical, which is also a priori unknown.
Then, in general, removing drift and offset before (double) integration is not enough to ensure reliable reconstruction. One should take into account any further available information about the kind of motion (e.g. 2D? 3D? circular? periodic? ...) and consider the possibility of measuring other variables. E.g., measuring also angular velocity by means of gyroscopes may help, but it isn't the eventual solution, because tangential and centrifugal accelerations depend on both angular velocity and trajectory radius, which is a priori unknown and may change at any instant during movement.
In conclusion, displacement plot reconstruction by double integration of acceleration measurement (and after removing drift and offset) may work only if you move either (i) along a straight line, with known and constant orientation with respect to gravity, or (ii) along a circular trajectory in a plane, with known radius and known orientation with respect to gravity, and you want to know where your object is at any instant. In all the other cases, you must get further information about the characteristics of the movement, and/or implement further processing of your data.
Examples of the errors in the reconstruction of body segments movement, measured by a set of either 6 or 9 3D accelerometers, can be found in Giansanti et al., IEEE Trans BME 50: 476-483, 2003.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Probability
- Probability Theory
- Reliability