Question/Issue: I am wondering if I can use data from two different devices to categorize a single element?
Context/Use case: I have two accelerometers/gyroscropes, one for each leg as a the dancer performs different moves. Is it possible to read in separate data from each sensor, and allow it to categorize one move? For example, if the dancer performs a spin, it looks different for the right and left foot, but I want the combination of what both sensors read to be categorized as a spin.
using a master and a slave device - where the master runs the inference from reading both its own sensor data and the slave data.
Using another device acting as a gateway.
In both cases, you need to be able to read the raw data at desired frequency, for accelerometer/gyroscope you are usually in the 100Hz range. Not a problem on the device, for the remote(s) devices, make sure your wireless protocol is supporting that and properly set up.
Thanks for getting back to me. To clarify, I am attempting to do this using prerecorded data and inputting it into the “Impulse design.” I have data for the right foot during a spin, and the left foot during the same spin. Rather than label them as right foot spin and left foot spin for the neural network to categorize, I want the impulse to be able to take in the right and left foot data and categorize the combined data as a spin.
You can name your axis like accX_left, accY_left, accZ_left, accX_right, accY_right, accZ_right, etc…
And then the process is just like any sensor fusion project: Sensor fusion | Edge Impulse Documentation
Is it alright that the time does not exactly match up for the right and left data? Specifically, even though I recorded data for both sensors when performing the same move, the timestamps do not occur at the exact same milliseconds. Furthermore, sometimes there is less data for the the left than there is for the right.