|
Maybe this is the wrong thread, but I want to visualize some data, the visualization isn't the problem (yet) the problem would be modeling data. I want to visualize accelerometer data as positions, (this data can be cleaned and improved with on device magnetometer, barometer, and gyroscope data), by integrating twice and doing more math. Normally this wouldn't be possible(remotely accurate) because the small % errors at each tick cause huge errors when integrated twice. However I know several facts about the movement I am trying to capture that might help me in modeling it. For example approximate distance and target location will probably be known ahead of time, and once moved to a location the device will return near to the starting point. Starting point is also 0,0,0. Is anyone familiar with work like this or could point to resources?
|
# ¿ Oct 13, 2014 04:55 |
|
|
# ¿ May 18, 2024 02:18 |
|
ohgodwhat posted:A Kalman filter, perhaps? Kalman came up, and I have started doing the research. A normal 2d free body might be a good start with an extra dimension for z added, and there are some good resources for that.
|
# ¿ Oct 13, 2014 05:45 |