So actually, the LPS data is already being fused with the onboard kalman filter on the crazyflie so it is already fused with the accelerometers. But ofcourse there might be improvements on how the measurement model is implemented or how the noise estimates are handled.
Here is where you can find how the measurement models used by the LPS systems is implemented:
(TWR) https://github.com/bitcraze/crazyflie-f ... ore.c#L366
(TDOA) https://github.com/bitcraze/crazyflie-f ... ore.c#L398
The documentation (https://www.bitcraze.io/docs/lps-node-f ... ter/index/) also explains how it has been implemented.
You can take a look at it and see if there is anything that can be improved in the code or the implementation. We are always happy to hear suggestions from the community about the LPS system, since we also really want it to have improved accuracy. It will probably never reach the same accuracy than Lighhouse or Mocap but probably the accuracy can be increased more than it is now.
Discussions about quadcopters/multi-rotors