Replace Const acc. model with CATR model
Description
@niosus @JWhitleyWork After discussions with @niosus, @nikolai.morin and I realized that the issue of the yaw spinning around when the ego vehicle is stationary is because the parametrization of the motion model is susceptible to noise. This problem should disappear when we choose polar coordinates and separate velocity into speed and yaw.
Expected behavior
The yaw estimate is stable when the ego is stationary
Definition of Done
-
const acceleration model is replaced by catr model in state estimation -
manual test with lgsvl shows expected behavior
Edited by Frederik Beaujean