How to detect the drift of the odometry?

[Lockwuud]

Thank you very much for your work. I am currently using Point-LIO as the odometry for the steering wheel chassis, which has multiple odometry units, with Point-LIO being one of them. I would now like to estimate the credibility of the odometry's coordinate data. I’ve tried using the covariance matrix from Kalman filtering to measure this, but I haven’t fully understood how Kalman filtering is implemented in the code. Could you recommend a good variable to detect the drift of the odometry? Specifically, could you point out which variable in the code can represent this level of drift?