EKF filter : give priority to a sensor
Hello,
Context : outdoor navigation with GPS, IMU and VO (visual odometry). The GPS is really accurate. Issue : The GPS covariance (provided by a node from the gps provider) is quite high relative to the covariance of the VO. Consequently, it looks like the Global EKF doesn't consider the GPS at all. Nevertheless, this is the most accurate sensor (the VO drifts a lot because of the space the robot works in).
Question : I was wondering if :
- it was possible to scale the covariances directly in the EKF ?
- it was possible to give some kind of priority to a sensor ?
- I should create a "covariance rescaler" node ?
Or maybe you have another solution/suggestion ?
Thanks in advance !