You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for your so wonderful project and share. I want to conduct your experiment but I face a localization problem really confusing me. When I use VINS Fusion and estimator td is 1 with initial value -0.01,td is always changed and cannot be stable at a constant. Does it mean VINS Fusion is not convergence? And how to make td stable? I use 200hz pixhawk imu and 30hz d435 infra stereo images.
I ask this question because when I want to hover my drone for a long time, this unstable td may suddenly change, causing VINS pos and vel estimator suddenly change in z axis. Then onboard controller becomes unstable and the drone vibrates. In this situation, imu output from fcu contains large noise which further deteriorates VINS estimator and induces more vibration and results in crash. I want to know whether there is a standard to confirm VIO is stable,accuracy and convergence? Or how to realise? Such as the setting of td.
The text was updated successfully, but these errors were encountered:
You can try to disable td estimation by setting estimate_td to 0 with the initial value of td = -0.01;
Stop estimating td will improve convergence but at the cost of a slightly lower accurancy.
Thanks for your so wonderful project and share. I want to conduct your experiment but I face a localization problem really confusing me. When I use VINS Fusion and estimator td is 1 with initial value -0.01,td is always changed and cannot be stable at a constant. Does it mean VINS Fusion is not convergence? And how to make td stable? I use 200hz pixhawk imu and 30hz d435 infra stereo images.
I ask this question because when I want to hover my drone for a long time, this unstable td may suddenly change, causing VINS pos and vel estimator suddenly change in z axis. Then onboard controller becomes unstable and the drone vibrates. In this situation, imu output from fcu contains large noise which further deteriorates VINS estimator and induces more vibration and results in crash. I want to know whether there is a standard to confirm VIO is stable,accuracy and convergence? Or how to realise? Such as the setting of td.
The text was updated successfully, but these errors were encountered: