I’ve been using the ZED 2i and trying out the different settings for pos_tracking_mode (GEN_1, GEN_2 and GEN_3) and I wanted to ask what the difference was between each of the settings?
I’m displaying a point cloud in RViz2. I’ve found that GEN_3 displays some jitters, where the camera pose can bounce around sometimes even though it is physically stationary. GEN_1 is much more reliable and aligns smoothly with the camera as I move it. I’m wondering if the jitters come exclusively from pos_tracking_mode or if its a combination of different settings (I’ve also been experimenting with depth_mode, where ULTRA and NEURAL_PLUS perform well in my setup).
I’m using the camera indoors in a well lit environment. And I’m using version 5.0.1 on a computer with NVIDIA-SMI 535.230.02 Driver Version: 535.230.02 CUDA Version: 12.2.
I recommend you upgrade to the latest v5.0.3 if you want to use GEN_3.
The new mode is in continuous evolution and each patch version brings new improvements
Great thanks I will try that.
I’m also interested to know if there’s any significant differences between the GEN’s, such as if one is better for a particular purpose than another. Or is each iteration just a general, all-round improvement on the previous generation? Thanks
GEN_1 is the first version of the positional tracking module, mostly based on visual odometry. Using the IMU only to improve the orientation estimation.
GEN_2 is the first “next-gen” algorithm with a full integration of Visual and Inertial information. It requires more compute resources than GEN_1, but it’s more reliable.
GEN_3 is the new evolution; it provides better accuracy than the other modes and requires lower performance. We expect to release a stable version with SDK v5.1.
Thanks for getting back to me on this, is helpful to know the distinction between the GENs. Is there a timeline on when the stable GEN_3 version will be available? I am able to keep using the earlier GENs for now but interested to know about GEN_3.