# Open-vins keeps diverging and re-initializing. Source: https://forum.modalai.com/topic/4485/open-vins-keeps-diverging-and-re-initializing Category: General Questions (https://forum.modalai.com/category/2/general-questions) Posted: 2025-06-05 19:44:29 UTC by Yufei Jiang Replies: 3 · Views: 965 ## Yufei Jiang · 2025-06-05 19:44:29 UTC Hi, I'm using VOXL2 with flight deck and trying to test the open-vins. Currently I'm only using one camera (tracking) and the output will diverge quickly once I slightly move the VOXL2. Is there any clue about why and how I can solve this? Below is my system: voxl2:/$ voxl-version ──────────────────────────────────────────────────────────────────────────────── system-image: 1.8.02-M0054-14.1a-perf kernel: #1 SMP PREEMPT Mon Nov 11 22:08:01 UTC 2024 4.19.125 ──────────────────────────────────────────────────────────────────────────────── hw platform: M0054 mach.var: 1.0.0 SKU: MDK-F0006-4-V2-C11-E1 ──────────────────────────────────────────────────────────────────────────────── voxl-suite: 1.4.5 ──────────────────────────────────────────────────────────────────────────────── Packages: Repo: http://voxl-packages.modalai.com/ ./dists/qrb5165/sdk-1.4/binary-arm64/ Last Updated: 2025-06-05 19:03:39 List: kernel-module-voxl-fsync-mod-4.19.125 1.0-r0 kernel-module-voxl-gpio-mod-4.19.125 1.0-r0 kernel-module-voxl-platform-mod-4.19.125 1.0-r0 libfc-sensor 1.0.7 libmodal-cv 0.5.18 libmodal-exposure 0.1.4 libmodal-journal 0.2.5 libmodal-json 0.4.7 libmodal-pipe 2.13.1 libqrb5165-io 0.5.0 libvoxl-cci-direct 0.3.2 libvoxl-cutils 0.1.5 modalai-slpi 1.1.20 mv-voxl 0.1-r0 qrb5165-bind 0.1-r0 qrb5165-dfs-server 0.2.0 qrb5165-imu-server 1.1.3 qrb5165-rangefinder-server 0.1.5 qrb5165-slpi-test-sig 01-r0 qrb5165-system-tweaks 0.3.5 qrb5165-tflite 2.8.0-2 voxl-bind-spektrum 0.1.1 voxl-camera-calibration 0.5.9 voxl-camera-server 2.2.1 voxl-ceres-solver 2:1.14.0-10 voxl-configurator 1.0.1 voxl-cpu-monitor 0.6.0 voxl-docker-support 1.3.1 voxl-elrs 0.4.7 voxl-esc 1.5.2 voxl-feature-tracker 0.5.2 voxl-flow-server 0.3.6 voxl-fsync-mod 1.0-r0 voxl-gphoto2-server 0.0.10 voxl-gpio-mod 1.0-r0 voxl-io-server 0.0.5 voxl-jpeg-turbo 2.1.3-7 voxl-lepton-server 1.3.3 voxl-lepton-tracker 0.0.4 voxl-libgphoto2 0.0.4 voxl-libuvc 1.0.7 voxl-logger 0.5.3 voxl-mavcam-manager 0.5.8 voxl-mavlink 0.1.5 voxl-mavlink-server 1.4.7 voxl-modem 1.1.6 voxl-mongoose 7.7.0-2 voxl-mpa-to-ros 0.3.9 voxl-mpa-tools 1.4.0 voxl-open-vins 0.4.19 voxl-open-vins-server 0.3.12 voxl-opencv 4.5.5-3 voxl-osd 0.1.7 voxl-platform-mod 1.0-r0 voxl-portal 0.7.10 voxl-px4 1.14.0-2.0.98 voxl-px4-imu-server 0.1.2 voxl-px4-params 0.7.3 voxl-qvio-server 1.2.0 voxl-remote-id 0.0.9 voxl-reset-slpi 0.0.1 voxl-state-estimator 0.0.5 voxl-streamer 0.7.5 voxl-suite 1.4.5 voxl-tag-detector 0.0.4 voxl-tflite-server 0.4.1 voxl-utils 1.4.8 voxl-uvc-server 0.1.7 voxl-vision-hub 1.8.21 voxl-vtx 1.3.14 voxl-wavemux 0.0.1 voxl2-io 0.0.3 voxl2-system-image 1.8.02-r0 voxl2-wlan 1.0-r0 ──────────────────────────────────────────────────────────────────────────────── voxl-open-vins-server.conf: { "en_auto_reset": true, "auto_reset_max_velocity": 20, "auto_reset_max_v_cov_instant": 0.10000000149011612, "auto_reset_max_v_cov": 0.10000000149011612, "auto_reset_max_v_cov_timeout_s": 0.5, "auto_reset_min_features": 1, "auto_reset_min_feature_timeout_s": 3, "auto_fallback_timeout_s": 3, "auto_fallback_min_v": 0.600000023841858, "en_cont_yaw_checks": false, "fast_yaw_thresh": 5, "fast_yaw_timeout_s": 1.75, "do_fej": true, "imu_avg": true, "use_rk4_integration": true, "cam_to_imu_refinement": true, "cam_intrins_refinement": true, "cam_imu_ts_refinement": true, "max_clone_size": 8, "max_slam_features": 35, "max_slam_in_update": 10, "max_msckf_in_update": 10, "feat_rep_msckf": 4, "feat_rep_slam": 4, "cam_imu_time_offset": 0, "slam_delay": 1, "gravity_mag": 9.80665, "init_window_time": 0.5, "init_imu_thresh": 2, "imu_sigma_w": 0.00013990944749616306, "imu_sigma_wb": 4.1189724174615527e-07, "imu_sigma_a": 0.0038947538150776763, "imu_sigma_ab": 5.538346201712153e-05, "msckf_chi2_multiplier": 1, "slam_chi2_multiplier": 40, "zupt_chi2_multiplier": 1, "refine_features": true, "pyr_levels": 6, "grid_x": 5, "grid_y": 5, "msckf_sigma_px": 1, "slam_sigma_px": 1.8, "zupt_sigma_px": 1, "try_zupt": true, "zupt_max_velocity": 0.03, "zupt_only_at_beginning": true, "zupt_noise_multiplier": 1, "zupt_max_disparity": 8, "init_dyn_use": false, "triangulate_1d": false, "max_runs": 5, "init_lamda": 0.001, "max_lamda": 10000000000, "min_dx": 1e-06, "min_dcost": 1e-06, "lam_mult": 10, "min_dist": 0.1, "max_dist": 60, "max_baseline": 40, "max_cond_number": 600000, "use_mask": false, "use_stereo": false, "use_baro": false, "num_opencv_threads": 4, "fast_threshold": 15, "histogram_method": 1, "knn_ratio": 0.7, "takeoff_accel_threshold": 0.5, "takeoff_threshold": -0.1, "use_stats": false, "max_allowable_cep": 1, "en_force_init": false, "en_force_ned_2_flu": false, "en_imu_frame_output": true, "track_frequency": 15, "publish_frequency": 5, "en_vio_always_on": true, "en_ext_feature_tracker": false, "en_gpu_for_tracking": true, "num_features_to_track": 10, "raansac_gn": false, "raansac_tri": true, "en_thermal_enhance": false, "en_overlay_landscape": false, "thermal_brightness": 1, "thermal_brightness_bos": 1 Output of open-vins-server: voxl2:/etc/modalai$ voxl-open-vins-server Setting printing level to: SILENT ================================================================= =========================AUTO RESET============================== en_auto_reset: 1 auto_reset_max_velocity: 20.000 auto_reset_max_v_cov_instant: 0.100 auto_reset_max_v_cov: 0.100 auto_reset_max_v_cov_timeout_s: 0.500 auto_reset_min_features: 1 auto_reset_min_feature_timeout_s: 3.000 auto_fallback_timeout_s: 3.000 auto_fallback_min_v: 0.600 en_cont_yaw_checks: false fast_yaw_thresh: 5.000 fast_yaw_timeout_s: 1.750 ===========================STATE================================= do fej: true imu avg: true use rk4 integration: true cam to imu refinement: true cam intrins refinement: true cam imu ts refinement: true max clone size: 8 max slam features: 35 max slam in update: 10 max msckf in update: 10 feat rep msckf: ANCHORED_MSCKF_INVERSE_DEPTH feat rep slam: ANCHORED_MSCKF_INVERSE_DEPTH cam imu time offset: 0.00000 slam delay: 1.00000 pyr levels: 6 grid x: 5 grid y: 5 ================================================================= =====================INERTIAL INITIALIZER======================== gravity mag: 9.80665 init window time: 0.50000 init imu thresh: 2.00000 init dyn use: false ================================================================= ==========================IMU NOISE============================== imu sigma w: 0.00014 imu sigma wb: 0.00000 imu sigma a: 0.00389 imu sigma ab: 0.00006 ================================================================= ========================FEATURE NOISE============================ msckf chi^2 multiplier: 1.00000 msckf sigma px: 1.00000 slam chi^2 multiplier: 40.00000 slam sigma px: 1.80000 zupt chi^2 multiplier: 1.00000 zupt sigma px: 1.00000 ================================================================= =============================ZUPT================================ refine features: true max runs: 5 try zupt: true en_ext_feature_tracker: 0 en_gpu_for_tracking: 1 raansac_gn: false raansac_tri: true zupt max velocity: 0.03000 zupt only at beginning: true zupt noise multiplier: 1.00000 zupt max disparity: 8.00000 ================================================================= use mask: false use stereo: false use baro: false ================================================================= ========================FEATURE INITIALIZER====================== triangulate 1d: false init lamda: 0.00100 max lamda: 10000000000.00000 min dx: 0.00000 min dcost: 0.00000 lam mult: 10.00000 min dist: 0.10000 max dist: 60.00000 max baseline: 40.00000 num_opencv_threads: 4 fast_threshold: 15 histogram_method: HISTOGRAM knn_ratio: 0.70000 track_frequency: 15.00000 publish_frequency: 5 imu takeoff accel thresh: 0.50000 force init on takeoff: false takeoff_threshold(m): -0.100000 publish stats: false max_allowable_cep: 1.00000 force FLU to NED transform: false publish IMU frame: true VIO always on (for bench testing): true num_features_to_track: 10 Enable Thermal Enhance: false Enable Overlay Landscape: false Thermal Brightness: 1.00000 Thermal Brightness BOSON: 1.00000 ================================================================= ================================================================= vio_cam config: cam 0: enable: 1 name: tracking pipe_for_preview: tracking pipe_for_tracking: tracking is_occluded_on_ground: 0 imu: imu_apps is_extrinsic_present: 1 extrinsic cam wrt imu: parent: imu_apps child: tracking T_cam_wrt_imu: 0.027 0.007 0.018 RPY_imu_to_cam: 0.0 45.0 90.0 R_cam_to_imu: 0.000 -0.707 0.707 1.000 0.000 -0.000 -0.000 0.707 0.707 cal file: opencv_tracking_intrinsics.yml is_cal_present: 1 cam cal intrinsics: width: 640 height: 480 fx: 277.659 fy: 277.763 cx: 349.269 cy: 219.049 n_coeffs: 4 is_fisheye: 1 D: -0.0240 0.0390 -0.0284 0.0077 ================================================================= vio cam: 0 is occluded: 0 [INFO] Camera: 0 -- converted extrinsics *in FLU* are: Roll: -0.000000, Pitch 135.000000, Yaw -90.000000 (-1.570796) WORLD QUATERNION: -0.000000 0.000000 -0.000000 1.000000 OpenVINS using fisheye camera OpenVINS FINAL Cam extrinsics to IMU for tracking: 7 -0.653281 0.653281 -0.270598 0.270598 0.0273 -0.0067 -0.0182 ++++++++++++++++++++++++++ OpenVINS reports number of cameras being used, regardless of delivery method (pipe, merge, etc...): 1 ++++++++++++++++++++++++++ Set camera intrinsics and extrinsics ====> Using Internal KLT feature tracker (w/ GPU) <==== initialized ocl manager >>>>>>>>>> Current reference counts: trackDB: 1 trackFEATS: 1 propagator: 2 updaterMSCKF: 1 updaterSLAM: 1 active_tracks_initializer: 1 initializer: 1 thread is locked to cores: 4 5 6 7 create_server_pipes connecting client pipes imu pipe name: imu_apps Number of Cameras active: 1 Camera merge --- > ch: 1 to cam id: 0 Opening camera pipe: tracking tmp_camera_pipe_names.push_back(): tracking client "voxl-portal66662777" connected to overlay [WARN] (0.005064 and 0.235771) Cannot initialize FRD to IMU transform--too much drift: your IMU may need to be recalibrated or is OVERHEATED! [WARN] (0.004868 and 0.241071) Cannot initialize FRD to IMU transform--too much drift: your IMU may need to be recalibrated or is OVERHEATED! [WARN] (0.003866 and 0.231757) Cannot initialize FRD to IMU transform--too much drift: your IMU may need to be recalibrated or is OVERHEATED! [INFO] world : 1.000000 1.000000 (2) [WARN] (0.003118 and 0.237088) Cannot initialize FRD to IMU transform--too much drift: your IMU may need to be recalibrated or is OVERHEATED! [WARN] (0.005283 and 0.226763) Cannot initialize FRD to IMU transform--too much drift: your IMU may need to be recalibrated or is OVERHEATED! FAILED TO FETCH IMG RINGBUF at time 0 -2 -2 -2 -2 -2 [INFO] FRD to IMU NED configuration done. g direction: -1 [INFO] gravity direction (2) : -1 [INFO] *** FRD to IMU transform READY! *** FAILED TO FETCH IMG RINGBUF at time 0 -2 -2 -2 -2 -2 waiting for enough clone states (1 of 5).... [WARN] Motion detected, going to full processing waiting for enough clone states (2 of 5).... waiting for enough clone states (3 of 5).... waiting for enough clone states (4 of 5).... [VIO_BAD_STATE] Quality less than 1. WARNING auto-resetting, Bad VIO, State: true Q: -1 Vel: 0.043736 Uncert: 0.949216 State: 0(1) [INFO] restarting managers STATELESS Set camera intrinsics and extrinsics ====> Using Internal KLT feature tracker (w/ GPU) <==== initialized ocl manager >>>>>>>>>> Current reference counts: trackDB: 1 trackFEATS: 1 propagator: 2 updaterMSCKF: 1 updaterSLAM: 1 active_tracks_initializer: 1 initializer: 1 Camera queue size after reset: 0 IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1959.206927 - 1958.867053 (0.339874) IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1959.237218 - 1958.867053 (0.370165) IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1959.282138 - 1958.867053 (0.415084) IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1959.311426 - 1958.867053 (0.444373) IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1959.336785 - 1958.867053 (0.469732) waiting for enough clone states (1 of 5).... [INFO] TAKEOFF: DISPARITY disturbed, possible motion detected. Avg: 434.788239 (8.000000) # feats: 2 (20). waiting for enough clone states (2 of 5).... [INFO] TAKEOFF: DISPARITY disturbed, possible motion detected. Avg: 0.000000 (8.000000) # feats: 2 (20). waiting for enough clone states (3 of 5).... waiting for enough clone states (4 of 5).... [WARN] Motion detected, going to full processing [VIO_BAD_STATE] Quality less than 1. WARNING auto-resetting, Bad VIO, State: true Q: -1 Vel: 7.194683 Uncert: 0.586549 State: 0(1) [INFO] restarting managers STATELESS Set camera intrinsics and extrinsics ====> Using Internal KLT feature tracker (w/ GPU) <==== initialized ocl manager >>>>>>>>>> Current reference counts: trackDB: 1 trackFEATS: 1 propagator: 2 updaterMSCKF: 1 updaterSLAM: 1 active_tracks_initializer: 1 initializer: 1 Camera queue size after reset: 0 IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1965.271825 - 1964.916372 (0.355453) IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1965.311861 - 1964.916372 (0.395488) IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1965.342133 - 1964.916372 (0.425761) IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1965.376322 - 1964.916372 (0.459950) IMU samples too small, make your window smaller (params.init_window_time), can't initalize 1965.411513 - 1964.916372 (0.495141) waiting for enough clone states (1 of 5).... ## Reply by zauberflote1 (ModalAI staff) · 2025-06-06 15:13:45 UTC Hello, Please refer to the following and see if the problem persists: https://forum.modalai.com/topic/4475/voxl-open-vins-server?_=1749222785639 Thanks, ## Reply by Yufei Jiang · 2025-06-06 20:10:57 UTC (in reply to zauberflote1) @zauberflote1 Hi, I recalibrated the imu and checked it using voxl-inspect-imu. The imu data looks good. In voxl-inspect-cam it can detect all the cameras. However, it's still diverging really fast once I moved the VOXL2. If I disabled the auto-reset of vins, it will keep drifting. ![image (5).png](https://forum.modalai.com/assets/uploads/files/1749240628426-image-5.png) ## Reply by Yufei Jiang · 2025-06-06 20:13:28 UTC (in reply to Yufei Jiang) @Yufei-Jiang And I used voxl-configure-cameras and voxl-configure-extrinsics D0006_Sentinel to generate the camera config.