@Sem-Andeweg

I don't work for ModalAI but can try to help.

You need to adjust the IMU0, and IMU1 to body RPY at minimum, which stands for Roll, Pitch, Yaw.

If you look at the Seeker example at https://gitlab.com/voxl-public/voxl-sdk/utilities/voxl-mpa-tools/-/blob/master/misc_files/usr/share/modalai/extrinsic_configs/seeker_v1_voxlcam.conf#L64

you'll see the VOXL is rotated 90 degrees "up" and yawed 90 degrees right. From there the tracking camera is then rotated "down" 45 degrees to account for the VOXL CAM being mounted on the front of the Seeker and its orientation relative to the VOXL inside the camera housing.

Hope this helps a bit, definitely watch the video at https://docs.modalai.com/configure-extrinsics/

Is this a unit direct from ModalAI? If it is, it should be tuned or you could load default parameters from https://gitlab.com/voxl-public/flight-core-px4/px4-parameters/-/tree/master/platforms for your version of PX4 and platform.

Then from there you load a helper through QGroundControl parameters page: https://gitlab.com/voxl-public/flight-core-px4/px4-parameters/-/tree/master/helpers

That should get you close to a decent tune.

If this is a custom drone, you need to configure your extrinsics, calibrate your camera, and validate QVIO first before attempting to fly which it looks like you are attempting to do. The VOXL portal qVIO overlay will show you points selected as well as distance traveled which you can validate against manually moving the drone around prior to any attempts to fly. The camera needs to have areas of contrasting features in view -- so reflective floors or some repetitive patterns can be a challenge for qVIO.

https://docs.modalai.com/configure-extrinsics/

https://docs.modalai.com/calibrate-cameras/

Late to the party on this -- but I have been able to build ipk's off of an M1 recently.

I followed: https://docs.modalai.com/install-voxl-docker/

And then also downloaded the image from the developer portal as laid out: https://docs.modalai.com/voxl-cross/

To load:
docker load --input voxl-cross_V1.8.tar

To run:
voxl-docker -i voxl-cross:V1.8

Then the docs in all of the Modal repo's work well.

I do have buildkit etc enabled as laid out here: https://medium.com/nerd-for-tech/developing-on-apple-m1-silicon-with-virtual-environments-4f5f0765fd2f but you shouldn't need any of that. Let me know if that helps you. Worked just fine for me -- no special rosetta2 terminal or Docker version.

I'll see if I can contribute something to the docs when I have a moment.

https://gitlab.com/voxl-public/voxl-sdk/services/voxl-portal/-/blob/master/src/video_manager.cpp#L77

Is an example of allocating a buffer and copying the frame into it. There are some other good examples in that file as well for processing frames and converting them to various formats within the callback. Hope that helps!

https://docs.modalai.com/setup-ros-on-voxl-2/

May help you with setting up ROS on VOXL2.

https://gitlab.com/voxl-public/voxl-sdk/utilities/voxl-mpa-to-ros may also help you bridge the Modal Pipe Architecture into ROS.

@rohitpillai

VOXL2 supports many camera configurations including front and rear stereo, a hires camera, and a tracking camera.

The drone development kits for VOXL2 such as the Sentinel (https://www.modalai.com/products/sentinel-development-drone?variant=40446488805427) and VOXL2 Flight Deck include the tracking camera at 45 degrees. You could adapt this to point straight down.

You could also put any camera on a gimbal or statically using the MIPI connectors for these "standard" cameras ModalAI provides (https://docs.modalai.com/voxl2-camera-configs/) or even interface over USB. https://www.modalai.com/pages/voxl-cameras

https://forum.modalai.com/topic/1328/camera-servers-docker-build-royale-331-spectre-4-7 may have some answers for you on voxl-camera-server which is something ModalAI is working on WRT the build.

In case anyone else comes on this thread --

https://gitlab.com/voxl-public/voxl-sdk/core-libs/libmodal-pipe/-/blob/master/library/include/modal_pipe_interfaces.h#L848 defines the VIO messages within MPA.

float T_imu_wrt_vio[3]; contains the translation in x, y, z and is what is printed in the overlay within VOXL Portal.

• Positive X = Forward
• Positive Y = Left
• Positive Z = Increase in Altitude

If you then look in the Odometry messages used by PX4 or run voxl-inspect-pose these coordinates will be in FRD/NED frame of reference which is:
Positive X = Forward
Positive Y = Right
Positive Z = Decrease Altitude
(I think I've got this right -- hopefully someone from Modal can confirm)

https://gitlab.com/voxl-public/voxl-sdk/services/voxl-vision-px4/-/blob/master/src/vio_manager.c#L350-393 walks through the publication of the odometry MAVLINK message to PX4 and rotation into appropriate frame for processing.

========
TLDR; Most of my issues for this platform were solved by disabling the QVIO server, keeping voxl-vision-px4 running, and dialing in PID for PX4 without any external inputs from vision or GPS. Once tuned, you can focus on calibrating cameras and etc/modalai/extrinsics.conf to ensure consistency in VIO inputs.

Shoutout to @wilkinsaf for tips and tricks on tuning.

All of the files for the demonstration TF Lite server models live in: https://gitlab.com/voxl-public/voxl-sdk/services/voxl-tflite-server/-/tree/master/misc_files/usr/bin/dnn

On the VOXL itself they reside at: /usr/bin/dnn

You can add your specific model and execute voxl-configure-tflite-server to point to the new model.
https://gitlab.com/voxl-public/voxl-sdk/services/voxl-tflite-server/-/blob/master/src/main.cpp#L238 shows an example with YoloV5.

I don't work for Modal but,
Questions 1,2,3:

I recommend checking out https://docs.modalai.com/voxl-tflite-server/#voxl-2 and following along with deploying the TF Lite server on your VOXL2.

The source is available here https://gitlab.com/voxl-public/voxl-sdk/services/voxl-tflite-server and includes many built-in models.

Questions 4:
Yes for both cameras, but likely will need some adaptation particularly for thermal.

Resurrecting this topic. We have two drones with identical hardware builds.
Only camera's plugged in are tracking cameras to J4.
Both are running the latest system image.

Drone 1's camera server never crashes while flying Indoors/outdoors.
Drone 2's camera server sometimes crashes and seems to be triggered by exposure changes when outside in bright environments. It is running a slightly newer version of voxl-camera-server.

The failing drone has this in the logs from logcat | grep "camera":

voxl:~$ logcat | grep camera


01-26 20:06:14.004  5732  5732 V HAL     : loaded HAL id=camera path=/usr/lib/hw/camera.msm8996.so hmi=(nil) handle=0x64450
01-26 20:06:14.005  5732  5732 I QCamera : <MCI><INFO> get_num_of_cameras: 2767: dev_info[id=0,name='video2']
01-26 20:06:14.005  5732  5732 I QCamera : <MCI><INFO> sort_camera_info: 2611: Camera id: 0 facing: 1, type: 1 is_yuv: 0
01-26 20:06:14.005  5732  5732 I QCamera : <MCI><INFO> sort_camera_info: 2617: Number of cameras 1 sorted 1
01-26 20:06:14.005  5732  5732 I QCamera : <MCI><INFO> get_num_of_cameras: 2785: num_cameras=1
01-26 20:06:14.006  5732  5732 I QCamera : <HAL><INFO> QCamera2Factory: 142: 1 camera devices detected!
01-26 20:06:14.107  5732  5732 E QCamera : <HAL><ERROR> initStaticMetadata: 9846: static int qcamera::QCamera3HardwareInterface::initStaticMetadata(uint32_t)@9846: from gCamCapability QCAMERA3_EXP_TIME_RANGE= (19425, 1020461640)
01-26 20:06:14.107  5732  5732 E QCamera : <HAL><ERROR> initStaticMetadata: 9877: static int qcamera::QCamera3HardwareInterface::initStaticMetadata(uint32_t)@9877: cameraId 0 raw_dim[0].w 640 raw_dim[0].h 480
01-26 20:06:14.107  5732  5732 I QCamera : <HAL><INFO> getCamInfo: 10272: camera 0 resource cost is 100
01-26 20:06:14.107  5732  5732 I QCamera : <HAL><INFO> cameraDeviceOpen: 414: Open camera id 0 API version 768
01-26 20:06:14.117  5732  5732 I QCamera : <HAL><INFO> openCamera: 851: [KPI Perf]: E PROFILE_OPEN_CAMERA camera id 0
01-26 20:06:14.197  5732  5732 I QCamera : <HAL><INFO> openCamera: 861: [KPI Perf]: X PROFILE_OPEN_CAMERA camera id 0, rc: 0
01-26 20:06:14.197  5732  5732 E QCamera : <HAL><ERROR> initParameters: 8374: int qcamera::QCamera3HardwareInterface::initParameters()@8374 HAL params are initialized


01-26 20:07:55.283  5732  5789 E QCamera : <MCI><ERROR> mm_camera_event_notify: 192: Camera Event DAEMON DIED received
01-26 20:07:55.284  5732  5788 E QCamera : <HAL><ERROR> camEvtHandle: 803: Fatal, camera daemon died

journalctl on the drone with crashed voxl-camera-server shows:

Jan 01 00:00:52 apq8096 systemd[1]: Started voxl-camera-server.
Jan 26 20:01:17 apq8096 bash[2966]: ERROR:   Recieved "Buffer" error from camera: tracking
Jan 26 20:01:23 apq8096 bash[2966]: ERROR:   Recieved "Buffer" error from camera: tracking
Jan 26 20:01:24 apq8096 bash[2966]: ERROR:   Recieved "Buffer" error from camera: tracking
Jan 26 20:01:35 apq8096 bash[2966]: ERROR:   Recieved "Buffer" error from camera: tracking
Jan 26 20:01:37 apq8096 bash[2966]: ERROR:   Recieved "Buffer" error from camera: tracking
Jan 26 20:01:42 apq8096 bash[2966]: ERROR:   Recieved Fatal error from camera: tracking
Jan 26 20:01:42 apq8096 bash[2966]: ERROR:   Sending request 7188, ErrorCode: -ENODEV
Jan 26 20:01:42 apq8096 bash[2966]: WARNING: Thread: tracking request thread recieved ESTOP
Jan 26 20:01:42 apq8096 bash[2966]: WARNING: Thread: tracking result thread recieved ESTOP
Jan 26 20:01:42 apq8096 bash[2966]: ------ voxl-camera-server INFO: Camera server is now stopping
Jan 26 20:01:42 apq8096 bash[2966]: WARNING: Deleting buffers: 24 of 32 still in use
Jan 26 20:01:42 apq8096 bash[2966]: Segmentation fault:
Jan 26 20:01:42 apq8096 bash[2966]: Fault thread: voxl-camera-ser(tid: 2966)
Jan 26 20:01:42 apq8096 bash[2966]: Fault address: 0x8
Jan 26 20:01:42 apq8096 bash[2966]: Address not mapped.
Jan 26 20:01:43 apq8096 systemd[1]: [[1;39mvoxl-camera-server.service: Main process exited, code=killed, status=11/SEGV[[0m
Jan 26 20:01:43 apq8096 systemd[1]: [[1;39mvoxl-camera-server.service: Unit entered failed state.[[0m
Jan 26 20:01:43 apq8096 systemd[1]: [[1;39mvoxl-camera-server.service: Failed with result 'signal'.[[0m

They have not been crashed... and the failure happens regularly when in bright environments. I matched up software versions as well and the crash still happens, so it has to be some kind of hardware issue and the below can be disregarded.

Hardware problems again with expansion or something weird in this specific camera/cable?

Drone 1 package versions (a bit older/stable repo):

libvoxl-cci-direct - 0.1.5
libvoxl-cutils - 0.1.1
voxl-camera-calibration - 0.2.3
voxl-camera-server - 1.3.5
voxl-cpu-monitor - 0.3.0
voxl-docker-support - 1.2.4
voxl-gphoto2-server - 0.0.10
voxl-jpeg-turbo - 2.1.3-4
voxl-libgphoto2 - 0.0.4
voxl-libuvc - 1.0.7
voxl-logger - 0.3.4
voxl-mavlink - 0.1.0
voxl-mavlink-server - 0.2.0
voxl-modem - 0.16.1
voxl-mongoose - 7.7.0-1
voxl-mpa-to-ros - 0.3.6
voxl-mpa-tools - 0.7.6
voxl-opencv - 4.5.5-1
voxl-portal - 0.4.2
voxl-qvio-server - 0.9.0
voxl-streamer - 0.4.1
voxl-suite - 0.9.4
voxl-system-image - 4.0-r0
voxl-tag-detector - 0.0.4
voxl-tflite-server - 0.3.1
voxl-utils - 1.2.2
voxl-uvc-server - 0.1.3
voxl-vision-px4 - 1.4.0
voxl-vpn - 0.0.6

Drone 2 package versions (a bit newer, dev repos):

libvoxl-cci-direct - 0.1.5
libvoxl-cutils - 0.1.1
voxl-camera-calibration - 0.3.0
voxl-camera-server - 1.3.7
voxl-cpu-monitor - 0.4.1
voxl-docker-support - 1.2.5
voxl-gphoto2-server - 0.0.10
voxl-jpeg-turbo - 2.1.3-4
voxl-lepton-server - 1.0.1
voxl-libgphoto2 - 0.0.4
voxl-libuvc - 1.0.7
voxl-logger - 0.3.4
voxl-mavlink - 0.1.0
voxl-mavlink-server - 0.2.0
voxl-modem - 0.16.2
voxl-mongoose - 7.7.0-1
voxl-mpa-to-ros - 0.3.6
voxl-mpa-tools - 0.8.4
voxl-opencv - 4.5.5-1
voxl-portal - 0.4.3
voxl-qvio-server - 0.9.0
voxl-streamer - 0.4.3
voxl-suite - 1.0.0
voxl-system-image - 4.0-r0
voxl-tag-detector - 0.0.4
voxl-tflite-server - 0.3.1
voxl-utils - 1.2.3
voxl-uvc-server - 0.1.3
voxl-vision-px4 - 1.4.1
voxl-vpn - 0.0.6

https://github.com/PX4/PX4-Autopilot/issues/20950

@Sem-Andeweg Just cross-posting here too for anyone that comes back to it. I'm interested in these issues as well.

Sem, (I don't work for modal!)

A few ideas..

Have you tried disabling the QVIO service and adjusting the EKF2_AID_MASK values to ignore the vision inputs across the modes you've tested with PX4?

See how the platform behaves and that should either point to VIO or rule it out. In stabilized mode and without GPS and VIO inputs the platform should be relatively stable.

Is it possible to run High-Resolution + TOF + Stereo on the VOXL? Very familiar with the configs at https://docs.modalai.com/voxl-camera-config/

Apologies for a bit of a noob question -- just tried it out and getting a segfault when J4 is set to stereo.
If it doesn't work now, that's fine just curious if it is fixable in software or a different port config before we dive in too far.

Thanks!

I don't work for Modal but,
Questions 1,2,3:

I recommend checking out https://docs.modalai.com/voxl-tflite-server/#voxl-2 and following along with deploying the TF Lite server on your VOXL2.

The source is available here https://gitlab.com/voxl-public/voxl-sdk/services/voxl-tflite-server and includes many built-in models.

Questions 4:
Yes for both cameras, but likely will need some adaptation particularly for thermal.

I do not work for ModalAI, but will attempt to answer.

The V2 ESC supports serial and PWM input. https://docs.modalai.com/modal-escs/
I am unsure if it supports DSHOT but you could configure it to use PWM. That said, if you're using a VOXL the serial connection is a better option.

As far enabling DShot in the firmware, you need to make sure the uart is disabled by setting UART_ESC_CONFIG to disabled. Then I believe the option for DShot/PWM will show up.
Check out https://docs.modalai.com/modal-esc-px4-user-guide/ for more info.

https://www.modalai.com/products/voxl2-io?variant=40610514206771 for generating PWM from the VOXL2 to command a larger 4-1 ESC.

I don't work for ModalAI, but a few quick questions I have to try to help:

VOXL, VOXL Flight or VOXL2?

What port do you have your PX4 Autopilot/Flight Core plugged into if running a VOXL?

I recommend running a voxl-configure-mpa and picking one of the options close to your platform or factory reset to configure the mavlink connection via UART for your platform. https://docs.modalai.com/voxl-configure-mpa/

I would also check /etc/modalai/voxl-mavlink-server.conf and make sure the telemetry port is set up. voxl-configure-mpa should do this for you, but just in case you should see something like:

/**
 * voxl-mavlink-server Configuration File
 * UART fields are for APQ8096 VOXL1 or QRB5165 VOXL2 with external fc
 * UDP fields are for PX4 on SDSP on QRB5165 only
 * External FC field is for QRB5165 only. Set to true to enable UART
 * communication to an external flight controller, otherwise a UDP interface
 * will be started to talk to voxl-px4 on localhost which is the default behavior.
 *
 */
{
	"px4_uart_bus":	5,
	"px4_uart_baudrate":	921600,
	"udp_port_to_px4":	14556,
	"udp_port_from_px4":	14557,
	"external_fc":	false
}

If you're running a VOXL with Flight Core, the connection is documented here: https://docs.modalai.com/how-to-connect-voxl-to-flight-core/

And the UART bus should be port 5 as shown in the config file above.

If you're using a Flight core, do you have blinking lights on it? Should be solid green, blinking blue.

I also recommend running image 3.9.0 based on our experience if you can make the upgrade.

Hope this helps!

With a fresh install on one of our platforms of System Image 4.0.0, the camera server crashes and QVIO becomes unresponsive. The system is configured with Stereo, Tracking, and TOF.

voxl-camera-server:

Jan 01 00:00:07 apq8096 systemd[1]: Started voxl-camera-server.
Jan 01 00:00:10 apq8096 bash[2543]: ------ voxl-camera-server: Camera server is now running
Nov 16 21:25:39 apq8096 bash[2543]: ERROR:   Recieved Fatal error from camera: tof
Nov 16 21:25:39 apq8096 bash[2543]: ERROR:   Sending request 397, ErrorCode: -ENODEV
Nov 16 21:25:39 apq8096 bash[2543]: WARNING: Thread: tracking result thread recieved ESTOP
Nov 16 21:25:39 apq8096 bash[2543]: WARNING: Thread: tof request thread recieved ESTOP
Nov 16 21:25:39 apq8096 bash[2543]: WARNING: Thread: stereo_child result thread recieved ESTOP
Nov 16 21:25:39 apq8096 bash[2543]: WARNING: Thread: tof result thread recieved ESTOP
Nov 16 21:25:39 apq8096 bash[2543]: WARNING: Thread: stereo result thread recieved ESTOP
Nov 16 21:25:39 apq8096 bash[2543]: WARNING: Thread: stereo request thread recieved ESTOP
Nov 16 21:25:39 apq8096 bash[2543]: WARNING: Thread: tracking request thread recieved ESTOP
Nov 16 21:25:39 apq8096 bash[2543]: WARNING: Thread: stereo_child request thread recieved ESTOP
Nov 16 21:25:39 apq8096 bash[2543]: ------ voxl-camera-server INFO: Camera server is now stopping
Nov 16 21:25:39 apq8096 bash[2543]: WARNING: Deleting buffers: 3 of 16 still in use
Nov 16 21:25:40 apq8096 bash[2543]: WARNING: Deleting buffers: 5 of 16 still in use
Nov 16 21:25:47 apq8096 bash[2543]: WARNING: Deleting buffers: 1179656 of 1179668 still in use
Nov 16 21:25:47 apq8096 bash[2543]: Segmentation fault:
Nov 16 21:25:47 apq8096 bash[2543]: Fault thread: voxl-camera-ser(tid: 2543)
Nov 16 21:25:47 apq8096 bash[2543]: Fault address: 0xa
Nov 16 21:25:47 apq8096 bash[2543]: Address not mapped.
Nov 16 21:25:49 apq8096 systemd[1]: [[1;39mvoxl-camera-server.service: Main process exited, code=killed, status=11/SEGV[[0m
Nov 16 21:25:49 apq8096 systemd[1]: [[1;39mvoxl-camera-server.service: Unit entered failed state.[[0m
Nov 16 21:25:49 apq8096 systemd[1]: [[1;39mvoxl-camera-server.service: Failed with result 'signal'.[[0m

voxl-qvio-server:

-- Logs begin at Thu 1970-01-01 00:00:04 UTC, end at Wed 2022-11-16 21:29:40 UTC. --
Jan 01 00:00:07 apq8096 systemd[1]: Started voxl-qvio-server.
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: setting scheduler
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: INFO: set FIFO priority successfully!
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: loading config file
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: =================================================================
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: imu_name:                         imu0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: cam_name:                         tracking
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: lens_cal_path:                    /data/modalai/opencv_tracking_intrinsics.yml
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: cam_extrinsics_name:              tracking
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: use_second_cam_if_stereo:         0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: odr_hz:                           30.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: use_camera_height_bootstrap:      1
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: camera_height_off_ground_m:        0.085
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: enable_init_while_moving:         0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: cam_imu_timeshift_s:              -0.002
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: cam_imu_timeshift_s_uncertainty:   0.001
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: T_cam_wrt_imu_uncertainty:         0.001  0.001  0.001
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: R_cam_to_imu_uncertainty:          0.005  0.005  0.005
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: accl_fsr_ms2:                     156.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: gyro_fsr_rad:                     34.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: accl_noise_std_dev:                0.316
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: gyro_noise_std_dev:                0.010
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: cam_noise_std_dev:                100.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: min_std_pixel_noise:               0.500
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: fail_high_pixel_noise_points:      1.665
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: limited_imu_bw_trigger:           25.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: gps_imu_time_alignment_s:          0.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: T_gps_wrt_imu:                     0.000  0.000  0.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: enable_mapping:                   1
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: enable_gps_vel:                   0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: show_extra_points_on_overlay;     0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: auto-reset params:
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: en_auto_reset:                    1
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: auto_reset_max_velocity:          10.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: auto_reset_max_v_cov_instant:      0.100
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: auto_reset_max_v_cov:              0.010
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: auto_reset_max_v_cov_timeout_s:    0.500
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: auto_reset_min_features:          3
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: auto_reset_min_feature_timeout_s:  1.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: =================================================================
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: loading extrinsics config file
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: #0:
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: parent:                imu0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: child:                 tracking
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: T_child_wrt_parent:    0.109  -0.012   0.001
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: RPY_parent_to_child: 180.0    45.0    90.0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: R_child_to_parent:     0.000  -0.707   0.707
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: -1.000  -0.000  -0.000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: 0.000  -0.707  -0.707
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: tbc:   0.109 -0.012  0.001
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: ombc:  2.504 -2.504  1.037 (axis angle)
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: ombc: 180.000 45.000 90.000 (RPY deg)
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: loading /data/modalai/opencv_tracking_intrinsics.yml
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: using fisheye camera intrinsics:
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: distortion: 0.008810 -0.003841 0.000000 0.000000
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: focal lengths: 269.295927 268.516850
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: principle points: 340.767010 221.351666
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: WARNING failed to init in time, trying again
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: LIBMV: AR INFO:
                                                FIT:FeatureName=>SensorFusion, TSBDrift=>0.00059 (0.004)
                                                FIT:FeatureName=>SensorFusion, TSBDrift0=>-0.000 (0.003)
                                                FIT:FeatureName=>SensorFusion, TSBDrift1=>0.000 (0.003)
                                                FIT:FeatureName=>SensorFusion, TSBDrift2=>-0.000 (0.002)
                                                FIT:FeatureName=>SensorFusion, RSBDrift=>0.00062
                                                FIT:FeatureName=>SensorFusion, AccelBias0=>-0.00278
                                                FIT:FeatureName=>SensorFusion, AccelBias1=>-0.00121
                                                FIT:FeatureName=>SensorFusion, AccelBias2=>0.12533
                                                FIT:FeatureName=>SensorFusion, GyroBias0=>-0.00470
                                                FIT:FeatureName=>SensorFusion, GyroBias1=>-0.00426
                                                FIT:FeatureName=>SensorFusion, GyroBias2=>-0.00017
                                                FIT:FeatureName=>SensorFusion, GravityBias0=>0.30783
                                                FIT:FeatureName=>SensorFusion, GravityBias1=>0.16675
                                                FIT:FeatureName=>SensorFusion, GravityBias2=>-9.78417
                                                FIT:FeatureName=>SensorFusion, FrameCount=>7
                                                FIT:FeatureName=>SensorFusion, ImuCount=>304
                                                FIT:FeatureName=>SensorFusion, AvgTimeFrameUpdate=>5.84ms
                                                FIT:FeatureName=>SensorFusion, NumResets=>0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: LIBMV: AR INFO:
                                                TrckLnDst:6 4 3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: LIBMV: AR INFO:
                                                Distribution of num of consec keyframes w/ GN failing all features:0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: LIBMV: AR INFO:
                                                Distribution of num of consec frames w/ increasing depth uncertainty:2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: waiting for imu
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: waiting for cam
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: connected to imu server
Jan 01 00:00:10 apq8096 voxl-qvio-server[2539]: connected to camera server
Nov 16 21:25:48 apq8096 voxl-qvio-server[2539]: LIBMV: AR INFO:
                                                FIT:FeatureName=>SensorFusion, TSBDrift=>0.76085 (0.011)
                                                FIT:FeatureName=>SensorFusion, TSBDrift0=>0.596 (0.008)
                                                FIT:FeatureName=>SensorFusion, TSBDrift1=>-0.458 (0.006)
                                                FIT:FeatureName=>SensorFusion, TSBDrift2=>-0.114 (0.005)
                                                FIT:FeatureName=>SensorFusion, RSBDrift=>0.30760
                                                FIT:FeatureName=>SensorFusion, AccelBias0=>0.12854
                                                FIT:FeatureName=>SensorFusion, AccelBias1=>0.19535
                                                FIT:FeatureName=>SensorFusion, AccelBias2=>-0.10591
                                                FIT:FeatureName=>SensorFusion, GyroBias0=>-0.00679
                                                FIT:FeatureName=>SensorFusion, GyroBias1=>-0.00544
                                                FIT:FeatureName=>SensorFusion, GyroBias2=>-0.00180
                                                FIT:FeatureName=>SensorFusion, GravityBias0=>0.49261
                                                FIT:FeatureName=>SensorFusion, GravityBias1=>0.21229
                                                FIT:FeatureName=>SensorFusion, GravityBias2=>-9.78607
                                                FIT:FeatureName=>SensorFusion, FrameCount=>2375
                                                FIT:FeatureName=>SensorFusion, ImuCount=>180880
                                                FIT:FeatureName=>SensorFusion, AvgTimeFrameUpdate=>178.39ms
                                                FIT:FeatureName=>SensorFusion, NumResets=>1
Nov 16 21:25:48 apq8096 voxl-qvio-server[2539]: LIBMV: AR INFO:
                                                TrckLnDst:182 46 36 69 45 37 32 43 20 22 21 16 16 11 8 11 4 8 7 2 4 2 6 4 2 2 5 3 2 2 3 0 2 3 4 1 1 1 4 1 2 1 0 1 5 1 5 1 0 3 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 1 0 1 0 1 0 1 1 2 0 1 2 1 1 0 0 1 0 0 0 2 0 0 0 2 0 1 0 1 0 0 0 1 0 0 1 0 0 1 1 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 108
Nov 16 21:25:48 apq8096 voxl-qvio-server[2539]: LIBMV: AR INFO:
                                                Distribution of num of consec keyframes w/ GN failing all features:42 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Nov 16 21:25:48 apq8096 voxl-qvio-server[2539]: LIBMV: AR INFO:
                                                Distribution of num of consec frames w/ increasing depth uncertainty:2354 15 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Nov 16 21:25:48 apq8096 voxl-qvio-server[2539]: WARNING: disconnected from camera server, resetting VIO
Nov 16 21:25:48 apq8096 voxl-qvio-server[2539]: client "voxl-portal0" connected to overlay

Here are our versions etc.

Service Name         |  Version  |  Enabled  |   Running   |  CPU Usage
---------------------------------------------------------------------------
 docker-autorun       |    1.2.4  | Disabled  | Not Running |
 docker-daemon        |    1.2.4  | Disabled  | Not Running |
 modallink-relink     |   0.16.1  | Disabled  | Not Running |
 voxl-camera-server   |    1.3.5  |  Enabled  |   Running   |    12.7
 voxl-cpu-monitor     |    0.3.0  |  Enabled  |   Running   |     0.0
 voxl-dfs-server      |    0.3.1  | Disabled  | Not Running |
 voxl-imu-server      |    1.0.3  |  Enabled  |   Running   |     0.0
 voxl-mavlink-server  |    0.2.0  |  Enabled  |   Running   |     1.8
 voxl-modem           |   0.16.1  | Disabled  | Not Running |
 voxl-portal          |    0.4.2  |  Enabled  |   Running   |     0.0
 voxl-qvio-server     |    0.8.2  |  Enabled  |   Running   |     9.1
 voxl-streamer        |    0.4.1  | Disabled  | Not Running |
 voxl-tag-detector    |    0.0.4  | Disabled  | Not Running |
 voxl-tflite-server   |    0.3.1  | Disabled  | Not Running |
 voxl-time-sync       |    1.2.2  | Disabled  | Not Running |
 voxl-vision-px4      |    1.4.0  |  Enabled  |   Running   |     0.0
 voxl-wait-for-fs     |    1.2.2  |  Enabled  |  Completed  |

This happens with or without the portal. Other units seem OK and this unit was OK in a different environment. Trying to isolate the issue.

I recommend following: https://docs.modalai.com/voxl-vision-px4-collision-prevention/ as a starting point before attempting any custom integrations.

And updating to the latest system image build from https://docs.modalai.com/platform-releases/ which is 4.0.0.

All of the files for the demonstration TF Lite server models live in: https://gitlab.com/voxl-public/voxl-sdk/services/voxl-tflite-server/-/tree/master/misc_files/usr/bin/dnn

On the VOXL itself they reside at: /usr/bin/dnn

You can add your specific model and execute voxl-configure-tflite-server to point to the new model.
https://gitlab.com/voxl-public/voxl-sdk/services/voxl-tflite-server/-/blob/master/src/main.cpp#L238 shows an example with YoloV5.