@modaltb Yes you're right. Its not connecting to QGC via usb too.

It is possible to run concurrent LTE and Microhard connections on VOXL. This can be accomplished using a combination of the two modems, either with an add-on LTE modem and a standalone Microhard modem plugged in to the add-on or an add-on Microhard modem with a standalone LTE modem plugged in to the add-on. In order for both networks to start correctly, the LTE network will need to be brought up first, then the Microhard network. This can be set up easily by creating a custom startup service. The LTE network will use the wwan0 network interface while the Microhard network will use the eth0 network interface.

@Alfonso-Ponce We don't have any recommendations on how to create such a mobile application. However, whatever path you choose, you will most likely need the app to be sending Mavlink packets to the drone to carry out the instructions. Also, QGroundControl can be run as a mobile app and it is open source so you could look to that as an example.

The only lower bound I've seen is 20cm here: https://gitlab.com/voxl-public/modal-pipe-architecture/voxl-vision-px4/-/blob/master/src/voa_manager.c#L165 So I'm still puzzled. If you debug stereo, what kind of data are you seeing? https://docs.modalai.com/voxl-vision-px4-collision-prevention/#debug-tools

You could also convert the script into a Windows batch file or some other Windows scripting language. But there are Debian specific things in the script file that you would need to convert to the Windows alternative. For example, there are package management commands (apt) that would need to be changed into something for Windows and you would need to use the Windows version of QEMU. This could be a challenging task!

@Voxlady Thanks for the clarification I'll report my findings here.

A lot of effort goes in to supporting a new modem on a companion computer. The kernel needs to be modified, drivers need to be developed. We sell specific accessories that we put that effort in to so they can work with VOXL. Without knowing what type of modem you are trying to connect, it is difficult to suggest where to begin. Some things that might help: What type of modem is it? What interface are you connecting with? USB, serial or Ethernet? Does it have an available Linux driver? If it is USB, and you type 'lsusb', does it appear?

The short answer is yes, the long answer is that it is probably a decent amount of work. PX4 1.9 builds and runs for this platform, but may not fly. With that build, you need UART based ESCs like we have here: https://modalai.com/esc Mid-PX4 1.10 development c++14 was introduced. The SLPI compiler does not support c++14. IMO it would be fairly straight-forward to either comment out that code (it's for simulation), or just use boost to fix the issue. We just haven't done that or put much effort into it. You can take a look here: branch https://github.com/PX4/Firmware/commits/excelsior_1.9 Build the firmware with "make atlflight_excelsior_default" to build the posix side and "make atlflight_execelsior_qurt-default"

No, those are PCIe interfaced, so form factor wouldn't plug in. The "WP" series is the correct form factor, we've tested against 7610 and 7607, but there are more listed: https://source.sierrawireless.com/devices/wp-series/wp7610/

Hi @Aldo-Castro-Freire , We don't support any 5G modems on VOXL today, however we expect to in the next few months.

@Chad-Sweet [image: 1620141446462-img_20210504_100858-1.jpg] These are default connection on the board yocto:/# voxl-configure-cameras 3 attempting to use camera configuration 3 adding override_cam_id value=1 to /etc/snav/camera.downward.xml adding override_cam_id value=2 to /etc/snav/camera.stereo.xml yocto:/# cat /etc/modalai/camera_env.sh export CAM_CONFIG_ID=3 export HIRES_CAM_ID=0 export TRACKING_CAM_ID=1 export STEREO_CAM_ID=2 export TOF_CAM_ID=-1 After setting configuration to 3 as well yocto# roslaunch voxl_cam_ros tracking ... logging to /home/root/.ros/log/1f7737e8-ad09-11eb-b4a1-ec5c68cd1745/roslaunch-apq8096-3742.log Checking log directory for disk usage. This may take awhile. Press Ctrl-C to interrupt Done checking log file disk usage. Usage is <1GB. started roslaunch server http://192.168.100.90:46578/ SUMMARY ======== PARAMETERS * /rosdistro: indigo * /rosversion: 1.11.21 * /tracking/voxl_cam_nodelet/cam_id: 0 * /tracking/voxl_cam_nodelet/cam_name: tracking * /tracking/voxl_cam_nodelet/frame_id: tracking_frame * /tracking/voxl_cam_nodelet/frame_rate: 15 * /tracking/voxl_cam_nodelet/height: 480 * /tracking/voxl_cam_nodelet/is_cam_master: False * /tracking/voxl_cam_nodelet/skip_n_frames: 0 * /tracking/voxl_cam_nodelet/snav_raw10_compatability: True * /tracking/voxl_cam_nodelet/width: 640 NODES /tracking/ tracking_nodelet_manager (nodelet/nodelet) voxl_cam_nodelet (nodelet/nodelet) auto-starting new master process[master]: started with pid [3761] ROS_MASTER_URI=http://localhost:11311/ setting /run_id to 1f7737e8-ad09-11eb-b4a1-ec5c68cd1745 process[rosout-1]: started with pid [3774] started core service [/rosout] process[tracking/tracking_nodelet_manager-2]: started with pid [3780] process[tracking/voxl_cam_nodelet-3]: started with pid [3788] [ INFO] [1620154027.140051685]: Initializing nodelet with 4 worker threads. [ INFO] [1620154027.459145704]: SnapCamDriver Starting [ INFO] [1620154027.459566279]: Monotonic offset: 1620153449.753732686 libcam.enable.publish 1 libcam.enable.publish.dump 0 libcam.publish.buffer.policy max libcam.publish.buffer.max 3 libcam.debug.level info libcam.dump.dir /data/misc/camera/dumps ERROR: could not open camera subscriber for cam id 0 [ERROR] [1620154044.976910041]: CameraManager::Initialize() failed [ERROR] [1620154044.977876660]: Unable to open camera.

@Alfonso-Ponce the -f is for a factory setup (a quick way for us to set the default configuration for the m500 drones)

nothing else has changed.

Hi @Sébastien-de-Pret-Roose-de-Calesberg , I'll get an email going to this address on this account, I'm not clear what's going on and perhaps we need to send it in. Thanks and sorry for the inconvenience, Travis

It is likely there were not stable visual features on takeoff. A field of grass is not that great for visual feature tracking. You could try putting something in front of the drone on the ground for it to look at and see if that helps

It sounds like you are putting the drone into offboard mode, with the offboard mode set to "figure_eight". https://docs.modalai.com/voxl-vision-px4-other-features/ You shouldn't need to be in offboard mode to use the apriltag relocalization, see here: https://docs.modalai.com/voxl-vision-px4-apriltag-relocalization/

Hi @kiprock , The motor test commands I've used were from the MAVLink shell, but using the pwm test command: https://dev.px4.io/master/en/middleware/modules_command.html#pwm I've seen similar issues when trying to use the built in QGC PWM calibration feature... Since the Flight Core v1 can't be powered over USB (an artifact of sharing the 5VDC rail with VOXL when paired together and the fact that 500mA isn't enough to juice VOXL), it requires the python util that you linked above.