@Taylor-Presley Please upgrade to the latest SDK, it should be resolved

@Gary-Holmgren Hi Gary, Flight Core simply has more I/O. So, if you need more I/O, like CAN, then Flight Core is a better option. VOXL 2 I/O just adds in PWM and S.Bus, so it is very limited

@tom Thanks for the reply. I thought the same at the start. I also tried connecting with the ethernet cable, and the results were the same. Do you mind if any other issues can cause the same behaviour?

@Alex-Kushleyev @Eric-Katzfey Thank you for clearing this up. I am trying to install SDK 1.1.2 but my host machine can't discover the VOXL2 as a fastboot device (I have made an issue here). I would appreciate if you have any recommendations to hop on there as it is an urgent issue.

@Amber-Elaine-Parker I think our Seeker was doing something similar but it is a little bit difficult to identify because of the viewing angle of the video. I can't find our original video but our Seeker was oscillating around X-axis (FRD reference frame) like this:

f8e87bf6-de8a-49c8-9e6c-013bc6cba5f7-image.png

@relevinsky , i don't have a good example to point you to.. you would need to modify the code on the web GUI side and the code that is running on voxl-portal Voxl.

VOXL sends all the information to the browser via web sockets. Here is where video stats (including fps) are sent : https://gitlab.com/voxl-public/voxl-sdk/services/voxl-portal/-/blob/master/src/video_group_manager.cpp?ref_type=heads#L313 and there is the browser code that parses the video stats https://gitlab.com/voxl-public/voxl-sdk/services/voxl-portal/-/blob/master/web_root/video.html#L588.

I would suggest first experimenting with existing data structure, such as VideoStats and see if you change the data on the VOXL side, make sure it shows up correctly in voxl-portal. After that, you could change the VideoStats structure add a message to it and also print it in the browser.

After understanding how the message passing works, you could add your own message definition on both sides and display the contents.

I hope this helps.

Alex

@planch98 Thank you so much! I have solved the problem.

@SMRazaRizvi , the reference frame should be NED (convention that is used in PX4) - can you please check if that makes sense?

For conversion, perhaps this would help : https://answers.ros.org/question/336814/how-to-change-ned-to-enu/

"SOLVED"

Hi everyone, the problem was due to our own error of not updating to the most current SDK. after the update everything was fixed, although there was still a faulty stereo sensor that will not function.

@brahim , Tom is on vacation for the holidays, so please be patient with us during this break.
Most of our team is taking some much needed time off to spend with family.

In the meantime, I'd go through the usual trial and error of different hardware, cables, and USB ports to see if things improve for you. Also check all connectors and cables are secure and pin-1 aligned.

Thanks!

@sarahl Hi Sara, Could you share any info, like a colab notebook or whatever, on how you trained a tflite that works on the voxls as i've had zero luck even though my trained models work on my linux and mac systems. I created a yolov5 and quantized it and it inferered correctly. I also was able to convert a yolov8 model to tflite using ultralytics converter. I also retrained my own moblie ssd model and quantized it and they all infer correctly except on the voxl2.

@Prawin-Sankar Software-wise, yes. However it would take a custom mounting solution and would add a lot of weight to the Starling so it would require a re-tune.

@coreyazion I was able to switch to position flight mode and the drone flew better in position flight mode. In my previous testing I had issues with position flight mode with drifting but in my most recent attempt it flew much better and was relatively stable. I did experience after a short while some more drift towards the back-left of the drone however I am unsure if this is due calibration or configuration. I also experienced the same issue with PX4 Hold mode.

@Vinny Extremely useful. Thanks Vinny!

Hi @rohitpillai
The intent of the power module is to take direct 2S-6S battery via the XT-60-M, and then pass that power input to two places:

a 5V DC/DC for VOXL / VOXL 2 using the Molex 4-pin with MCBL-00001 (https://docs.modalai.com/cable-datasheets/#mcbl-00001), not the XT-60 Output using the XT-60-F to an ESC/Other VBAT based item in your platform.

Both are measured by the sense resistor when connected properly as we instruct.

In some instances, as in the case of a VOXL 2 Mini, we have configured the Power Module to not output a downstream VBAT using the XT60-F (https://docs.modalai.com/voxl2-mini-power/#voxl-power-module), but a user is easily able to do so.

Hope this helps!

@m4v3r ,

I think the ESC params installation failed during an upgrade. Not a big deal. Here is a procedure you should follow to get your ESC to a good state. We have been improving the update process, so hopefully this does not happen in the future.

get the latest voxl-esc tools on your voxl2. You can clone directly to voxl2 or to your PC and copy to voxl2 if your voxl2 does not have internet access git clone https://gitlab.com/voxl-public/voxl-sdk/utilities/voxl-esc.git -b dev stop px4 systemctl stop voxl-px4 make sure you can scan / detect ESCs . Note that based on your current state, the script should find at least one ESC running at 57600 baud rate, which would mean the param install failed (57600 is the fall back baud rate). This is ok. cd voxl-esc/voxl-esc-tools ./voxl-esc-scan.py install the params based on your motor / propeller configuration or any custom ESC param that you created. This script should find all the ESCs even if they are running at different baud rates ./voxl-esc-upload-params.py --params-file <path_to_esc_params_file.xml> install latest firmware (this script will auto detect the board and pick the correct firmware). Please note that if the ESCs are detected at different baud rates, that needs to be fixed with the correct params (command above). This script will not automatically find the ESCs running at different baud rates (we will fix this soon). ./voxl-esc-upload-firmware-all.sh

Then you can start px4 again if needed

systemctl start voxl-px4

HI @iksk
That stiffener on the older units is not a shield. It's just floating metal (stainless steel). We have since moved away from all metal stiffeners due to electrical shorting risks in tight drone assemblies, and all new flexes have the laminate backers now. The laminate is slightly thicker at 0.8mm instead of 0.2mm.

If you are seeing performance issues between the two, it is purely coincidence and likely more result of manufacturing variance or handling resulting in degradation. We have built and tested thousands of the new ones and there are no stability differences between them.

Hope that helps.

@SMRazaRizvi ,

Please see the following post that answers your question : https://forum.modalai.com/topic/2551/add-mask-file-to-qvio-algorithm .

Specifically, you would need to add the name of the static mask file here (ideally full path of the file) : https://gitlab.com/voxl-public/voxl-sdk/services/voxl-qvio-server/-/blob/master/server/main.cpp?ref_type=heads#L384 (staticMaskFileName which is initialized to Null).

See the other thread for explanation what the mask format should be.

@tiralonghipol , great to hear! if you have this issue again, you should be able to recover the ESC using the steps I described.