@Chad-Sweet I found this. The 5G modem board also has a USB 3 port. [image: 1675315192006-%EC%BA%A1%EC%B2%98.jpg] Then, all I need is this cable. https://www.modalai.com/collections/cables-1/products/mcbl-00022

Please directly move to 22 version if possible

These are usually hardware issues. Some possibilities crimped flex cable exposed copper on the flex cable loose or damaged connector

Hm thats super strange but I am glad you got it working! Keep us posted on how the project turns out

It worked, thank you

It was a custom PCB that we do not have available as a product

@Dylan-Lau said in Safety Setup parameters missing from firmware: @Adrian-Hidalgo Here is the px4 parameter file: link text Hi @Adrian-Hidalgo, may I know if there is an update? We thought of some workarounds but since it deals with safety, it is better to implement something with the support team's expertise. We found https://gitlab.com/voxl-public/flight-core-px4/px4-parameters/-/blob/master/platforms/master/temp1.params and https://gitlab.com/voxl-public/flight-core-px4/px4-parameters/-/tree/master/platforms/v1.13/M500 and so should we use this instead of the default one we have from the system image? Cheers Dylan

yessss 6x9 and a white background both improved detection rate, not just latency. Thanks for your help!

It looks like configuration 28 is what you're looking for with the standard 3840x2160. That's the resolution I've tested the 214/377 sensors with.

@ShauliXtend said in Building a custom docker image with VOXL libs: on the VOXL itself you use 32bit libraries while the system is 64bit Yes, unfortunately the QVIO algorithm is proprietary to qualcomm, we don't even have the source, just the 32 bit binaries that they've provided. maybe there is a way to still use the 32bit file You'll need to compile a 32-bit program to use a 32-bit library, there's unfortunately no way around this without some sort of inter-process communication (hence our standalone qvio-server)

@Alex-Gardner So I guess I can't really do much to debug this (without knowing more about the proprietary implementation), true? Point being... if there is documentation you can share, I'll gladly dig into this.

Thanks for the hint @Alex-Gardner, I tried two separate modules on 3.3.0, 3.6.0, and 3.8.0 and they worked just fine, it is only with system image 4.0.0 and voxl-sdk 0.9 that neither one works which makes me conclude that it's not the sensor but something with the software. Also both modules works just fine when run by themselves, it is only in conjunction with other sensors like stereo/tof that the hires is causing issues.

@Chad-Sweet Thanks for your response. I apologise for the late response. As you correctly pointed out, we resolved the issue after a re-calibration of the front stereo pair in natural lighting and by further tuning the 'blur_size' and 'post_median_size' parameters in the DFS server config file. Now the data reflects actual physical obstacles quite reliably.

@Youngjoo-Kim J18 is the port normally used for the UART based ESC. It is connected to the DSP, not the Linux processor. You cannot directly access it from Linux. Why do you need to run Mavlink over a serial port? We support and recommend using UDP.

You should be using docs.modalai.com , the beta-docs are only published for specific, pre-release components. You cannot plug image sensors in where ever you like. You need to follow a supported configuration here: https://docs.modalai.com/voxl2-camera-configs/

Hi Chad, Thanks. One more question: Once new features are necessary during development or feasibility check, how to get ModalAI direct support? by contract or only through ModalAI forum? Best Regards, Dan

thanks for the suggestion, note added to hellocamera

Hi Jose, there's really not enough information for us to help here. What flight controller are you using? how are you connected? What did you change to get it into this state? that error implies you do not have GPS connected, or it is not outside to receive position information

You need to calibrate the thermal camera to generate the intrinsic. There is no off the shelf way to do this. You will need to create a to-scale checkerboard that the sensor can see