Hi @Matthew-Howlett ,
No the m500 currently has a BMP388.
Engineer at ModalAI
Hi @Matthew-Howlett ,
No the m500 currently has a BMP388.
The official product documentation is located at https://docs.modalai.com/
The purpose of this forum is to supplement that documentation, and to update the official docs when we find information missing.
Hi @rapit ,
OK we validated that a USB3 device can show up, here's a "guide", but there's no new info really, just an example: https://docs.modalai.com/voxl-as-usb-host/
-Travis
Hi @mrawding ,
The current early release on the EVT hardware relies on gstreamer.... for example:
cd /data/misc/camera
rm *
gst-launch-1.0 -e qtiqmmfsrc camera=0 name=tracker ! video/x-h264,format=NV12,width=640,height=480,framerate=30/1 ! queue ! h264parse ! avdec_h264 ! videoconvert ! pngenc ! multifilesink enable-last-sample=false location="/data/misc/camera/cam0_img_%d_.png"
This will save images to the path specified. We're actively working on getting the same interfacing as VOXL up on this ;). Next few weeks should have some good PUBLIC REPO updates.
Hi @gauravshukla914,
If voxl-vision-px4 is setup for VOA it will connect to the stereo and 'take over', if that's the case can you try the following before launching?
systemctl stop voxl-vision-px4
FYI for the future, the MPA architecture (https://gitlab.com/voxl-public/modal-pipe-architecture) will provide 'fan out' and allow multiple clients for a given camera.
@Gaurav-Borade , we have one in the works that should be ready shortly, I'll send a link once we get it posted!
Hi @mrawding ,
You are correct, there's something wrong, and I'm investigating this now.
The UART is active after bootup but then something fishy is happening.
When used with PX4 (ran on bootup) there's no issue.
I'll post here when there's an update to fix.
Hi @gauravshukla914 ,
Our flight controller is a little unique in that it can't be powered off USB so yes, it will lose power and connection during the standard process. (The reason being our Flight Controller shares the 5V rail with the companion computer and 500mA off USB won't suffice!)
To get around this, we have a tool that has a user guide here:
https://docs.modalai.com/flight-core-pwm-esc-calibration/
This allows you to keep the battery connected during the ESC calibration process.
This is for the STM32 based FlightCore.... not SLPI which I think you're super aware of but making sure!
Hi @Kevin ,
The VOXL2 (M0054) and RB5 Flight (M0052) images aren't compatible as they have different kernel and rootfs. So flashing the VOXL2 onto the RB5 leads to bricking.
But, on the RB5, you can get to a 'factory state' following this:
https://docs.modalai.com/Qualcomm-Flight-RB5-QDL/
That image is a factory image that is pretty old and not fully useable, so we need to flash an update.
You then have two choices.... reflash the same (9.1 based) image and use the RB5 Flight software (now in maintenance mode).
Or, use the ModalAI maintained VOXL SDK. To use the VOXL SDK, you need to update your RB5 Flight hardware with this from downloads.modalai.com :
We are currently working on an improved update guide and will post shortly.
Hi @mlarsen ,
(I'm still a Bitbake/Yocto/Poky padawan but can report what I know!)
Our production VOXL2 build uses:
https://git.codelinaro.org/clo/ype/external/yoctoproject.org/poky/-/blob/f5a57e939e626a5b7c6de5b51799ca602ed355ed/meta-poky/conf/distro/poky.conf#L4
DISTRO_VERSION = "2.6.1"
DISTRO_CODENAME = "thud"
(I'm also pushing towards a new build with ubun20/kernel5.4, ~alpha level right now)
https://git.codelinaro.org/clo/ype/external/yoctoproject.org/poky/-/blob/yocto-upstream.lnx.4.0.r13-rel/meta-poky/conf/distro/poky.conf?ref_type=heads#L4
DISTRO_VERSION = "3.1.14"
DISTRO_CODENAME = "dunfell"
@mrawding was logged into wrong account for above, feel free to hit up email as well!
There's a directional level shifter on the RX pin, and it seems like our default state isn't what I have documented, and I need to investigate that!
As temporary work around, you can set the pin state with onboard utility:
voxl2-mini:/$ voxl-gpio -w 67 0
For others, you can check loopback if you have it (between pins 4/5 of J10):
voxl2-mini:/$ qrb5165io-uart-test -d 0
[INFO]: success!
Sorry for the confusion! We'll fix.
Hi @JoeC , this IMX377 is a bit of a dormant binary that we've not put any active support or testing into and aren't planning on putting anytime into at this point. We have put time into IMX412 though so that's a valid path!
@AP123 because we can't provide the bootloader binary as a download, any existing M0065 HW needs to be replaced.
You can use contact@modalai.com to arrange a replacement.
It's technically possible to update in the field but we can't provide the bootloader binary so it needs to be sent into the modalai office.
Hi @dongsup-seo thanks for reporting, things should be up again here: https://gitlab.com/voxl-public/system-image-build/qrb5165-kernel
@Felipe ,
The Starling has this breakoutboard attached:
https://docs.modalai.com/usb2-type-a-breakout-add-on/
It has an apps proc UART available at /dev/ttyHS1
https://docs.modalai.com/usb2-type-a-breakout-add-on/#j5---uarti2cgpio
Perhaps you can use this UART, and voxl-mavlink-server has external FC support:
https://docs.modalai.com/voxl2-external-flight-controller/#voxl-2-connecting-an-external-flight-controller
Hi @Enrico-Bandera ,
I've seen this to, and I believe my work around was to sort of remove the working output/or start from a clean git clone.
There's some bitbake/poky cache something or other that causes this problem that I'm not clear yet how to fix... holler if you are still stuck.
the px4io implementation changed in px4 1.12 to 1.14... SDK 1.0.0 moved to 1.14 and the protocol busted...
We are planning to support SBUS in a SDK 1.0.1 release (as an engineer, I will say it's coming out soon, but I'm optimistic always... so I can't put a date down just yet).
As far as PWM support, we are looking at ripping px4io totally out of the voxl2io board and putting our own FW inside so we aren't tied to px4io at all.
Sorry for the inconvenience here, the support matrix got way to out of control and we are trying to get PWM/RC support back going with voxl2io properly and not use px4io which we can't control...
We can provide updates as we have them, including swapping out voxl2io boards in case you can't reflash in the field.
@Mastermind thank you so much for documenting this, we are going give this a go and get it into our docs! Thanks for spending the time (and sticking up for us in the other post )