@Eric-Katzfey If that works then try using only the -o
option and removing the -g
and -m
options so that you have no gps or magnetometer coming in, only VIO odometry. With that also change the PX4 parameters SYS_HAS_GPS and SYS_HAS_MAG to 0 and EKF2_MAG_TYPE to 5. Then try to fly VIO only with that.
Posts made by Eric Katzfey
-
RE: VOXL2 - HITL/VIO
-
RE: VOXL2 - HITL/VIO
@leonardo-Andrade-0 Can you change the start command to remove the
-o
option from dsp_hitl start? That will try prevent VIO odometry from being published and only leave GPS and magnetomer for position hold. See if that works. You can see that home position is not valid in the failsafe flags but it should be if GPS is coming in. -
RE: VOXL2 - HITL/VIO
@leonardo-Andrade-0 It shows you are just running gazebo from the command line. You need to be running our version of gazebo from the Docker. See https://docs.modalai.com/voxl2-PX4-hitl/#gazebo-simulator-setup-on-host-computer
-
RE: voxl2 rangefinder (lw20)
@voxltester Again, LW20 is not in the px4 build. Until you rebuild px4 with that driver you cannot use it within px4. So, you have 2 choices. Rebuild PX4 and try to get LW20 working within PX4 or take a look at the source code for voxl-rangefinder-server and see if you can add support for LW20 into that. It currently supports the vl53l1x rangefinder. Perhaps you can just switch to that rangefinder model? Otherwise the source code it here: https://gitlab.com/voxl-public/voxl-sdk/services/qrb5165-rangefinder-server
-
RE: Deep learning and MAVSDK on VOXL 2
@achanana Docker containers are just really useful when you have dependency problems. You can set up the container with all of the proper dependencies. When running directly on VOXL 2 you don't necessarily have all of the correct dependencies and trying to install them all will often be difficult and at times impossible (If there are conflicts). In this case voxl-cross may not be the correct compiler required for MAVSDK.
-
RE: voxl-camera-server bugs with snapshot capture for mapping
@AndriiHlyvko Snapshots commands are handled by the voxl-mavcam-manager application. It has it's own mavlink connection to QGC. You can enable debug messages in voxl-mavcam-manager by adding the
-d
option to see what commands it is getting. The images will have timestamps based on the system time at the voxl, not of the external flight controller. -
RE: Outdoor navigation problem
@Amin-Bassiri So, does the mag problem happen on all 7 drones?
-
RE: voxl2 rangefinder (lw20)
@voxltester The
voxl-configure-rangefinders
refers to range finders connected to the apps processor outside of px4. If you use that then thevoxl-rangefinder-server
will communicate to px4 with mavlink messages to indicate distance. -
RE: voxl2 rangefinder (lw20)
@voxltester In order to use the LW20 with i2c interface you would need to add that driver to your build (e.g. in
boards/modalai/voxl2-slpi/default.px4board
). And then you would need to start that driver in the startup script (e.g. inboards/modalai/voxl2/target/voxl-px4-start
). -
RE: lis3mdl magnetometer
@mkriesel We don't have any immediate plans to add dsp i2c access to libqrb5165-io.
-
RE: lis3mdl magnetometer
@mkriesel You cannot access the DSP i2c ports from the applications processor.
-
RE: lis3mdl magnetometer
@mkriesel Yes, you would need to compile the DSP code with the Hexagon SDK.
-
RE: lis3mdl magnetometer
@mkriesel The low level driver is Qualcomm proprietary so we cannot share it. But you don't need to look at that code to use an i2c port. You just need to use the interface functions provided by px4.
-
RE: VOXL2 - HITL/VIO
@leonardo-Andrade-0 Is px4 generating a log for the failed takeoff attempt? It's best to go into /data/px4/log and clear out any logs that may be saved there. Then run your test and verify that a new log was created. That log should indicate the reason for the arming failure. You can also just look at topics like
failsafe_flags
andvehicle_status
with thelistener
command to see what they are indicating as failures. What does the start command look like in/usr/bin/voxl-px4-hitl-start
? Did you change it from the defaultqshell dsp_hitl start -g -m -o
? -
RE: Outdoor navigation problem
@Amin-Bassiri I think the first thing is to figure out what has gone wrong with the mag. You need to get it flying nicely in position mode with GPS / Mag / baro first. VIO is trickier, especially outdoors. Since it relies on computer vision for it's tracking it can fail for quite a few different reasons and is more difficult to debug. VIO indoors where the environment can be controlled more closely is ideal. Outdoors VIO can be a lot more challenging and there will certainly be cases where VIO will fail. Outdoors it's probably best to use both GPS and VIO simultaneously and have PX4 fuse these measurements internally depending on the quality of each one at any particular time. Regardless, let's figure out what is going wrong with the mag first. Does the mag problem happen on all 7 drones? Are you able to calibrate the mag on any of them?
-
RE: Lepton board
@Projects-a You can connect the i2c to the J5 connector on M0141 and spi to J10 on VOXL 2 board if you want to stay with M0141
-
RE: VOXL2 microdds agent namespace
@Darshit-Desai Seems like some bash scripting issues. Perhaps
echo $PX4_NAMESPACE
before the microdds_client start line to see what is in that variable. -
RE: VOXL2 - HITL/VIO
@leonardo-Andrade-0 Doesn't look like the HITL driver is receiving any data over the serial port. That's why you see those failures about baro, gyro, and accel data missing. Can you execute the command
qshell dsp_hitl status
and see that it says? -
RE: Lepton board
@Projects-a If you go with an LTE add on module or a Microhard add on module then you can use one of the USB ports on them to attach the Lepton.
-
RE: VOXL2 microdds agent namespace
@Darshit-Desai Rather than modifying voxl-px4.conf to use EXTRA_STEPS, can you try directly modifying
/usr/bin/voxl-px4-start
and adding that command?