Hi @Gary-Holmgren. This is likely because you're running the build script from an x86 system (docker requires that you build on the same architecture as the base image, in this case arm64). You can solve that by cloning the repo onto the voxl and running build.sh from there.
Note: make sure your voxl has a wifi connection before running ./build.sh.
@GlennTee So that example requires you have the MicroDDS installed and running to communicate with PX4 (guide here).
Alternatively, you can run offboard commands via mavsdk (example here). If you do it this way, make sure to edit that example to connect on port 14551 and set en_localhost_mavlink_udp to true in 'etc/modalai/voxl-vision-hub.conf'.
Let me know if that helps.
@GlennTee What do you see when you run ros2 topic list?
Have you run "apt-get install voxl-ros2-foxy && apt-get install voxl-mpa-to-ros2" followed by "voxl-configure-mpa-to-ros2".
@GlennTee said in VOXL2/ROS2 Drone Runs Figure 8 Offboard Program Instead of My Program:
I thought I have reverted everything to how it was before, but the drone is now back to not flying at all. I also noticed that my list of ros2 topics has also shrunk down as well.
This likely means that those topics got populated from the offboard script not the px4-ros bridge. Could you check that voxl-microdds-agent is running when you see this happen.
Check with "voxl-inspect-services"
Hi @KnightHawk06. If you're looking for all the data stored in the ai_detection_t struct we've done it in the past by opening up a libmodal pipe to the tflite detection path defined like this:
#define TFLITE_DETECTION_PATH (MODAL_PIPE_DEFAULT_BASE_DIR "tflite_data/")
Let me know if that suits your needs and if you need help opening the pipe.
Hi @Jing-Yu,
It looks like your TOF sensor might be out of focus.
This is what normal output looks like:
You can change the focus by rotating the ring around the TOF sensor. You might want to use a set of pliers but be careful not to scratch the glass.
Hi @Roya-Salehzadeh,
Yep, the points showing up in voxl_mapper_aligned_ptcloud are static (ie. not tied to the vehicle's current pose)
I've never tried to do that so I'm not 100% sure but recording the points with a ros2 bag seems like a reasonable approach
3a. I'm not sure why you would need to do any transformation to plot the data on RViz. Voxl-mapper already does transformations to put the data in a static frame of reference. If you're only seeing it as a live stream this is probably because it's playing one message one after the other. I'm not sure but there might be a setting in RViz to ignore the timestamps or you could potentially do something like set all the timestamps to 1.
3b. Voxl-mapper uses the tof data to generate the map. What are you trying to do by visualizing the tof data?
3c. For visualizing the path I'm not sure if you would need to do transformations or not. This is the line that digests the pose data in voxl-portal. After a little investigating it seems like the /world might be the correct frame_id.
If it were me, I'd start by getting the accumulated map on RViz then plot the odometry inside that map with the /world frame_id and then check if you need to do any additional transformations.
@brahim what're your thoughts on the RViz frame_id for odometry?
Let me know if that helps,
Ted
@Jing-Yu what SDK are you using? My TOF is working with SDK 1.4.5. I'll try to replicate your setup the best I can but I'm suspicious because the TOF IR is working and the TOF depth pipe is receiving frames in your setup. In the meantime, can you try pointing the TOF sensor at something very far away?
That looks good except voxl-mapper doesn't look like it's installed or running. You can install it using the command below. Make sure your starling is connected to wifi else you can adb push the voxl-mapper deb.
$ wget http://voxl-packages.modalai.com/dists/qrb5165/sdk-1.5/binary-arm64/voxl-mapper_0.2.3_arm64.deb
$ dpkg -i voxl-mapper_0.2.3_arm64.deb
After it's installed and running check that you see the map updating on portal and you should see the voxl_mapper_aligned_ptcloud pipe show up as a ros topic.
Hi @taiwohazeez,
That is the cable that powers the VOXL. I'm suggesting you leave that plugged in and disconnect everything else. I see you still have the VOXL 2 WiFi Add-on board connected. Disconnect that as well all the cameras on the bottom of the VOXL, ESC, and GPS.
Please have the VOXL powered off while you disconnect everything.
Let me know how it goes,
Ted
Hi @Jing-Yu,
Does the feed for tof_depth change when you move your drone?
Hi @taiwohazeez,
I'd start with unplugging all the peripherals (cameras, wifi, gps, rc, etc) so you just have a bare voxl connected to a power module and check if it boots then.
For "voa_pc_out" the default on voxl-vision-hub allows input from both stereo and tof. You can check which inputs you have enabled by checking the voa_input in "/etc/modalai/voxl-vision-hub.conf".
Correct, I would just use "voxl_mapper_aligned_ptcloud"
Can you share your output from voxl-inspect-cam -a to make sure your stereo cameras are setup correctly? I would debug this on voxl-portal before moving onto the ROS portion. Do you have tracking cameras setup, and will you be running openvins or voxl-qvio-server. Once that's verified check that you're seeing that pose on voxl-mapper. You should see the orientation show up like in the red box in the picture below or by running voxl-inspect-pose -l.
Hi @Judoor-0,
Cool! Feel free to make a MR or post the link to your code if you want someone to review it.
Hi @Judoor-0,
Interesting I'm assuming you're looking at the rangefinders pipe. I can take a look at your code, or you can make a MR back into mpa-to-ros2.
Also, something to consider is using microdds and subscribing to the "fmu/in/obstacle_distance" topic.