Skip to content

issues.md

Latest commit

 

History

History
193 lines (129 loc) · 7.66 KB

issues.md

File metadata and controls

193 lines (129 loc) · 7.66 KB

This augments the documentation found here: https://git.hsr.io/tmc/hsr-ros2-setup/-/blob/main/README-EN.md

Also in this repo here

Steps that we needed help with

Unplug HDMI

Description: instructions say we must disconnect the head display HDMI Solution: remove cover and unplug hdmi cable labeled 1030 on the cpu board on the right side (looking at it from the front)

Long term: need male VGA to female HDMI adapter that connects the VGA port by the E-Stop to the disconnected male HDMI

Reinstalling OS

Description: doing fresh ubuntu installation we got intermitent errors preventing loading into new boot

Solution: do a clean installation with the distribution listed

We had lots of trouble getting into the computer after installation
No fix was found, but it eventually just booted in properly part of the time

  • Continuing the installation seems to have resolved the issues

If you come to a blank black screen, press any key and it will continue -- this happened a few times after booting in recovery mode

Network configuration

Description: post running ansible on host PC network not working

Solution: Ran the following command to bring the network back up and running before being able to connect again:

sudo ip link set wlp3s0 up

Also make sure wifi is on with:

sudo nmcli radio wifi on

Even with that fix, can't use the wifi options in settings, must use the sudo wpa_gui mentioned in the docs

Also turning of wifi in settings and turning back on breaks it, but fixed with a reboot

Rewriting Arduino

Description: arduino didn't have access to the serial port Solution: add the user to the dialout group in order to have access to the serial port /dev/ttyUSB0:

sudo usermod -a -G dialout <USER>

Then run newgrp dialout to apply the changes

Restoring Arduino Software for ROS1

We didn't do it, figuring that we do want the ROS2 not ROS1 arduino

  • Might revisit though because of Calibration section specifically

Creating Docker Image

Description: permission denied for ./make.sh Solution: use sudo ./make.sh

Additionally This took over an hour, so be patient.

Running ROS2 Setup with Ansible

Description: permission denied when running ansible commands Solution: add sudo

Setting up Cyclone DDS

Description: would error out (forget what specifically) Solution: use ROS_DOMAIN_ID != 0

Python Interface

Description: errors when running

docker exec -it docker.humble.diag.service /bin/bash -c ". /etc/opt/tmc/robot/ros2_setup.sh;ros2 run hsrb_interface_py ihsrb.py"

Solution: needs something running on host pc?

  • seems to time out waiting for a service
  • Errors with ddsi_udp_conn_write to udp/ fixed by ROS_DOMAIN_ID != 0 I believe

Can't access hand camera

Description: hand camera not accessible through rviz

Also can't access through cheese or kamoso

  • though also aren't able to access head center cam which we can access through rviz so maybe just prohibited by some hsr thing --- Fixed itself later

Tried:

v4l2-ctl --device=/dev/video2 --stream-mmap --stream-count=3 --verbose

Throws:

        VIDIOC_REQBUFS returned 0 (Success)
		VIDIOC_QUERYBUF returned 0 (Success)
		VIDIOC_QUERYBUF returned 0 (Success)
		VIDIOC_QUERYBUF returned 0 (Success)
		VIDIOC_QUERYBUF returned 0 (Success)
		VIDIOC_QBUF returned 0 (Success)
		VIDIOC_QBUF returned 0 (Success)
		VIDIOC_QBUF returned 0 (Success)
		VIDIOC_QBUF returned 0 (Success)
		VIDIOC_STREAMON returned -1 (Input/output error)

Interesting:
This gives same error for video2 (center cam that works) But it works on 41...

Tried:

gst-launch-1.0 pipewiresrc path=34 ! videoconvert ! autovideosink

** path=34 is from path= listed from gst-device-monitor-1.0 Video/Source

Get: black screen, not recieving data...

Even nothing when doing gst-launch-1.0 pipewiresrc path=34 ! videoconvert ! video/x-raw,format=RGB ! fakesink dump=true which should cat the raw bits.

Update following day:

These commands now work for video2 at least, but cheese still isn't working

Next steps:

  • check these commands on 41
  • open up case and check wiring (hsrb.io has some guidance here)

Current status:

Cheese itself now works on both hsrs

BUT switching feed to (82) hand camera freezes last head camera frame (presumably no data is being sent to overwrite)

Therefore likely disconnected cable (1011 or 1008)

  • 1006 seems fine found on the bottom of the cpu (bottom)

Partial solution:

This seems to fix camera for command line streaming, still NOT working in rviz. Make sure that the device is the hand cam should be video0, but might be swapped to video2.

v4l2-ctl --device=/dev/video0 --set-ctrl=exposure_auto=1
v4l2-ctl --device=/dev/video0 --set-ctrl=exposure_absolute=300
v4l2-ctl --device=/dev/video0 --stream-mmap --stream-count=3 --verbose

Potentially needed to be executed prior to the above commands (resets usbs, but also might switch video0 and video2):

echo 0 | sudo tee /sys/bus/usb/devices/1-1.4/authorized
sleep 5
echo 1 | sudo tee /sys/bus/usb/devices/1-1.4/authorized

v4l2-ctl --list-devices should result in the following if the video channels are right:

NCM13-J-02 (usb-0000:00:1a.0-1.4):
	/dev/video0
	/dev/video1
	/dev/media0

NCM13-J-02 (usb-0000:02:00.0-1.3.1):
	/dev/video2
	/dev/video3
	/dev/media1

Likely explanation:

We were getting an initial frame some of the time, but only the first frame or two. Claude's explanation of this phenomenon is below.

When a UVC camera powers up, the image sensor goes through an initialization sequence — it charges up, sets exposure, and produces a burst of frames before any control negotiation happens. Those first 2-3 frames are essentially "startup frames" from the sensor's cold start. After that, the camera firmware tries to apply the UVC control settings (including control 8, exposure), fails with the -32 error, and the sensor either shuts down or enters a fault state. This is why you see data briefly then nothing.

Final Solution:

The usb-b mini cable in the hand was bent, swapping that and plugging into usb port on the robot allows the camera to function. (USB bar on top didn't have enough bandwidth, but the silver ones on the bottom right did)

TODO: Full solution will likely be stripping the cable and splicing on the new head as rerouting cable seems extremely difficult. Awaiting instruction from Taskin.

Can't access stereo head cameras

Description: Can't access cameras from rviz

Attempts: The command flycap allows us to access the individual cameras... so not hardware (can be used to identify left and right)

KEY ERROR: Not found key 'cameras' or 'cameras' node type is not sequence.

This is from the src/tmc_drivers/tmc_pgr_camera/src/tmc_pgr_camera/yaml_point_grey_camera_system_setting.cpp file

Error is while parsing camera ids in line 192.

This led me to looking at the stereo camera yaml file in /etc/opt/tmc/robot/conf.d/calib_results/stereo_camera_params.yaml (specified in line 47 of src/hsrb_robot/hsrb_bringup/hsrb_bringup/launch/sensors/stereo_camera.py). The file is in the wrong format according to the comments in yaml_point_grey_camera_system_setting.cpp.

Solution: Replace with the proper file that only includes the camera id's (found in flycap).

The LEFT camera (the one on the right if you are looking at the face) should be listed as master. Make sure to check that the id's are correct, probably not the ones listed.

Note: The point grey camera system mentions additional parameters, but that was deprecated and relocated to src/tmc_drivers/tmc_pgr_camera/launch/capture.launch.py

Unsolved