v2.0.0

What's Changed

• Refactor the repository structure
• Support L4T R36
• Fix various bugs

Important

This refactoring causes files to be changed, moved or event deleted, which may require changes to your scripts or build configuration.

• Sync with the internal repository by @pfpacket in #31

Full Changelog: v1.5.0...v2.0.0

v1.5.0

What's Changed

This version newly supports the C3 camera. Additionally, it adds new L4T version support for RQX-58G and RQX-59G.

• RQX-58G + L4T R35.4.1
• RQX-59G + L4T R35.4.1

Important

Vecow EAC-5000 users and Connect Tech Anvil users cannot use this release. For these ECUs, you should use the pre-build driver instead of this package. The pre-build driver will be available separately from the release page.

• Release/v1.5.0 by @bookbridge in #27

Full Changelog: v1.4.5...v1.5.0

v1.4.5

What's Changed

This release adds the support L4T R35.4.1 (JetPack 5.1.2) for the Nvidia Jetson AGX Orin development kit.

• Feat/support orin devkit 3541 by @bookbridge in #26

Full Changelog: v1.4.4...v1.4.5

v1.4.4

Summary

This release fixes the unintended clock output frequency setting from the MAX9295 serializer to the image sensor.
This sets the frame rate correctly in the case of master mode. For instance, the C1 camera runs at 30fps instead of 31.25fps in master mode.
This fix is applicable for both C1 and C2 cameras.

What's Changed

• Add Vecow EAC-5000 support by @bookbridge in #20
• Feature/anvil by @bookbridge in #23
• Fix/max9295 clock output setting by @bookbridge in #25

Important

Vecow EAC-5000 users and Connect Tech Anvil users should use the pre-build driver instead of this package. The pre-build driver is available from the release page separately.

Support Connect Tech Anvil

Overview

This release supports Connect Tech Anvil..

Important

Unlike the other ECUs, we provide a pre-build kernel module (.ko files) for Anvil. Please download the pre-build binary from this release page and follow the instructions.

Install

Currently, Anvil has issues with building and loading the out-of-tree modules.

We provide the pre-built drivers and the installation script to avoid dealing with the complicated kernel stuff.

1. Download tier4-camera-drivers-for-anvil.tar.gz.
2. Extract the tarball and run ./install.sh to install the camera drivers.
3. The installation script does the following steps for you:
   • Install the camera drivers' configuration files
   • Install the pre-built camera drivers
   • Install the pre-built device-tree blob
   • Create a new boot entry in the /boot/extlinux/extlinux.conf
4. After running the installation script, you can run /opt/nvidia/jetson-io/configure-by-hardware.py
5. Select the GMSL port assignment by using configure-by-hardware.py. For the details, please refer to README.
6. Reboot
7. Make sure that the camera device file (/dev/video*) is created

After the installation script is finished, you must ensure extlinux.conf is well-formed and valid.
This step may require your manual editing.

Support EAC-5000

Overview

This release supports Vecow EAC-5000..

Important

Unlike the other ECUs, we provide a pre-build kernel module (.ko files) for EAC-5000. Please download the pre-build binary from this release page and follow the instructions.

Install

Currently, EAC-5000 has issues with building and loading the out-of-tree modules.

We provide the pre-built drivers and the installation script to avoid dealing with the complicated kernel stuff.

1. Download tier4-drivers-for-eac-5000.tar.gz.
2. Extract the tarball and run ./install.sh to install the camera drivers.
3. The installation script does the following steps for you:
   • Install the camera drivers' configuration files
   • Install the pre-built camera drivers
   • Install the pre-built device-tree blob
   • Create a new boot entry in the /boot/extlinux/extlinux.conf
4. After running the installation script, you can run /opt/nvidia/jetson-io/configure-by-hardware.py
5. Select the GMSL port assignment by using configure-by-hardware.py. For the details, please refer to README.
6. Reboot
7. Make sure that the camera device file (/dev/video*) is created

After the installation script is finished, you must ensure extlinux.conf is well-formed and valid.
This step may require your manual editing.

Source

https://github.com/tier4/EAC-5000_src

v1.4.3

• Fixed an issue where the camera could not be recognized properly at boot-up.

What's Changed

• Fix/cache coherency in #18

Full Changelog: v1.4.2...v1.4.3

Initial RQX-59G support

This release initially supports ADLINK RQX-59G. Verified with BSP version 1.4.2 (based on Jetpack5.1.2)

Please download the pkg file from here and follow the getting started guide.

Important

When you apply configure_by_hardware.py to configure the camera assignment, please note that you need to apply the following command (not 2= but 1=)

sudo /opt/nvidia/jetson-io/config-by-hardware.py -n 1="<YOUR PREFERRED SETTING>"

This pkg is still under evaluation for other ECUs (e.g., RQX-58G). Please refer to the latest release for the other ECUs.

Note

Please be noted, that dedicated t4cam-ctrl tools are required for RQX-59G. Please download the deb package from this page.

What's Changed

• Update for RQX59G and FPGA Fsync by @QQting in #12

New Contributors

• @QQting made their first contribution in #12

Full Changelog: v1.4.2...tentative_support_59g

v1.4.2

Updates

Camera driver

• Fix the bug related to enabling/disabling distortion correction for both C1/C2 cameras.
• Support connecting two C2 cameras to one deserializer with 10fps mode with distortion correction.

t4cam-ctrl

• Fix the bug regarding manual exposure control for C1
• Fix the bug regarding C2 control

Test environment

We've verified the driver function with these setups.

• ROScube RQX-58G + L4T R32.6.1
• Nvidia Jetson AGX Xavier Devkit+ L4T R35.1
• Nvidia Jetson AGX Orin Devkit + L4T R35.2.1

Refer to the documentation https://tier4.github.io/edge-auto-docs/ for the details

Test Result

Warning

• Port 7 and Port 8 cannot be used in Xavier Devkit
• Xavier/Orin devkit only supports master mode (free-run mode) for both C1 and C2.

In the case of C1

Test \ Port	port 1	port 2	port 3	port 4	port 5	port 6	port 7	port 8	Result
1	C1								OK
2		C1							OK
3			C1						OK
4				C1					OK
5					C1				OK
6						C1			OK
7							C1		OK
8								C1	OK
9	C1	C1							OK
10			C1	C1					OK
11					C1	C1			OK
12							C1	C1	OK
13	C1		C1						OK
14	C1			C1					OK
15		C1	C1						OK
16		C1		C1					OK

Note
Test 13-16 could cover port 5 to port 8 as the structure is identical with the port1 to port 4

In the case of C2

Test \ Port	port 1	port 2	port 3	port 4	port 5	port 6	port 7	port 8	Result trigger_mode=0/1	Result trigger_mode=2/3/4/5
1	C2								OK	OK
2		C2							OK	OK
3			C2						OK	OK
4				C2					OK	OK
5					C2				OK	OK
6						C2			OK	OK
7							C2		OK	OK
8								C2	OK	OK
9	C2	C2							OK *	NG
10			C2	C2					OK *	NG
11					C2	C2			OK *	NG
12							C2	C2	OK *	NG
13	C2		C2						OK	OK
14	C2			C2					OK	OK
15		C2	C2						OK	OK
16		C2		C2					OK	OK

Note

• Currently, in the case of Test 9 ~ 12 (2 cameras with single deserializer), the framerate must be set to 10fps.
• Test 13-16 could cover port 5 to port 8 as the structure is identical with port1 to port 4

In the case of a combination between C1 and C2

Test \ Port	port 1	port 2	port 3	port 4	port 5	port 6	port 7	port 8	Result trigger_mode=0/1/2/3/4/5
1	C1		C2						OK
2	C1			C2					OK
3		C1	C2						OK
4		C1		C2					OK

Note
Tests 1-4 could cover port 5 to port 8 as the structure is identical with the port1 to port 4

Release v.1.4.1-fix-1

Modifications

• Several bug fixes for C2 camera (#14):
  • Fixed the number of lanes in the case of Nvidia Xavier devkit
  • Fixed the bug regarding the activation of distortion correction.
  • Fixed the bug regarding the trigger_mode setting (in the case of setting 0)

Documentation

Edge.Auto documentation site