Upgrade to py39 / tf 29

[DocGarbanzo]

Upgrading Python to 3.9 and Tensorflow to 2.9

This doc update covers the change in Donkey Car when we merge autorope/donkeycar#1119 into main. The PR upgrade will bump the Donkey Car version to 5.0.X. In this PR for the documentation change, there are essentially 3 changes:

• Upgrading of python to 3.9 (note, only the jetson jetpack 5.0.X install is using python 3.8 which is the system version and therefore Nvidia provides the required tensorflow and Open CV packages). Upgrading tensorflow to 2.9.

• Upgrading the Raspberry Pi to Raspberry Pi OS Bullseye 64 bit. We are a making use of the new graphical installer for that.

• Upgrading to Jetpack 4.6.X on Jetson Nano and 5.0.X on Jetson Xavier (and newer Jetson boards, like the Orin)

Also changing the theme from readthedocs to material. Note the doc change here still keeps the old installation instructions if you want to stick to Donkey Car <= 4.5.X

[BrianHenryIE]

Can <user name> be replaced with $USER?

[DocGarbanzo]

Hm, I think the $ is wrong there anyway and we should rather put

-D "${HOME}/mambaforge/envs/donkey/bin/python3.9"

Could you check that? There are 4 more references to mamba forge and opencv, I have replaced all of them.

[BrianHenryIE]

$ make -j4 is missing here.

I think it would be better to leave out this and the following steps in this code block, and leave only the cmake command which differs from the Qengineering install steps

[DocGarbanzo]

I think @TCIII ran into problems when compiling with -j 4 and even with -j 2 which was also mentioned in the Qengineering instructions. In the end we gave up and run w/o any parallelisation, which was slow but completed. Did -j 4 work for you without problems?

[BrianHenryIE]

Mostly in addition to the comments above, I got Open CV to install on my third try.

1. I had an existing Donkey install on 4.6.1 which I tried to use as a base. Failed between 30/40%
2. A fresh install just for this, where I tried to follow the steps of the Q-engineering tutorial but it failed between 30/40% (different error)
3. Success: I downloaded the Q-engineering install script and edited in the cmake changes in the documentation.

Important to note (1) was its own legacy Donkey install on SD card, and (2) and (3) were the same new install..

[DocGarbanzo]

@BrianHenryIE - thank you for all your good comments and the thorough review of the doc update. I have updated the files as mentioned in the comments above. If you could confirm the changes are good and work for you - that would be great.

[francisduvivier]

Hi,
I've been working on this branch with a raspberry pi and working great for me.
Some notes from my side:

• There is no Camera option in the interface settings in raspi-config, only the legacy camera option and I don't think that one needs to be enabled because it works without for me. So I'd remove the part saying you need to able the camera in raspi-config.

• Regarding Donkey on Raspberry pi OS Lite, I had to do the following additional steps, might be worth mentioning:

  • sudo apt install pi git -y
  • pip install virutalenv

Additional note, but not sure if anything about this needs to be in the donkey docs:

• when using Raspberry pi OS Lite and you want to connect through ssh for the first setup, the installation steps have changed. You cannot simply create a ssh file anymore to enable ssh, you now also need to create a userconf.txt. But all of that is described on the Raspberry Pi documenation pages: https://www.raspberrypi.com/documentation/computers/configuration.html#setting-up-a-headless-raspberry-pi

[BrianHenryIE]

I just sat 20 minutes waiting for this to finish... It was prompting me to confirm overwriting files (although it was a fresh SD install)

Maybe we should add the -y flag: sudo apt update -y sudo apt upgrade -y.

I haven't tested it yet, particularly the fullscreen dialog: "Configuring docker.io" "Automatically restart Docker daemon?"

[BrianHenryIE]

-y flag did not work to automatically accept prompts.

~75%

Setting up nvidia-l4t-core (32.7.3-20221122092935) ...

Configuration file '/etc/ld.so.conf.d/nvidia-tegra.conf'
 ==> Modified (by you or by a script) since installation.
 ==> Package distributor has shipped an updated version.
   What would you like to do about it ?  Your options are:
    Y or I  : install the package maintainer's version
    N or O  : keep your currently-installed version
      D     : show the differences between the versions
      Z     : start a shell to examine the situation
 The default action is to keep your current version.
*** nvidia-tegra.conf (Y/I/N/O/D/Z) [default=N] ? 

77%

Setting up nvidia-l4t-oem-config (32.7.3-20221122092935) ...

Configuration file '/etc/systemd/nv-oem-config-post.sh'
 ==> Deleted (by you or by a script) since installation.
 ==> Package distributor has shipped an updated version.
   What would you like to do about it ?  Your options are:
    Y or I  : install the package maintainer's version
    N or O  : keep your currently-installed version
      D     : show the differences between the versions
      Z     : start a shell to examine the situation
 The default action is to keep your current version.
*** nv-oem-config-post.sh (Y/I/N/O/D/Z) [default=N] ? 

81%. This is a fullscreen prompt

 Configuring docker.io 
Automatically restart Docker daemon?