Jetracer Python examples

Hello

I am writing a report on Autonomous Vehicles and have a Jetracer kit from Waveshaw with a jetson nano. I am using Jupyter and python to test a few things and want to know if I can actually have programmatical control over the throttle and steering without using any widgets or traitlets. I think this is possible by using the motors direct but not sure how to do this at the moment. While some of the demo code is ok I would like to do this from scratch!

Hi

Please refer to below threads to see if can help:



Nope, none of this helps. I can use the motors fine but what I want to do is write my own code. I have used adafruit for the front steering servo and this works really well. The two rear wheel motors using this do not work no matter what channel I use. If I use the supplied python code and then car.throttle = 1 it works fine.

Is there a way of programming the steering and the throttle without using any traitlets or widgets just by pure python code?

This has been sorted now, the only issue I have at the moment is I cannot run a CNN model on the jetson Nano as the cuda drivers will not load.

Ok so my question is how can I install tensorflow-gpu==1.15 and all the dependencies? currently all i see is errors as no tensorflow versions available :(

Please refer to https://elinux.org/Jetson_Zoo#TensorFlow

Sadly following this through I have the same message I receive on all versions of TensorFlow when installing - Ettot: No matching distributions found. or installed v1.13 errors when running the cnn (very small model to test even fails).

Error: could not find the algorithm which may be because the cudnn failed to initialise

If I cannot have this working in the next 24hrs my thesis is ruined. I chose the Jetson as on paper it had every everything I needed to write my own autonomous driving model but this does not appear so as I the architecture I need to run fails!

Hi,

If you are using JetPack4.4, please use the v1.15.3+nv20.8 package.
Please noticed that it’s required to use the package built with the same JetPack version to make it compatible.

You can find more information in our document:
https://docs.nvidia.com/deeplearning/frameworks/install-tf-jetson-platform/index.html

$ sudo pip3 install --pre --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v44 ‘tensorflow<2’

Thanks.

Many thanks for a quick response, One other question the image jetcard_v0p0p0 what jetpack should be pre-installed? Is there a way on the Nano to confirm this?

I have been to link and followed the instructions to no avail and will try the above again one last time!

Many Thanks

Graham

Still the same error!,

Hi,

The error indicates that the image doesn’t build on JetPack4.4.
Please help to confirm this with the following command:

$ cat /etc/nv_tegra_release

For JetPack 4.4, it should be R32.4.3.
Thanks.

Hi

This is the results of the command

Thanks

Graham

Hi,

Based on the log, you are using rel32.1.0, which is released a long time ago.

We don’t have any prebuilt TensorFlow v1,15,1 for the OS.
Would you mind to reflash your device with the latest JetPack first?

Thanks.

To my knowledge this was the latest version! Could you let me have a link to the latest version and I will gladly re-flasy!!

Thanks

Graham

Hi,

The latest OS version is rel-32.4.4.
Please reflash it from the SDK manager here:


Thanks.

Hello

I cannot install the SDK would this be because the Jetson nano only has 4GB RAM it in the requirements it says 8?

Thanks

Graham

Hello

Trying to install the SDK, I understand I need to do this on a linux PC not the Jetson. I have loaded virtual box with ubuntu installed but then it says no Jetson Nano connected!!?? Then My laptop blue screens and reboots. Is there another way to do this with out using the SDK Manger?

Many Thanks

Graham

Hi @graham.hilton23, for Jetson Nano you don’t need to use SDK Manager - you can simply flash your Nano’s SD card from a PC running Windows/Mac/Linux. You take your Nano’s SD card out and put it into your PC, and then use Balena Etcher from your PC to flash it. You can follow the procedure in the Jetson Nano Getting Started Guide here:

Hi,

Yes I understand this but I am being told the version I have is really old and may be the cause of the issues I have!

Cheers

Graham

You can update to the latest JetPack version by re-flashing your SD card with the latest image:

https://developer.nvidia.com/jetson-nano-sd-card-image

It is a little old L4T version.
If you want to run JetPack 4.4 on your JetRacer, you should use other image.
Houw about this?

1 Like