Matlab support for Raspberry Pi 4

Hey,

I'm making a robotics project for my BSc and was wandering when could we expect support for Raspberry Pi 4? This page said that RPi's are currently supported only up to 3B+, 4 being quite new is not even mentioned.

On the other hand, are there some better or equal SoC for robotics projects that are supported by matlab?

Thanks in advance!

[Alessio Gagliardi]

I am interested as well,

There is a plan for releasing the Raspberry Pi 4 Support Toolbox? Thanks in advance.

Alessio

[Rik]

Even if sometimes staff members look on this forum and post from time to time, the help on this forum is mostly provided by people who are doing this as a volunteer. You can contact Mathworks support directly with the Contact Us button at the top of the page if you want an official answer.

Answer 1

[Prasanth Sunkara]

Hello Pawel,

Raspberry Pi 4B is a significant upgrade over earlier versions. As of today(Sept 2019), the latest release, MATLAB R2019b supports upto 3B+. The development team is actively looking into the request for 4B support and it will be made available in one of the future releases.

Thanks,

Prasanth

[Florian Bayer]

Hello Prasanth,

is there any update regarding the support of Raspberry Pi 4B ? Will the Raspberry Pi be supported in the next Release 2020a ?

Thanks,

Florian

[Prasanth Sunkara]

Hello Florian,

There is a workaround for supporting 4B on R2019b.

It is based on "OS customization" during "Hardware Setup" process.

Steps:

1) Flash the SD card with latest Raspbian Buster image from https://www.raspberrypi.org/downloads/raspbian/. (The shipping Raspbian image with R2019b Raspberry Pi support package is based on Debian Stretch, for Pi4 Debian Buster is required. )

2) Install a fresh Raspberry Pi support package

3) Run the Hardware Setup wizard. Please see the below link for more info on starting hardware setup.

https://www.mathworks.com/help/supportpkg/raspberrypi/ug/install-target-for-raspberry-pi-hardware.html

4) In the hardware setup screens, select "Raspberry Pi 3B+" and then select “OS Customization Workflow". Please see the below image.

5) Click 'Next' and follow the instructions provided on the screen. In the OS customization workflow, the linux libraries and packages required by the support packages are installed.

6) Once the process is done, execute the below line on MATLAB to ensure the installation is successful.

>> r=raspi('IP_ADDRESS','username','password')

Note that you may see that the above command may not return the board name as Raspberry Pi 4B. If you are using simulink blocks of raspberry pi, please select the board as "Raspberry Pi3/3B" as a workaround.

We need to wait till March 2020 for an update on the official support for 4B. Until then, I hope this workaround helps.

Feel free to comment, if you face any issues in this process.

Thanks,

Prasanth

[Florian Bayer]

Hello Prasanth,

thank you very much for your fast reply. I will try it with these information.

Thanks,

Florian

[takamitsu hatakeyama]

Hello Prasanth

I want to ask in case of using MATLAB2018b.

counter measure you mensioned above is still effective in case of using MATLAB2018b?

Thanks,

Takamitsu

[Pawel Gakan]

Hello Prasanth,

thanks for answering my question! I got around those limitations by running ROS network on my RPi with nodes generated from simulink.

However, what my solution lacks is real time support. Will using your solution guarantee RT?

Thanks,

Paweł

[Prasanth Sunkara]

Hello @Takamitsu,

The above suggested OS customization workflow would work for versions from R2018b - R2019b. Please try it out and let us know if you face any issues.

Hello @Pawel,

ROS nodes, to put it in a simpler way, are event driven. However if you deploy a simulink model with non-ros blocks on to Raspberry Pi, thery are scheduled to run at the sample time Simulink allots during the compilation of the model.

But since Raspberry pi runs a non real-time OS, the OS may not always guarantee the Simulink generated executable runs at the specified sample time. But in most cases the OS is likely to respect the specified sample time. It varies based on factors like, the block's algorithm computational complexity vs sample time specifed for that block, other programs running on the board etc.

Hope this helps!

Thanks,

Prasanth

[Pawel Gakan]

Hello @Prasanth,

Yes, this the exact behaviour I've encountered.

I've compiled and successfully installed preemptive kernel on RPi. Do you think, that this is something I can take advantage of by using MATLAB/Simulink?

Thanks,

Paweł

[takamitsu hatakeyama]

Hello @Prasanth

Thank you for your repry.

I have try raspbery pi4 on MATLAB2018b.

It work well for now.

If I found any informative issue, I'll share infomation.

Thanks,

Takamitsu

Answer 2

[Laalitya Acharya]

Prasanth Sunkara - I have pi3 and pi4 and both of them are getting errors of

$sudo apt-get install matlab-rpi

Reading package lists... Done

Building dependency tree

Reading state information... Done

E: Unable to locate package matlab-rpi

I am following instructions on https://www.mathworks.com/help/supportpkg/raspberrypiio/ug/connect-to-raspberry-pi-hardware-board-in-matlab-online.html

I am trying to do this for matlab online.

[Christopher Hamilton]

Looks like the package installer calls for an older version of libgnutls-openssl. Version 10u4 is the only one on the repo. Kinda new to linux, is there a way I can download/install the older version?

E: Failed to fetch http://raspbian.raspberrypi.org/raspbian/pool/main/g/gnutls28/libgnutls-openssl27_3.6.7-4+deb10u3_armhf.deb 404 Not Found [IP: 93.93.128.193 80]

[Prasanth Sunkara]

If you are using Desktop MATLAB software(not MATLAB Online), then please use either MathWorks Customized image or Customize your existing Raspberry Pi OS with the additional requirements.

For more info, please see: https://www.youtube.com/watch?v=32ByiUdOwsw&ab_channel=MATLAB

Please do not use "matlab-rpi" linux package installation for Desktop MATLAB. It is only applicable for Online MATLAB.

Answer 3

[Tohru Kikawada]

Now MATLAB and Simulink supports Raspberry Pi 4 Model B from R2020a.

https://jp.mathworks.com/hardware-support/raspberry-pi-matlab.html

[Christopher Hamilton]

Looks like the package installer calls for an older version of libgnutls-openssl. Version 10u4 is the only one on the repo. Kinda new to linux, is there a way I can download/install the older version?

E: Failed to fetch http://raspbian.raspberrypi.org/raspbian/pool/main/g/gnutls28/libgnutls-openssl27_3.6.7-4+deb10u3_armhf.deb 404 Not Found [IP: 93.93.128.193 80]