Today I talked on a social event with a number of professors for control engineering about Matlab and Julia. They all told me, they would love to replace Matlab with Julia, but they have dSpace control hardware that only works with Matlab/ Simulink.
What could be a good answer?
I guess that the assumption that Matlab/Simulink is required to use dSpace hardware is wrong:
To configure the systems, you can use the dSPACE software tools ConfigurationDesk, ControlDesk, and the Bus Manager in combination with commonly used tools such as MATLAB®/Simulink® from MathWorks, FMI-based models, or Linux-based frameworks such as RTMaps or the Robot Operating System (ROS).
I have to find out more details how we actually use this hardware…
Hi,
Found this topic as I’m having same problem as the users you talked to. Use Julia a lot but when it comes to using our dspace hardware I have to go back to matlab/simulink and that’s a nightmare of licenses and so on … So would be happy to help ! I great would be some flags on the new juliac to compile direct code for dSpace platforms.
What I know is that Simulink generates some C code with specific libraries and compile it using gnu compiler (for some platform).
BR
Yves-Julien
This package creates C code from LTI systems: GitHub - JuliaControl/SymbolicControlSystems.jl: C-code generation and an interface between ControlSystems.jl and SymPy.jl
Cross compilation of Julia code is probably a few years away, though.
In some situations it’s possible to use emulation to perform cross-compilation, i.e., to emulate the target platform on your desktop/laptop PC. There was a related talk on juliacon this year
Why use dSpace, when linux is real-time now? PREEMPT_RT - Wikipedia
Maybe it is possible to use emulation to cross-compile to a Raspberry Pi with preempt_rt?
dSpace boxes have very nice hardware inputs- and outputs. Some even have FPGA chips for fast data processing. And many universities have them already and do not have to buy them. But if you have to buy them they are expensive.
The problem is somewhat similar with Speedgoat systems.
Yes speedgoat faces same problems plus it is even more expensive and AD/DA are somewhat noisier.
For dSpace a way would be to use FMI.jl to generate an FMU based on julia code then use configuration desk (dspace proprietary) to link I/O, compile and load it to hardware. But for this to be a solution it needs FMI.jl to be able to export “se” models which it does not seem to be planned as it does not appear in the development table depicted here: Introduction · FMI.jl
…BR