ACU Development, want some feedback

Hey guys, I'm still working on the FPGA based ACU, but I'm having a few troubles.  Since I know no one else here is into FPGA design, I wanted to feel out the crown for an ARM based solution.  Quad or Octa core, beefy enough to compare to FPGA IO's, etc.  This will still be a custom SBC for MRL, so feel free to chime in.  I'm moving into a new house this month and once I close, I'll have my bench set back up for testing.  I'm still leaning towards FPGA backend, but it may be unecessary, at least for now....  So, in closing, my original design was FPGA connected by USB to a SBC/laptop/etc, and now I"m considering FPGA+ARM on the same board, in essense, a SoC design, using the FPGA and a ARM together, on the same board.


Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
kwatters's picture

verilog, vhdl, xilinx, altera...

Hey Cyberhedz,

I've done a bit of FPGA stuff.  Just curious, what do you need an FPGA for?  Could a simple microcontroller do what you need?  Arduinos are pretty good at controlling servos and other sort of devices..

That being said, I think you could probably create some nice control structures that would control the servos for you, but in the end, you're still limited by your input/output pins.  

If it's I2C support you seek.. Arduino's have a library for that.. 

FPGAs are cool, but there might be an easier way... 

 

 

cyberhedz's picture

Refresher on the ACU

Thanks for the feedback, it's been a while so I'll just go ahead and refresh everyone on the "goal" of the ACU (Android Control Unit).  The main focus is th create a "motherboard" for the Inmoov and other humanoid robots.  I chose FPGA for teh robustness and teh capability to grow, expand, and sheer power.  Currently I need high speed UART, I2C, and general IO, and while yes an arduino can..it doesnt mean it should.  If you want to read my full rant, its in my blog here.   Back to the topic at hand though.  If I am creating a motherboard for us, it needs to be self dependant, powerful, reliable, and expandable.  I'm starting to really lean towards a hybrid ARM+FPGA.  I had looked into the Cyclone V SoC, but with only a dual core A9, I dismissed it and am now thinking about sticking with the Cyclone IV FPGA, and using the ARM Cortex A-17 for the SOC portions.  As far as interconnects, well, that depends on the ARM really.  If I can find a way to use PCI-E, that is the preferred method, but until I get datasheets and find a distributor, its all just scratchpad.

cyberhedz's picture

Not dead, just not around

Just updating the folks here.  I'm still working on the ACU, but with my recent temporary move to Australia, I had to put soem projects on hold.  I have been whiteboarding and engineering some new ideas for the ACU and right now I'm debating between using the Cyclone V and the Tegra X1.  I'm still thinking for a true "brain" the FPGA will be better due to its ridiculously fast backbone.  IO's are a plenty on both, adn if I went with teh X1, I would end up having to pair an MCU with it to get the same nuber of IO's that the Cyclone has onboard.  Always happy to hear opinions, and would love to find a VHDL engineer willing to work for next to no pay :)

Ray.Edgley's picture

Just looking into FPGA's as well.

I know this is an old thread, but.

I noticed that Arduino have now released a new board, The Arduino MKR VIDOR 4000

https://store.arduino.cc/usa/mkr-vidor-4000

This is an Arm Cortex-MO+ processor (ATSAMD21) coupled with an Intel Cyclone 10CL016 FPGA.

Looking into FPGA's and I can see some major potential for ML application.
The Intel Cyclone is a relatively small FPA with 16K Logic Elements.
However if you consider the logic elements all process in parralell, there is scope of increadible overal processing speeds depending on how many elements are required to contruct a Neural Network Node.

Out of curiosity, are you still in Australia and if so, what part?
I'm in Central Victoria.