2

Author Topic: 32-bit ARM robot controller board  (Read 1003 times)

0 Members and 1 Guest are viewing this topic.

Offline jsengTopic starter

  • Beginner
  • *
  • Posts: 2
  • Helpful? 0
32-bit ARM robot controller board
« on: May 11, 2013, 11:25:00 PM »
Hi all,
  I just launched a Kicstarter project for a 32-bit ARM robot controller project.  The board is called the Aithon board and has a STM32F407 with accelerometer/gyroscope, 2x5A h-bridges, LCD port, USB device/host ports.  Check out the link:

http://www.kickstarter.com/projects/590504545/aithon-32-bit-arm-microcontroller-board

Offline jwatte

  • Supreme Robot
  • *****
  • Posts: 1,343
  • Helpful? 81
Re: 32-bit ARM robot controller board
« Reply #1 on: May 12, 2013, 02:32:49 PM »
Because you posted here, I will give my un-varnished opinion of this idea:

In my opinion, there are too many boards like this already, each of which requires the user to cobble together a development environment of baling wire, bubblegum, and rusty nails.

Sure, summon-arm-toolchain may help, when it works, but the problem is not "I wish I could find a board with an ARM on it and some peripherals," the problem is "I wish the development environment was at least half-way sane."

For non-software-people, the Arduino environment is good, and it's portable across platforms. It also has libraries for the things most people want to do. This is a Good Thing (tm). If you want to tie into that ecosystem, you should use the same setup as the Arduino Due (an AVR Cortex) and peripherals that are already supported by the Arduino libraries.

For software people, avr-gcc + avrdude still beats most other setups in how easy it is to get going. For USB interfacing, there's also LUFA. Anything else I've tried has been like pulling teeth by comparison.
Also, consider the price point: The Raspberry Pi has an 800 MHz ARM with 512 MB of RAM running Linux, for $35. Add SPI/I2C peripherals and you're ready to go. The BeagleBoard Black has slightly higher specs, for $45. Again, running Linux, and has lots of I/O ports for peripherals.

In my opinion, adding on to an existing, well supported ecosystem, like Arduino Due (by making your board compatible,) or Raspberry Pi (by building a shield for the Pi, rather than a full board) would be a much better option to get some traction, and make it easy to get going!

Offline obiwanjacobi

  • Full Member
  • ***
  • Posts: 57
  • Helpful? 0
  • You can PIC any micro controller - what AVR!
Re: 32-bit ARM robot controller board
« Reply #2 on: May 13, 2013, 09:04:45 AM »
- never mind -
« Last Edit: May 13, 2013, 09:08:24 AM by obiwanjacobi »

Offline johnwarfin

  • Full Member
  • ***
  • Posts: 120
  • Helpful? 3
Re: 32-bit ARM robot controller board
« Reply #3 on: May 13, 2013, 11:20:03 AM »
i agree pretty with much everything j said. arduino, or more accurately atmel avr, does the job for nearly every one of my projects. including most that are unrelated to robots. but without knowing the level of technology required its hard to make such a blanket statement. avr is far from the most powerful or feature per dollar choice. it would fail miserably in things like machine vision or navigation on the level of say Grand Challenge trials. where it does excel is ease of use in both hardware and software which is what most in this hobby are after.

btw interesting to note the original pi WAS avr. just like the original paparazzi autopilot (these guy just dont know how to leave well enough alone).
« Last Edit: May 13, 2013, 11:21:40 AM by johnwarfin »

Offline paulstreats

  • Supreme Robot
  • *****
  • Posts: 1,381
  • Helpful? 21
Re: 32-bit ARM robot controller board
« Reply #4 on: May 13, 2013, 01:07:11 PM »
Sure if you guys want to get stuck in the dark ages, slate everything that comes along that you haven't learnt about from beginners hobbyist websites (probably here as far as your AVR business goes and probably most other robot related stuff goes).

 For those of us who do like to do something more advanced, these kind of things are interesting to hear about and how dare you comment on the lack of a decent toolchain or development system unless you know whats actually included and tried and tested it for yourself. Really sometimes its best to keep your comments to yourself. Even the toolchains that you were talking about (that aren't necessarily used by this product) might be simple enough to actually use, just because its something that you don't understand doesn't mean that others aren't going to.

 To the original poster, keep on with what your doing and dont be put off, disregard comments made by people who have no vision for the future, they'll still be playing with 8bit controllers and building 16 x 16 maps while the rest of us develop the things that are actually useful and realistic in today's world even if those people construct theyre replies that make it sound like they are the god of knowledge on all subjects (most of it probably googled to find a response in the first place).

Offline johnwarfin

  • Full Member
  • ***
  • Posts: 120
  • Helpful? 3
Re: 32-bit ARM robot controller board
« Reply #5 on: May 13, 2013, 01:48:24 PM »
some toolchains take days or weeks to get running and equal time to learn features, some with flash dongles that cost many time more than others. other toolchains take minutes to get going and architectures and instruction sets that can be learned just as quick. dongles that can be built for pennies (ie avr lpt). some projects warrant the difficultes others not so much.

there will always be those who must check out every new device and stay on the bleeding edge. live and breath datasheets and details that will never see practice. others need to get something done and take path of least resistance (and expense). its usually obvious whos whos within the first few sentences of a post.

Offline paulstreats

  • Supreme Robot
  • *****
  • Posts: 1,381
  • Helpful? 21
Re: 32-bit ARM robot controller board
« Reply #6 on: May 13, 2013, 03:15:15 PM »
nope not at all, I still use my PIC's and their associated IDE which is much simpler than the AVR but there really is only so far you can go with them. They are still used widely in my projects but for for things like making decent sized maps, decent AI, image processing, dynamic motion routines and the such they really dont have the capability so you will eventually need to use more modern devices not to be on the cutting edge but simply out of necessity. Sometimes even with software, I'll use something like PERL as a language to provide proof of concept because of its simplicity but when it comes to writing it up for actual use, I'll then convert it into a much more specified and higher end language that is appropriate for the task.

 From a beginners point of view, the toolchains and suchlike can seem daunting but the knowledge gained from experience will give you the ability to seamlessly use such systems without them taking weeks to set up. Just because somebody doesnt have real world experience or knowledge from experience doesnt mean that they can demean such things... really keep it to yourself, one day you might be using those systems yourself and try to help them out.

 One of the projects im involved with at the moment has recently gained a grant towards a robotics club of sorts where weve achieved funding that will allow all young people with an interest to build their own robots without cost to themselves following blue prints and fun guides made by myself and another member, were keeping the costs down so most things are being developed ourselves and built by the "students" and low end controllers are going to be used, but there will be options for people to develop their favourite area further so we are also looking at higher end things that we can make and develop toolchains that are going to be easy enough for teen aged students to follow. If i was less experienced and read that response then i would be put off from doing that, not only meaning that my own development suffered but also the development, skills and possibly future in robotics and engineering of the people I'll be guiding will suffer too.

It annoys me

Offline johnwarfin

  • Full Member
  • ***
  • Posts: 120
  • Helpful? 3
Re: 32-bit ARM robot controller board
« Reply #7 on: May 13, 2013, 05:30:30 PM »
my own development suffered but also the development, skills and possibly future in robotics and engineering of the people I'll be guiding will suffer too.

well... that is unfortunate but doesnt sound too surprising. (jk!)

weve followed similar paths. i started with pic when the company was general instruments and they were mask programmed for stb. 40 years ago. then avrs when at90s was new. not a single other person on bbs's or in the real world had heard of them. pic reigned supreme and even though pic and 8051 were my forte at the time when avr came along i switched in a ny minute. it was like a breath of fresh air. dont get the the same feeling for arm. st, and most of the other new chips.

currently my non-avr tools included both 8 and 32 bit discoveries, arm-tiny-usb, and sam setups so granted not as thoroughly indoctrinated as youself but ive spent enough time there to get the picture. your comment on pic is not too surprising either. im forced to use mplab/pickit3 but by no means consider it "simpler" than arduino. or as well supported. and even ignoring the 10 cent lpt avr interface note that you can buy 10 usbasp for the price of one of the cheapest microchip dongles. it seems like they spec an  different incompatible flash voltage and protocol with each new device.

as primary professor engaging 4 semesters of ee and 2 cs ive had opportunity to try various embedded proto kits and see the result. imo swamping newcomers with 32 bit flash and flare causes most to glaze and dash. on the other hand great interest and sparks over arduino. although not particulary a fan myself. but then you must realize, like the majority of robot hobbyists, few get involved with cameras, heuristics, kalman, pid, or other sophisticated motion/sensor processing. like most of the guys in this forum they just wanna build a sumo or pick up a block.

if you read my first post you should realize im more moderate than youself or jwatte when it comes to seeing the "big picture" impacting most in this hobby.

Offline jwatte

  • Supreme Robot
  • *****
  • Posts: 1,343
  • Helpful? 81
Re: 32-bit ARM robot controller board
« Reply #8 on: May 13, 2013, 10:34:26 PM »
Sure if you guys want to get stuck in the dark ages, slate everything that comes along that you haven't learnt about from beginners hobbyist websites (probably here as far as your AVR business goes and probably most other robot related stuff goes).

But that's not what I suggested. I suggested using the Arduino Due compatible ARM chip from Atmel, which is pretty similar to the STM chip suggested on this board. The reason being that tool chains are better supported.

Quote
For those of us who do like to do something more advanced, these kind of things are interesting to hear about

You're missing the point. Progress is great! How is this specific board different from an Arduino Due, or a TI Stellaris Cortex-M4, or any other similar board that's come out in the last two years? I mean, except from being bigger, because the peripherals are hard-coded onto the board, rather than modular, and the board being produced by a smaller company with a smaller community and higher prices? If the board really is original and special, then it would be great! But the kickstarter doesn't talk much about that, so I have to assume it isn't. (It says it includes "easy to use libraries" but what that means, who knows?)

And the Arduino Due is seeing significant, hard competition from boards with 10x the capability and half the price, namely the Raspberry Pi and the BeagleBone Black. That's really the competition here, and I'd like to understand how this product is superior than those ecosystems. The kickstarter documentation doesn't give me any reason to believe it is, and that's the developer putting the best foot forward.

(Btw: The main reason the RPi and BBB are easier to develop for because they come with self-hosted tools.)

Quote
and how dare you comment on the lack of a decent toolchain or development system unless you know whats actually included and tried and tested it for yourself.

Note that I commented on the tool chains I've seen for ARM, not on the specific tool chain. If there's a magic solution that's suddenly much better than the TI, Atmel, and open source hackers of the world can produce, I'm all ears! But that's not mentioned in this Kickstarter.

Quote
Really sometimes its best to keep your comments to yourself. Even the toolchains that you were talking about (that aren't necessarily used by this product) might be simple enough to actually use, just because its something that you don't understand doesn't mean that others aren't going to.

Personal attacks generally reflect more on the attacker than the target. Although, you seem to not have read the questions and feedback I actually provided, so it's not clear we can actually have a productive discussion.

Offline jsengTopic starter

  • Beginner
  • *
  • Posts: 2
  • Helpful? 0
Re: 32-bit ARM robot controller board
« Reply #9 on: May 19, 2013, 11:45:38 PM »
At least I got a discussion going here. :)  We wanted to put together something a little more integrated than an Arduino.  It is true that the toolchain is not as straightforward as Arduino, but we hope that our libraries can overcome some of the startup difficulty.

Offline jwatte

  • Supreme Robot
  • *****
  • Posts: 1,343
  • Helpful? 81
Re: 32-bit ARM robot controller board
« Reply #10 on: May 20, 2013, 11:19:13 AM »
I understand about the ease of use of having integrated peripherals, from a hardware point of view.
But why would I want to use this board, rather than a Raspberry Pi or BeagleBone Black, if I really need the ARM performance?

 


Get Your Ad Here