Re: [Gumstix-users] Advice on Gumstix

[Dave H. <dhy...@gm...>]

Hi Liam,

On 9/19/06, Kilmartin, Liam <lia...@nu...> wrote:
> Hi,
>
> I am responsible for running an under-graduate project course at a university and I am interested in getting advice on the most suitable Gumstix platform to use for this module. The main use of the platform will likely be basic robotic projects (e.g. simple autonomous vehicle with collision avoidance). The students doing this module have reasonable micro-processor experience but mostly assembly language based around an 8051 clone as well as good general C programming skills.
>
> The robostix - wifi pack seems to be close to a good platform for us to use but my biggest concern here would be the probably difficulties the students (particularly weaker ones!) would encounter with the developing software to run of the two different processor platforms (ARM on motherboard and Atmel on robostix expansion) within the system. I fear that there may be a substantial learning curve associated with this approach in terms of familiarisation with the system architecture and the various development tools.

Since the robostix can be programmed in C, the real effort will be in
learning the AVR architecture. You may wish to take a look at the SRS
Workshop Bot, and have a agnder through the manual for it. A big chunk
of the manual deals with the actual assembly of the circuit board,
which you can largely ignore.
http://www.seattlerobotics.org/WorkshopRobot/index.php
Page 19 starts with a bunch of good slides for programming the
ATMega16, which is a subset of the ATMega128 used with the robostix.
There are also a number of sample programs available, which could
easily be adapted to the robostix.

There is also a bunch of sample code available for the Robostix, and
I've published the source code for one of my robots:
http://www.davehylands.com/Robotics/Stinger/

> A possible alternative, if I understand the overall architecture correctly, would be to replace the robostix module in the above pack with a breakout gs expansion board to expose the GPIO pins for the processor. Obviously with such a solution, the students would then be required to built their own motor control/sensor interface hardware to be managed by ARM based application DIRECTLY through these GPIO pins. I believe a 24 pin interface provided through the breakout board would be sufficient (assuming there most/all of these pins are configurably input or output) for our projects. Clearly, the disadvantage of this approach is the added complexity of hardware development required from the student. Perhaps I have not correctly understood the functionality of the breakout gs expansion card here is which case this may not be an option ???

Personally, I think it would be much simpler to develop the code on
the robostix. You don't need to deal with linux, and all of its
constraints. The ADC and PWM will be easier on the robostix as well.

In fact, you could probably get away with just using the robostix by
itself until you reach the point that you want the bots to communicate
with each other (over Wifi).

In any event, you'll still need an H-Bridge, although you could pick a
nice simple one like the L293D or the SN754410 which is pin
compatible. This is the same H-Bridge used on nthe SRS Workshop Bot
and is suitable for driving small motors.

-- 
Dave Hylands
Vancouver, BC, Canada
http://www.DaveHylands.com/