Microcontroller Project

[tjump]

Greetings to the Group,

I am currently developing a pre-engineering curriculum to be used at the high school level. The curriculum is fairly extensive, and has been deployed in part over the last 5 years. The overall program is a three year, three tier system that begins with the LEGO platform and advances to full embedded systems (PC104 and mini-ITX). However, there exists a gap between the LEGO platform and the embedded systems that I am looking to fill, thus the query that follows.

I need to put together a microcontroller package that incorporates several elements (listed below); and I am starting with the idea that this will be a multi-microcontroller system. Any suggestions and links to different pieces of this puzzle would be apprecitated.

Requirements:

Wireless connectivity –
Remote control options for handheld controllers (game controllers and/or R/C radios)
Remote control from a PC
Data exchange between a PC and the robot, possibly including data processing on the PC
Video streaming to the PC

Motor Control –
R/C PWM and servo motor

I/Os –
Analogue and Digital sensors (communication protocol?) One element we really need here is the ability to take scan data and generate a map/profile image of the “room” (small size, maybe 1m boundary). Our advanced robot uses an Hokuyo URG.

Other sensors we employ: IR, Thermal, CO2, Ultrasonic, Encoders. We are also working on adapting a laser mouse for position tracking.

[hudbrog]

Wireless connectivity –
Remote control from a PC

Nice, cheap and interesting way - rf tranceivers in ISM band. For example - TI CC2500. Easy to use, low external component count. Module with PCB antenna will cost about 3$ (not counting the cost of pcb itself).
Easier and a little more expensive - bluetooth modules (BTM-111 for example). Pictures of both could be found here:
http://roboforum.ru/download/file.php?id=6098

Remote control options for handheld controllers (game controllers and/or R/C radios)

Well... don't know if there are implementations already, but it is pretty easy to do. Maybe you should search for combat robots, afaik they use off the shelf rc hobby controller for them.

Data exchange between a PC and the robot, possibly including data processing on the PC

depeding on amount of data - defferent solutions. Small amounts - see above - cc2500 or btm-111
large amounts - bulk usb connections

Video streaming to the PC

Easiest way - off the shelf wireless cams... any other - much more expensive.

Motor Control –
R/C PWM and servo motor

Basically any uC + h-bridge(depeding on motors - l293d, mosfets or something like VNH3SP30-E)

I/Os –
Analogue and Digital sensors (communication protocol?)

Just plain uC..

One element we really need here is the ability to take scan data and generate a map/profile image of the “room” (small size, maybe 1m boundary). Our advanced robot uses an Hokuyo URG.

These are not trivial.. never done smthing like that..

[tjump]

Thanks for the above, but I am really looking at a more specific target answer (i.e. microcontroller brand x). Which platform is going to give me access to all my needs. I am focusing quite a bit of attention to the Axon, but there are some others that may be a better fit. I just don't have any development time/experience with the product range so I'm hoping the group members can share their experience with some of the options to help narrow my search. i just don't know the capabilities of each microcontroller to make a quick, informed decision.

Thanks,
T. Jump

[hudbrog]

Exact uC - Atmel AVR (doh.. I'm begining to hate myslef)
Board.. I really don't know any popular american brands for boards. Arduino sounds nice and popular.. with moto shield and stuff like that. For all board I know it will be a bit difficult with wireless communication.
Or you can buy some from Russia(several possible variant which will suit your needs almost perfectly - i.e. with two H-bridges onboard, with wireless onboard and stuff like that). For that PM me, I'll show you boards you might like and can help arranging stuff.

[ArcMan]

There are 2 camps on this board.

One is the AVR, as you saw.

There's also the Microchip PIC.  Typically I buy PIC proto boards from Sparkfun, which include the uC and crystal, along with RS232 or USB interface.

The AVR and PIC have very similar capabilities.  One of the big differences is the C compiler.  The AVR's is open-source.  The PIC's is commercial.  Mind you, I'm only talking about the good compilers.  I use the CCS compiler, as do most PIC people.

I'm not aware of any all-in-one solutions like you may be looking for, but I've never looked for one.

[Admin]

For wireless data transfer, use bluetooth:
http://www.societyofrobots.com/electronics_bluetooth_robot.shtml

Its really easy to implement, and it worked great on my ERP.


I already sent this in an email to you, but I'll repost it here for others that are interested:

The competitive teams for USAR use FGPA's, PC104 stacks, laptops (with
interfacing microcontrollers), and ARM processors. Its worth looking
in to, but for highschoolers all of that is probably a bit too
advanced.

My Axon can handle all of those requirements, and is the easiest to
learn option. However my concern is that the Axon processing speed
(16MHz at 16 MIPs) and memory (8kb RAM) is limited for what you may
want to do.

If you use PIC's or AVR's, you'll be limited on what algorithms you can use. BTW, what algorithms do you plan/hope to implement? Were you looking into SLAM? Check out this forum post that demonstrates VSLAM (V for visual), it might have helpful info.

[tjump]

Thanks,

My work e-mail is down so it may be a while before I get anything coming across.

I did see the Bluetooth article on the SoR site and it put the final pieces together for me, so I think the Axon is the way to go. I'm pretty sure it will meet all my requirements. I would like to know if there is a particular Bluetooth module that interfaces directly to one of the UART terminals on the Axon rather than having to add a serial adapter.

As for mapping, we have steered away from SLAM, even on our advanced robot. It poses a problem that is too advanced for my students/my knowledge set/our time limitations and broader learing goals. We have developed a snapshot/cut and paste process whereby we take a stationary scan and plot the map in a grid (the on board compass always aligns scans to north). As we collect multiple scans we just have to drag them together (x and y translation only since the scans always align to north...no rotation of images necessary). We then drag victim tags into the map from a pulldown menu. Still quite a bit of programming and data management; certainly impressive at the high school level. Our maps still outscore many of the SLAM teams at RoboCup.

Anyway, I'm going to bring a batch of the Axons into the lab and we will see what the kids can do with them. I will let you know about speed and memory issues.

Cheers,
T. Jump

[Admin]

I would like to know if there is a particular Bluetooth module that interfaces directly to one of the UART terminals on the Axon rather than having to add a serial adapter.

Any bluetooth you get should already have serial rx/tx on it. Check sparkfun.com for a bunch of them. You can directly connect them to the UART of the Axon without any converters.

We have developed a snapshot/cut and paste process whereby we take a stationary scan and plot the map in a grid

A great first step! Optionally, you can have your robot measure its motion using encoders and then move the robot position within that map. New scans overwrite old scans. If you don't do slam, error will exponentially build up. However a map with 50%+ error is better than no map at all! Its really about what your students learn and not how they place