intelicopter

hi there, we're designing an autonomously controlled helicopter at uni (as you may have read in earlier posts) basically to initially take a series of waypoints we give it and move along them. At the moment it is unlikely we're going to have any vision or other complex peripherals just yet, but thats not to say in subsequent years they won't be incorporated. I have been scouting around for what computing power we will require in the machine. The PC 104 standard seems to be the way forward simply because so many other projects use it. Does anyone know why pc 104's are better then other embedded computer computers? The viper board (PC104) from arcom looks promising as it runs a 400MHz intel chip and is reasonable inexpensive, $400 or so. However looking at what other people have been using there is a broad range from a 33MHz 386 up to a 700MHz PIII!
Does anyone know how much processing power we are going to require?
Many thanks.
Intelicopter

greenboot

My Feedback: (1)

The original space shuttle had less computer power than the original IBM PC. I would start with that and add more as necessary.

Tom

steve69

As I read it you have said that neither you nor your collegues have any experiance with RC heli's.

Therefore in all honesty I think your looking to start your project from the wrong end.

First thing you need to work out is how a heli works, how it's put together and how to fly it and the control required.

No-one, no matter how experienced they are, can explain these things to you on a forum, so looking at what type of computer and what processor to use is at this point a near useless exercise.

Start with the basics first and learn how to fly, you wont be able to do any prototype testing if you cant assist the heli with manual inputs, have you ever even been near a large RC heli thats spooled up, it's not something you can just 'give' to the computer to control with no other manual interface (ie a normal transmitter)

I appriciate that you may already have these things in hand however you have made no mention of them in any of your posts.


Best of luck in your project, this post is not intended to try put you off of to make fun of you or your team but unless you have tried to fly a RC heli you will have no idea of the complexity of the control required to maintain a stable flight/hover

Jyrki

Howabout building a self controlling helium blimp?

By its nature its much more stabile to control, no serious vibration, plenty of time to make corrections, good payload, relatively safe to fly near people. Still challengin enough i believe.

If you manage to build a "Zeppelin" that holds reliable and still in a certain area with help of gps for hours or so, it would be great platform for aerial video or whatever purpose like replacing a mast in temporary use. For example carrying A WLAN-BRINDGE during a nerd camp. Or just as a eye catcher with lots of marketing surface.

A blimp would carry a gas generator, lots of fuel etc. I think you could control it with electric motors but use the gas generator to make electrics for them.

And a great benefit is also that you CAN CALCULATE the project parameters. Calculating heli parameters is difficult. But for a blimp you can for example solve the needed velocity so easy.

I bet you can easilly use sensors that behave odd in a heli. Like SMM acceleration sensors. If a sensor is slow but accurate, its nice for a blimp, but useless in a heli. You could even drive it with gsm short messages ) by giving coordinates.

Jyrki

Jyrki

I thought a second and must add a comment to my own post to make my point:

As a result, a reliable self controlling blimp would be much more valuated than a complex heli which just and just manage to hover a few minutes.

Jyrki

Jyrki

about vision:

detecting specially good contrast objects from ccd camera image should be simple. For example you can mark the landing point (or whatever point) with IR diode. mod the ccd camera to work at IR.

From the image you just pick up the max illumination coordinate.

I have no experiment nor details, but only the idea that you could also slowly blink the IR diode to send a "ID" of the point to CCD by rougly measuring the blink rate.

Jyrki

dave jones

Guys
The AMA election is coming down to the wire and we need to keep the momentum going if you would like to continue your work with UAVs and video enhanced flight then you need to get your vote in for Mr. Dave Mathewson if you have not cast your vote for the office of AMA president then please do it now.
Please pass this on to all of your modeling friends and your club members regardless if they are involved in autonomous or video enhanced flight or not, they make up the vast majority of our membership please ask them to visit Mr.Mathewson web site at www.mathewson4pres.com. and if they do not have a a computer then please ask them to vote for him in this very important election. Ask them to past this on to all of there modeling friends. It's very important that we reach as many members who have not yet exercised their right to vote. It's very important to point out in your email that the ballots need to be postmarked by no later then next Friday, November 12, 2004 and will only count if they're mailed to the address on the front of the ballot no postage required. You do not have to renew your AMA member to vote in this 2004 election. Please get those ballets in the Mail now. One last thing if we have to 4 more years of Dave Brown you could very well see your work with autonomous and video enhanced flight come to a screeching halt, even if you are not flying at an AMA site so please get those ballets in and vote for Dave Mathewson.
Dave Jones

killfile

The pc104 is a shrunken x86 architecture designed for general purpose applications. Embedded systems are typically specialized designs created to excute a specific function. In this case, you need a realtime architecture with fast sensor input, data processing and output control and you'll need a battery-powered lightweight package. An x86 IMO is less than ideal for these reqt's.

Consider a microcontroller/processor like the ARM cores and the Ti DSP's with particular attention to the input/output that you'll need.