By: Gillerire

Gillerire — Thu, 01 Jan 1970 00:00:00 +0000

Nope, but Kym’s comment below does

By: ash

ash — Thu, 01 Jan 1970 00:00:00 +0000

Holy crap dude, the website has blown up. WTF???

Does this comment take up the whole screen for you??

By: PirCapAdi

PirCapAdi — Thu, 01 Jan 1970 00:00:00 +0000

[I've just edited this to try to correct some craziness that happened at the bottom due to embedded object].

Most clubs seem to have their Magellan comps on paved roads or in parking lots, while the Robogames has it in a park. I think the park setting is much more realistic than these, more obstacles, uneven terrain etc. In a parking lot like that you could probably do the course with just gps, and a cam.
http://curry.jandmworks.com/robogames-2008.html

Hmmm... my YouTube object is not working? Worked in the preview...

By: kymhorsell

kymhorsell — Thu, 01 Jan 1970 00:00:00 +0000

I think the problem may be — I was replying to and article that had a "funny" embedded object. I commented it out and everything should be now cool.

Someone said:

"Hmmm… my YouTube object is not working?
Worked in the preview…"

And I replied with:

Welcome to Drupal!

Sure, car-parks (or even going around the block is kid’s play
compared with a grassy knoll or twisty turny paths all looking the same
.

I can imagine the problems a nice, neat inertial nav system will
have when a small wheelbot starts slipping and sliding or lurching,
falling or tumbling in a "real world" course.

I’m hoping making a large wheelbot will help slightly with the scale
of obstacles in some of the robomagellan courses — even if I end up
making structural components out of foam to fit within weight rules.

But GPS isn’t exactly a "saviour" for nav problems. It has limitations and can even cause its own problems.

In the vid I pointed at you can see the bot apparently gets confused
and dashes across the diag when I *think* the object of the exercise
was to do a square (i.e. go to the nearest non-visited cone until none
left).

The size of this couse would limit the use of GPS that has errors of
10 m or so when working perfectly, and dozens of metres when not
working.

Someone ("Jeff") pointed me to a track he’d made of a GPS vs his real
path around a 1/2-km square on his property. The GPS was
(systematically) off by 50 m in parts.

Someone else ("Ross") has told me large errors can happen if (e.g.)
the GPS antenna is too close to the ground — as on a wheelbot.

So GPS is like the "long range sensor" of navigation, albeit if
smoothed properly can be used as the "only" nav instrument for UAV or
some other kinds of work where 10 m is small bikkies compared with avg
over-ground speed.

Comments on: Truckbot mission #3

By: Gillerire

By: ash

By: PirCapAdi

By: kymhorsell