Subject:
|
Dead reckoning (was Re: Rover Programming)
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Thu, 27 Jun 2002 09:27:40 GMT
|
Original-From:
|
Simon Brooke <SIMON@JASMINE.ORG.antispamUK>
|
Reply-To:
|
SIMON@JASMINE.ihatespamORG.UK
|
Viewed:
|
794 times
|
| |
|
|
On Thursday 27 Jun 2002 8:30 am, you wrote:
> Simon Brooke wrote:
> > That's one solution, but it isn't essential. The other solution is
> > to assume every square is empty until you've explored it.
>
> So how do you remember what you've already explored?
A systematic search pattern is one solution. Not being too anal about
it is another - random walking the domain will eventually explore the
whole of it, although increasingly inefficiently as time goes on.
> Still, I think the poor performance you can get from dead-reckoning
> is what kills you. This kind of thing would be fine with a kind of
> GPS type system - but with a bunch of rotation sensors as our only
> navigation mechanism, I think we're doomed.
It's killing me. My first rotation sensor robots were inaccurate, so I
started again with a new design with the rotation sensors on idler
wheels rather than on the driven wheels. That was more inaccurate. So
I've gone back to having the rotation sensors geared off the driven
wheels, and I've at last achieved a robot which can (on a polished
wooden floor) walk a one metre square and return to within 10cm of it's
starting position. Wow! Achievement! However, like my first attempts,
it turns more than the calculated amount on polished floors (presumably
the inertia of the turn causes slip) and less than the calculated
amount on carpet (presumably because of drag).
So I'm now looking at software solutions that reduce inertia - i.e.
gradually ramping up and down speed in turns. We'll see how this goes.
I'm currently using the 8.5 centimetre wheels on my dead reckoning bot.
Has anyone tried building a dead reckoning bot with tracks? How did it
perform?
--
simon@jasmine.org.uk (Simon Brooke) http://www.jasmine.org.uk/~simon/
;; An enamorata is for life, not just for weekends.
|
|
Message has 2 Replies:
 | | Re: Dead reckoning (was Re: Rover Programming)
|
| (...) Yes - managing accelleration is a 'must'...both in speeding up and slowing down - and easing into turns. An abrupt change of speed will generally result in wheel slip. (...) It was good in a straight line - but *hopeless* in a turn because the (...) (22 years ago, 28-Jun-02, to lugnet.robotics)
|
Message is in Reply To:
| | Re: Rover Programming
|
| (...) So how do you remember what you've already explored? I think you *need* three states if you want your robot to curiously explore it's environment without continually re-exploring areas that it's already visited but which proved to be empty. If (...) (22 years ago, 27-Jun-02, to lugnet.robotics)
|
24 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
