Subject:
|
Re: A robot who knows his position (fwd)
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Fri, 30 Apr 1999 16:13:19 GMT
|
Viewed:
|
849 times
|
| |
|
|
In lugnet.robotics, lego-robotics@crynwr.com (Ralph M. Deal) writes:
> On Fri, 30 Apr 1999, Mike Moran wrote:
>
> . . .
> > Now I come to think of it, if you allowed the light sensor to move such that
> > it could follow the head of the arrow, then we could arrange it so that the
> > light sensor starts off above the bright arrow. However, when the arrow • moves
> > as a result of movement then the light sensor assembly starts "seeking" • right
> > and left until it encounters the bright arrow again. Since the assembly • knows
> > how far it has travelled, and in what direction, it can give a reading on
> > bearing. Unfortunately this would require use of a motor and some fidgeting
> > around with some code to drive the thing. However, it sounds like it may be
> > do-able and useful.
> > Can anyone see any problems with this?
>
> Yes. It is unnecessarily clumsy and heavy and would require considerable
> computer attention to track rotation successfully.
That sounds like a challenge to me! ;-) I may just try and build something to
prove it can be done or can't be done easily. I believe the attention required
from the computer could be minimal. A first go at the algorithm would be:
#define INCREMENT <some time>
#define CIRCUMFERENCE <some number of increments>
sub moveLeftAnInc {}
sub moveRightAnInc {}
int aboveArrowLightLevel;
sub aboveArrow {}
int direction;
sub init {
// find the arrow
while (!foundArrow) {
moveLeft();
}
direction = 0;
aboveArrowLightLevel = <current level>;
}
int tmpDirection;
int seekTo;
int foundArrow; // 0 = false, 1 = true
int lostArrow;
int count;
task trackArrow {
init();
foundArrow = 1;
lostArrow = 0;
while (!lostArrow) {
wait (!aboveArrow()); // dunno if you can do this in wait
// seek the arrow
tmpDirection = direction;
seekTo = 1;
foundArrow = 0;
while (!foundArrow) {
if (abs(seekTo) >= CIRCUMFERENCE) {
lostArrow = 1; break;
}
// look left
while (count != seekTo && !foundArrow) {
moveLeftAnInc();
if (aboveArrow()) {
foundArrow = 1;
}
count += 1;
}
if (!foundArrow) {
// look right
< ... similar to look left ... >
}
seekTo *= 2;
}
}
}
Thier are some bugs in this impl. but basically its: first sample to find where
the arrow is, then as soon as you lose it, look further and further away until
you find it. It doesn't matter *too* much if you missed an intermediate value,
since you still know where you are pointing *know*. I think I'll have a go at
this at the weekend.
> I believe digital
> encoding would be better.
Of course it would, but beggers can't be choosers; I, personally, don't know
enough about electronics to build things from a circuit diagram and don't have
enough dosh to buy some of the sensors that have been mentioned.
|
|
Message is in Reply To:
 | | Re: A robot who knows his position (fwd)
|
| On Fri, 30 Apr 1999, Mike Moran wrote: . . . (...) Yes. It is unnecessarily clumsy and heavy and would require considerable computer attention to track rotation successfully. I believe digital encoding would be better. Anyone know how one determines (...) (25 years ago, 30-Apr-99, to lugnet.robotics)
|
6 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
