To LUGNET HomepageTo LUGNET News HomepageTo LUGNET Guide Homepage
 Help on Searching
 
Post new message to lugnet.roboticsOpen lugnet.robotics in your NNTP NewsreaderTo LUGNET News Traffic PageSign In (Members)
 Robotics / 18301
18300  |  18302
Subject: 
Re: Ant pheromones
Newsgroups: 
lugnet.robotics
Date: 
Fri, 5 Jul 2002 13:52:58 GMT
Viewed: 
678 times
  
In lugnet.robotics, "Jonathan Spitz" <jspiar@yahoo.com.ar> writes:
I have built my robot but I'm having some problems with the light sensor.
Can you give me some advice on wich colors to use to avoid problems or how
to put the sensor. Maybe it's just a problem of the RCX but I can't get it
to see a thin line at a relatively low speed.

OK. Two factors here:
1) Status of pens - colour, newness etc. Reflectance of the base surface -
glossy, etc.

I am using a white board for my playing surface, which is quite reflective.
Furthermore, dry wipe pens are translucent - otherwise OHPs etc wouldn't
show colour, they would show black! -  so, I found that there was little
difference between the line and the bare board. Avoid red and green if you
can, and go for blue and black, wihch are best. With fresh pens with good
nibs and with some pressure down on the board, there was just enough of a
mark for detection with careful calibration.  My ant robot actually has a
platform that raises and lowers the pen and pushes down almost so hard that
the robot stops moving; consequently, it makes a great clear line, but slips
a bit when turning.

2) Position and programming of the sensor - relative or absolute
calibration, etc

For one experiment I used three light sensors in a V shape at the front of
the robot, with about 5mm clearance and just in front of the turning point
of the robot. It was tracked and this nearness to the pivot allowed a small
change in sensor orientation when the robot turned. Too far from the centre,
the sensors move rapidly away from the line too quickly. This is less of a
problem for straight motion, but can still happen. Try gearing down your
motors to make smoother smaller movements. And consider where your light
sensors are with regard to the centre of rotation. Also, you don't say how
many light sensors you are using. I would suggest that two is almost a
requirement,, as you can traverse the line with better control. I know that
it  is possible with one sensor - especially if you use two colour lines -
but this is much harder. Buy the extra sensor, is my advice.

Note: I multiplexed three touch and three light sensors onto the same
inputs, using 100% light as an indication of the touch switch being pushed.
This allowed my robot to have three eyes and to feel its surroundings at the
same time.

Now, you need to consider the programming: I mostly use Robolab, which has
variables, so the robot can self-calibrate, and some Lejos too. The key
question is what programming structure are you using to detect the line?
relative sensing - e.g. 5% reduction in light, etc can be easier but is
problematic. Absolute sensing - e.g. <32 - is more controlable, but prone to
lighting changes, etc. We had a fairly controlled environment and went for
absolute, which worked very well.

Hope that helps

Mike



Message is in Reply To:
  Re: Ant pheromones
 
I have built my robot but I'm having some problems with the light sensor. Can you give me some advice on wich colors to use to avoid problems or how to put the sensor. Maybe it's just a problem of the RCX but I can't get it to see a thin line at a (...) (22 years ago, 4-Jul-02, to lugnet.robotics)

5 Messages in This Thread:


Entire Thread on One Page:
Nested:  All | Brief | Compact | Dots
Linear:  All | Brief | Compact
    

Custom Search

©2005 LUGNET. All rights reserved. - hosted by steinbruch.info GbR