Subject:
|
Ranging techniques
|
Newsgroups:
|
lugnet.robotics.handyboard
|
Date:
|
Tue, 1 Jun 1999 13:24:38 GMT
|
Original-From:
|
Bob_Kelly/First_Union@firstunion-reit.+ihatespam+com
|
Viewed:
|
1076 times
|
| |
|
|
Hello everyone...
If I may be permitted to ask your input on some of the following...
I have implemented a Polaroid sonar module and a GP2D02 IR ranging unit on
a R/C servo actuated pan/tilt mechanism. I can put in "scan_long(45.0)" or
"scan_short(45.0)" and a sensor reading 45 degrees to the right of the
centerline of the robot that supplies me with the average of 10 "hits" of
the "sonar_sample()" or "sonar_closeup()" and a GP2D02 reading at short
range.. Everything appears to work well and I am very satisfied to this
point with the progress I am making on my robot. However....
1). The Polaroid module seems to be very linear in its output, as would be
expected by the time of flight method. I am using a factor of sonar output
divided by 3636.3636 to get the range in feet. The two methods of
measurement each seem to have their own useful envelope. The
"sonar_closeup()" function is good betwee about 6" and about 2 1/2 feet.
The "sonar_sample()" routine appears to work well between 1 1/2 feet and
about 6 1/2 feet. This is fine as the ranges overlap and I can write a
routine to use the correct function to provide a good time of flight range.
My question is, if the sonar is supposed to be good out to about thirty
feet, why do I get alot of "-1" returns from the sonar function out past
about 8 feet? And why, between about 6 feet and 8 feet do the readings
become somewhat inconsistent? Overall, I find the current capabilities fine
for what I am doing and the max reasonably accurate range of 6 feet
acceptable, but hey, there's always room for improvement.
2). The GP2D02 works well too and is very useful in the short range and
provides a very accurate (read: consistent) reading between about 1 1/2
feet and 2 or 3 inches. This is excellent and fill s a gap left by the
sonar. Plus, while the software driver is active, the global value is
updated constantly. The problem is, the output of the sensor is not linear
and I want to convert the output into inches or feet so it can be used
along with the sonar output (I plan to implement some sort of comparison
between sensor readings and have what seems to be the most accurate sensor
for a given range used). I have logged the average of three readings for
each inch of range over 18 inches and have found the sensor readings have
larger changes at short ranges than for the same change in acual distance
at longer distances. I graphed the readings in excel and it is a nice,
smooth and predictable curve. I can't find my high school algebra books
anywhere but I just know there is a method of converting the non-linear
output to linear inches. (By the way, I modified the output of the sensor
by subtracting it from 238, which is what appears to be the closest range
reading obtainable before the sensor becomes unreliable - this gives me an
output that ranges from zero at close range and about 155 at long range).
Has anyone had any success at converting the output of this sensor to
inches or some other linear unit of measure? It would be very helpful if
somebody assisted my mathematically challenged self.
Thank you all in advance for your input,
Regards,
Bob Kelly.
|
|
1 Message in This Thread:
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
