srf04/ sonar detection problems
Archiver
Posts: 46,084
Hi all: I also have a robot with both srf04 detection and the Gp2d02 Ir sensor.
I have reached a unsolvable stumbling block with both sensor's for object
detection.
1) the Gp series will randomly miss detection depending on Reflectivity of
the object. This applys to simple fixed range IR detectors as well.
2) The srf04 sonar will miss detection depending on angle of approach to a
wall, or object.
I proved this tonight, and found this to be a reproducable error.
I have a hole sort routine in my code. If I have a detection ahead, my robot
stops, rotates the sensor head thru 180 degrees, find's the largest hole and
turns that direction.
If the robot comes into a white wall at approximately 45 degrees *with the
right side of robot nearest wall), it will not see the wall at all. My IR
detector stopped the robot fortuantly. The robot then ran thru the hole sort
routine using its sonar srf04.
The srf04 would NOT detect the wall in front of and to the right of the
robot, even though it was almost touching the wall. Pulling my hair out, I
used my simple srfo4 range test/lcd display program. I proved that at
exactly 45 degrees, the sonar will hit the wall, bounce off, travel down the
wall, and then get sent back to the sensor by whatever is at the end of the
wall. I rotated the sensor head thru a range of degrees, and found that
anywheres from 30 degrees on either side of square/center to the wall, the
sonar may bounce off and travel along the wall, giving a false indication of
long range. At exactly center, I get a good/valid indication.
In addition when the robot sits very near the wall, and rotates the sensor
head, it may or may not get a bad reflection depending on angles inclement
to the wall.
How the heck would anyone prevent a problem such as this?
The wall is a repeatable/provable failure, Sonar is going to do the same
thing depending on angle to box's chairs etc as well.
I could buy more srf04 sensors, but how to place them on the robot to
prevent such a situation? Also, how many does a person need to buy? 50 of
the things to reach semi reliabilty?
We have talked about the polaroid 6500 series, but once again it is sonar,
and I expect it will have the same problem as the srf04. Where does this
leave a person for robotics? No sensor out there is even 50% reliable under
real world conditions. I do not want to shell out the money for a polaroid
sensor, only to find it do the exact same thing.
Is there ANY sensor that is semi reliable? For any kind of money? Other than
a bumper switch?
Thank you to everyone.
-Yours with much regard
-Kerry
I have reached a unsolvable stumbling block with both sensor's for object
detection.
1) the Gp series will randomly miss detection depending on Reflectivity of
the object. This applys to simple fixed range IR detectors as well.
2) The srf04 sonar will miss detection depending on angle of approach to a
wall, or object.
I proved this tonight, and found this to be a reproducable error.
I have a hole sort routine in my code. If I have a detection ahead, my robot
stops, rotates the sensor head thru 180 degrees, find's the largest hole and
turns that direction.
If the robot comes into a white wall at approximately 45 degrees *with the
right side of robot nearest wall), it will not see the wall at all. My IR
detector stopped the robot fortuantly. The robot then ran thru the hole sort
routine using its sonar srf04.
The srf04 would NOT detect the wall in front of and to the right of the
robot, even though it was almost touching the wall. Pulling my hair out, I
used my simple srfo4 range test/lcd display program. I proved that at
exactly 45 degrees, the sonar will hit the wall, bounce off, travel down the
wall, and then get sent back to the sensor by whatever is at the end of the
wall. I rotated the sensor head thru a range of degrees, and found that
anywheres from 30 degrees on either side of square/center to the wall, the
sonar may bounce off and travel along the wall, giving a false indication of
long range. At exactly center, I get a good/valid indication.
In addition when the robot sits very near the wall, and rotates the sensor
head, it may or may not get a bad reflection depending on angles inclement
to the wall.
How the heck would anyone prevent a problem such as this?
The wall is a repeatable/provable failure, Sonar is going to do the same
thing depending on angle to box's chairs etc as well.
I could buy more srf04 sensors, but how to place them on the robot to
prevent such a situation? Also, how many does a person need to buy? 50 of
the things to reach semi reliabilty?
We have talked about the polaroid 6500 series, but once again it is sonar,
and I expect it will have the same problem as the srf04. Where does this
leave a person for robotics? No sensor out there is even 50% reliable under
real world conditions. I do not want to shell out the money for a polaroid
sensor, only to find it do the exact same thing.
Is there ANY sensor that is semi reliable? For any kind of money? Other than
a bumper switch?
Thank you to everyone.
-Yours with much regard
-Kerry