Society of Robots - Robot Forum
Electronics => Electronics => Topic started by: amadcow on October 08, 2008, 06:17:41 PM
-
Using one of these sensors:
http://www.acroname.com/robotics/info/articles/sharp/sharp.html
Is it possible to detect a round surface (or even a slanted surface) with a single Sharp IR sensor? By rotating the sensor back and forth, would it be possible for a robot to detect the shape of an object?
I'm uncertain about this because according to the documentation, the Sharp IR sensors use triangulation to detect distance from an object. So would the infrared beam always be required to land perpendicularly to a surface to give an accurate distance measurement?
-
Also if it isn't possible, with what sensor would it be possible to do this?
-
I don't think it is possible.
-
If you work with the code and make the robot go around the object you may be able to make it find the shape of an object but it seems hard and might not even be possible.
-
Well I'm thinking very simple shapes, like the incline in a wall or perhaps an open door.
-
It probably depend on the surface. For example, it probably won't work with a mirror finished metal plate, but a wooden board, I think for a reasonably small angle it should work.
Laser displacement sensors are like sharp sensors but with laser beams, with has longer range and higher accuracy, though they are very expensive... You might be able to make a webcam laser range finder, which would probably work for the application you need.
-
Its entirely possible and I show how to do it here:
http://www.societyofrobots.com/sensors_sharpirrange.shtml
Save your data into an array, then do some math on it to to determine which shape it is. Basic shapes are incredibly easy!
-
Thanks for that link. My main goal is to enable mapping without having the robot actually move around and trace objects. So if I had a stationary servo with a Sharp IR sensor on top being rotated back and forth, could I map something like a triangle or a ball placed in front of it? Of course it wouldn't be able to map parts of the object light couldn't touch (like the back of the object), but could it still get the general shape of the front of the object?
Which is why this GIF interests me a great deal:
http://www.societyofrobots.com/images/sensors_IRSLAM.gif
As you can see, the robot's path doesn't need to trace the shape of each room in order to determine its shape. If you look closely, you can see it even mapping the open doors to some of the rooms. I was curious of where that image came from. Was it a concept drawing or was it an actual demonstration of a SLAM robot? And if so, was it done with the Sharp sensors?
-
Of course it wouldn't be able to map parts of the object light couldn't touch (like the back of the object), but could it still get the general shape of the front of the object?
yeap
As for SLAM, its not something easy to do. You'll have to spend a lot of money on sensors, and put a laptop on your robot to have enough computationally capacity, too. Search the forum/google to learn more, but stick with the basics first.
-
Thanks, appreciate the answers everyone.
Yeah, the project I'm working on isn't exactly a basic robot.
I've built a few robots before (and some with these sensors), but they were used for simple things like obstacle avoidance. For this new project I'm going to rely heavily on their precision which is why I'm so concerned about their capabilities.
As far as computational capacity goes, I think I'm going to be using a ZigBee device to connect the robot to a host computer.
-
No it is not possible using a sharp IR sensor,
Instead you could use an ULTRASONIC SENSOR