About my job, I am given a robot and what I need to do is simple. Program it.
This small robot here is fabulous. It's equiped with a built in linux based operating system, a flash card slot, its multidirectional, it carries 9 distance sensors, a webcam, a bumper strip sensor, 3 encoders, 3 actuators, a gripper, 2 color sensors, a metallic strip sensor. Oh, not forgetting its built in LCD screen and mini keypads!!
But yet, the final aim of my internship, is to program the robot so it could navigate itself using localisation systems(which i need to come up with), as well as mapping the environment(which means, taking pictures of its surroundings with its webcam, and then map out its surrounding in the robot's memory)!
So to achieve that, 1st you need to know the robot, its sensors and actuators and so on. Then you need to work with the webcam so it could take pictures. Then, you do image processing, to take out all extra information and extract the information you want from the image. Lastly, you store that information in its memory and construct its environment using that information.
Well, its just a few lines of words to describe my work. But after one month here, I only started on image processing. And the best part is, I need to complete this in half a year's time!!
Today, I managed to classify the objects on a table and extract them out from the image. Since I never do image processing before, it took me 3 days just to construct an algorithm for this.
Here's a picture of the original image captured from the robot's eye(webcam).
Next is a picture of the processed image with all the details cut out, remaining the objects edges.
Then, using this image, I need to count how many items there are on the table. To do that I have to classify each connected component as an object.
Here's the extracted picture of each object.
The pen!
Mr.Eraser 1.
And Mr.Eraser 2. In step image, which means the pictures consist of two levels only, either black or white.
The result.
I can now count the number of objects found, and from there manipulate the object data and so on and so forth.By the way, I designed the interface. That button allows control over the robot's movements.
On a side note, this is the motion detector that I did for testing last few weeks.And that's my job generally. Also, lately, more and more of my colleagues are getting involve in this project(reads helping me).
I guess it's a good sign!


1 comments:
good thing !!
I also still do program my robotino with C++, but it's not finish yet.
it's nice, so you program it with C++??
what application you used to program it?? is it visual studio??
you use GUI from MFC or from the Robotino example program??
I have some problem for image processing?? can you help me??
thanks
Post a Comment