Society of Robots - Robot Forum

Software => Software => Topic started by: visshukl on September 15, 2010, 10:13:29 PM

Title: Getting started with Computer vision
Post by: visshukl on September 15, 2010, 10:13:29 PM
Hi All,

First of all I am glad to find this forum.. and looking at all the posts, I am sure it's a great place to get information.

Now this is regarding getting my robo's computer vision. I have looked into different options and I feel that roborealm should do the job. The current robo I have is a Sonar based navigation running on Basic stamp-2 (coded in P-basic). I know there will be lots of work to change it based on coputer vision, but I want to start rather just thinking about it.

I want members help here to get me started, some basic document which I should start reading to get a clear understanding on what all I will need for following:

1) Parse the image and get it to roborealm -- webcam?
2) How I can process that image for navigation -- once the image is in roborealm, how I can process it using code?
3) Based on the image properties , how I can generat signals for Servo motors -- I mean, if roborealm runs on a laptop, how the signals will reach to servos -- ??
4) Some Axon type of board would be needed.. right?

Thanks a lot in advance!

Title: Re: Getting started with Computer vision
Post by: SeagullOne on September 16, 2010, 01:48:58 AM
Hi there, welcome to Society of Robots! :D

I think I should be able to answer some of your questions.

I've experimented a bit with Roborealm and OpenCV for robot computer vision. My robot has the laptop mounted on it, so the laptop basically acts like one big controller. Here's what you could do though: use telemetry to control the robot from your laptop--which runs roborealm. Your robot will have the camera or webcam mounted on it and will send video feed to roborealm via telemetry. The laptop will process the data and then relay a course of action to the robot's microcontroller via telemetry feed.

Forgive me if I shouted a bunch of ignorance there. I've read about telemetry in robotics, but don't have any hand's on experience. ;D

Processing vision in code is easy in roborealm. You can use Visual Basic or the Python API, or just about any other kind of API roborealm supports--I used the python API on roborealm before switching to OpenCV.

To communicate with servos and motors, roborealm can do that extremely easily as well, if plugged into a Lynxmotion SSC-32 (for servos) or a Sabertooth motor controller for example. I'm not sure if roborealm can plug into the Axon...I think it can, actually.

You could use the Axon board for your robot's microcontroller. I've heard great things about it, but never used it myself. I've just always used a bunch different boards for different things on my robot all hooked up to my laptop via USB cables (which my application just requires).

Have you built your robot yet, or it is it still in the planning stages? Because if you're laptop isn't too heavy, you could just mount it on your robot...but of course, its your project. Use your imagination and ingenuity as you will. Who knows what you can come with?  ;)

Hope I could help at least a little.
Anyone else?