Society of Robots - Robot Forum

Electronics => Electronics => Topic started by: SomeSaba on October 22, 2007, 06:00:17 PM

Title: best way to detect a ball?
Post by: SomeSaba on October 22, 2007, 06:00:17 PM
Hello

Ok, so theres a capret and a tennis ball on the carpet.

If i aim an IR sensor straight down and take a sort of "threshold" value of the distance to the ground, and use that to detect the tennis ball would it be better than an sonar sensor due to the curvature of the ball? Im just worried about the tennis ball is not reflective enough

thanks <3!
Title: Re: best way to detect a ball?
Post by: Steve Joblin on October 22, 2007, 06:21:19 PM
I would think that sonar would be a bit more reliable, but difficult to say... have you considered trying to detect the tennis ball by its different color than the background?  I wonder if a vision or color sensor wouldn't be easier and more reliable?
Title: Re: best way to detect a ball?
Post by: Admin on October 22, 2007, 07:54:46 PM
sharp IR would work . . . but as steve hinted at, Id use a colored ball with the CMUcam. This would be very reliable and not very difficult.

For example:
[youtube=425,350]rE6SJZC9eeo[/youtube]
Title: Re: best way to detect a ball?
Post by: SomeSaba on October 22, 2007, 10:08:31 PM
thanks for the replies!

At previous competitions the carpets have been anywhere from gray to green, if its green im afraid the tennis ball might blend in  :P
Title: Re: best way to detect a ball?
Post by: bulkhead on October 22, 2007, 10:43:10 PM
Just wondering, are you trying to build a completely autonomous robot to pick up the objects?  Because I think it would be much easier to just have someone drive it...most likely quicker too.
Title: Re: best way to detect a ball?
Post by: SomeSaba on October 22, 2007, 11:04:53 PM
i totally agree... but i want serverly challange myself and make a completely atuonomous robot designed for a competition where the robot has always been driven by the competitor.
Its my senior year :) i g2g go with a bang!

Imagine the event supervisors/audience watching all the robots, and then there's the sudden awe when they realize im not controlling it!

of course there's a very high chance for error... but i hope to work it out all the little details to make it perfect, and of course it wont a big enough bang if i dont get first place  :D

I'm very excited about this project <3

(BTW: are u a competitor too? which school are you from?)
Title: Re: best way to detect a ball?
Post by: bulkhead on October 23, 2007, 12:25:07 AM
That's cool.  I am a former competitor...mine was almost autonomous one of the years I did it, but that was before they moved the box.  I didn't get to run mine autonomously or even very quickly for that matter, but my team won so it was completely worth it. 

Making it autonomous this year is real challenge...have fun with that!
Title: Re: best way to detect a ball?
Post by: Steve Joblin on October 23, 2007, 08:11:00 AM
If the carpet is grey or green, it shouldn't matter as most tennis balls are fluorecent yellow, which is quite easy for a color webcam to discern.
Title: Re: best way to detect a ball?
Post by: MaztorShake on May 26, 2011, 09:15:59 AM
I know I'm new to anything robotics, but how exactly is the gameboy being used? Is there some link someone has bookmarked that I can read?