Whahooooooowwwwww!!!!! I am sooooo happy! In the wee hours of the morning today, I wrapped up the Calculon’s (that’s the name of our robot) vision system for UCF Robotics! And on top of that, I got a little video that shows “what” Calculon sees (3.2 MB)! We have some videos we recorded of Calculon going around on a little obstacle course (remote control). I ran that through the vision system and saved each frame to a BMP. The vision system averaged about 13 FPS on 360×240 resolution (half NTSC), which is really good. So about a thousand images were generated and then I ran them through a BMP -> AVI converter. I also ran it through Windows Media Encoder to compress it. But man, after 8 months of solid development, it is finally coming together! A good thing too, since competition is about 3 weeks away 😉 Still, I was soooo completely stoked!
Just for those who might not know, Calculon is a wheelchair based autonomous robot. That means that while we can use a wireless joystick to drive it, the goal is for it to be able to drive itself and navigate through an outdoors obstacle course. The obstacle course contains stuff like construction cones, 5 gallon white buckets, spray painted lines, and other stuff. So a vision system is crucial if Calculon is to see buckets and lines and then avoid them (like an intelligent robot). We treat the video as a bunch of frames coming in and process each one individually. Without going into too much detail (I’ll leave that for later), the vision system looks for buckets and orange cones. If it sees one, it colors it yellow so you know what it is looking at. A blue box is also drawn around it so it doesn’t interfere with the line finding. The line finding looks for lines in the rest of the image. If it finds them, it draws over them with red. Super cool, huh?