Friday Night Robotics

Posted by Erin, the RobotGrrl on Friday, December 26th, 2008

Merry Christmas and Happy Holidays to everyone!
Best of all… it’s ROCK EM SOCK EM ROBOTS DAY! (Boxing Day)

Just as a fair warning, if this entry has more bizarre grammar mess-ups than usual, it’s because I ate too many jelly beans, and all I can think of is jelly beans! Robot jelly beans, jelly bean slushie, a meadow of jelly beans, jelly bean snow… *goes on and on* I think in pictures… and all I can think of right now is a DANCING JELLY BEAN ROBOT! Hahaha, joking. (Or am I?) ^_^

I hardly realized it was Friday today, so it doesn’t really feel like a FNR.

Last Friday, although I didn’t blog it, I was working on the LEDs that you’ll see on the stick, and the program! This Friday I tested and twirked the program so I can blog it.

Here is MANOI!

Do you notice something different about the stick?

The stick now has two perf boards on it! These perf boards have three yellow LEDs surrounding a LDR which has black construction paper around it.

The reason why I chose these yellow LEDs was nothing scientific. These were the only ones I had where I could have some consistency across the two boards. The LEDs on the left are inside of a yellow casing. The ones on the right are in a clear casing. There is a little bit of a difference in the readings of the LDRs, but they both work the same.

I would have really liked to do a different version of this but with an IR sensor and a FSR, but, as I mentioned above, these were the only things I had, and they will for sure get the job done!

What happens is the LDRs “spit out” a reading of the amount of light around it. I use the cardboard to ensure it is directed at what we want to be observing. When there is no object between the two LDRs, the reading is very high. Yet, when there is an object between the two, the reading drops a fair amount. This is because the LDR cannot sense the light from the opposite LEDs. I also tested this with a fairly white object (my DAD’s lead solder spool (eeew lead)) and it does work, meaning the reflected light from the LEDs back into the LDR (on the same board) does not obscure it. You can see for yourself in the video below.

I tinkered with the LDRs’ amount that it to create a threshold that will trigger MANOI’s shot. From what I learned at Stanford, this is simply called a neuron. It has input values, and if they meet a certain threshold,it will do something. This is a very primitive form of AI, but big things come in simple steps. =)

Here is a video of MANOI action! Pretend that the roll of tape is a hockey puck… it almost looks the same!

MANOI Hockey Robot AI from RobotGrrl on Vimeo.

It is also on youtube, here.

At the beginning, you see the LEDs flash. This means that the Arduino is soon going to evaluate the LDR levels. It takes five samples from each LDR, one every 500 ms. It then makes an average, which is the baseline used for comparing the LDR value against the threshold.

Yes, I agree, that this can completely become messed up. I will probably implement a markov approach to create (and update) the baseline. However, I’ll probably only do that after I make a modified Bayes algorithm for the AI.

You might have noticed something different about MANOI’s other hand. I’m giving it a gripping claw so that it can grip stuff. It won’t be meant to hold on to anything precious, but it could emphasize an effect here or there.

While I was adding on the boards, there was just not enough room for all of the stuff that I needed… like +5V and -Gnd. I made a thing with headers and inverted headers so I can easily plug it in!

It saves much needed space on the wave shield:

Just to add, today I was using an Arduino with an ATmega328.

There are more pictures that you can look at in this photoset on Flickr!

This project is almost finished! I just will add in an algorithm improving the AI, make more videos… and that’s it! I will continue the project later on, though with a few more things. I’ll add servos that can rotate the leg so I can make the robot actually skate, and many more things that I am still thinking of.

Can you believe that the Trossen Robotics robot contest deadline is really soon? Yikes!!
I just have to take a few more videos and work out what I will say in my post… what do you think the judges will be looking for? Hopefully it won’t end up like BubbleBoy in crabfu’s competition… x_x

Though, when crabfu was on Daily Planet (I actually saw it on TV (like two months before all of the robot websites started to go crazy over it), not on youtube) he described his robots as having character because you can interact with them through a control pad.

I think that is completely opposite, because then it is just the human expressing their character through a machine. When you don’t have to use a remote, yet the robot still interacts with its environment, that’s when a robot truly has character. :d

Now we know why BubbleBoy obviously lost… He can have his definition, and I’ll have mine. His works are very interesting though! I never knew you had to start a fire to make a steam thingy move. I figured you would just boil water or something… hahahaha (I never thought as far as you would need fire to make water boil, though… EPIC NOT-WIN!) XD

I hope you really enjoyed this blog post. It makes a lot more sense now that one can see what the finished robot will look like!

Posted in: MANOI, Projects, Robot.
Tags: AI, Arduino, artificial intelligence, ATmega328, beans, boxing day, dancing jelly bean, FNR, friday night robotics, holiday, jelly, jelly beans, LDRs, LEDs, light, MANOI, MANOI AT01, neuron, perf board, pictures, reflection, Robot, robotics, SSC-32, steam thingy, video, waves

Leave a Reply

Name (required)

Mail (will not be published) (required)

Website