I am interested in building a robot and was looking for some direction on what i need to learn and buy specifically for this project. I am not really looking to make a hobby out of this but rather looking to make a little friend to religate a very specific task to.
The basic idea is that i want a robot that is going to do some menial labor in a game for me. He would need to be able to see a picture and press a button based on which specific picture it is.
I have something like this in mind where a webcam would be looking at an iphone and based on what is displayed work one of 8 servos.
I have taken a single programming class in c++ several years ago, so I do not really remeber any of it, however I am familier with the basics and could probably write some very elementary stuff. However I was hoping there might be some already made control software that might be already suited to this task. Any help or direction would be appriciated:)
"Seeing" a picture will probably be the difficult part. Depending on the game, you could place color sensors in strategic locations to determine when certain game pieces/sprites are at that location, and use and Arduino to trigger the servo motion. (Keeping in mind some touch screens will need a clever workaround to get inanimate objects to control them.) Having an actual camera vision system will probably be best done via your PC through an Arduino or similar. You could look into the OpenCV library which is a fairly powerful computer vision library for C/C++. I’m sure there will be others along, with more than clever solutions. Good Luck.
I do see your point as I am basically asking to do exactly what you warn about in the video. However I am very persistant and really don’t have anything better to do.
I guess i should be more specific in my questions.
There are a whole bunch of aurdino boards and what I am hoping not to do is experiment. I would ask what board would be best for this particular application?
The depiction of the fingers in the picture is just something I threw together quickly. The final product will have a system that pulls them completely out of the way of the optical system when in an inactive state. The fingers will have a conductive material attached to them. I am not sure if it needs to have some kind of current applied, however having looked into that there are other examples of people using robots to play games like rock band, so I know it is feasible.
The pictures will be stagnate, no moving parts or anything of the like, just a picture that needs to be recognized and trigger the proper output.
I looked into opencv a bit and don’t fully understand at the moment how I would integrate that into an interface that runs a robot. However if that is the only software capable of recognizing a picture as an input I will make a stab at learning it. I was hoping to get some information on other programs I could use for processing an image from a camera and turning that into the commands. Hopefully something that doesn’t require a years worth of work just to get started. OpenCV seems like the kinda stuff you need to make quadracopters make complex optical calculations, I was hoping for something a tad more simple.
Yes, sounds kinda ridiculous I know, but when you spend 30 hours over a week and lose at a game, the next logical step is to learn how to build a robot:)
a little experiment to trick the iphone screen, attached a peice of fritos bag to about 200 ft of #6 ground wire. The wooden dowel is so that I could manupulate it while being sure I was not the capacitive element. The bag was necissary to make the proper shape as the phone will ignore anything not resembling a finger.
I understand the general negitve sentiment as I am sure I am not the first to come here with a grand idea and without the means to accomplish it. I am definitly trying to be upfront about my inadequecies, however I am certainly not looking to join the airforce.
What I want to get input on an avoid by speaking with the experts is buying a 50$ program like the one I was speaking about in one of my previous posts only to have it take 30-40 seconds just to process an image.
I already played around with the ez-robot software and it seemed like a dead end. It just does not have the cabability to recognize a picture. The only images it can process are a face or a predefinied glyph. I need to basically be able to define my own “glyphs”.
Was hoping mostly someone had come across a simular problem before, however if thats just not possible then project scrapped and I will move onto something else.
I guess I should also mention that I have taught myself excel programming and coding for the zmud gaming interface, so I am not a stranger to the basics of coding. I just really don’t want to spend 6 months building only to find out that I bought the wrong board, wrong software and wrong everything else.
The french fry actually just conducts your body to the screen. I used 200ft of ground wire as it needs to be something large enough to register. This thing doesnt actually need to go anywhere so I will just have a coil their with it and use conductive rubber to touch the screen with.
I had thought about using color sensors, but in many of the cases the sensor would have to be directly over the area where the servo would need to touch. I guess if it came down to it I could utilize a system that moved the sensors out of the way, brought the conductive material down, and then replaced the sensor in the appropriate place. However I was hoping for the much more elegant solution of image processing since there is only a finite amount of images that need to be recognized.
How well do the color sensors work? Can the distinguish between a wide range of colors?
hehe, well there is slightly more to the game, however a lot of it is extremely time consuming. However mostly its about the project itself. I like to do things I have not done before for the callenge. Currently I am building a spear… Once that is done I am going to need something else to do. I am hoping to have that done before the end of January. After I am done with this project I will think of something else to do.
I have been playing with my I have been playing with my rpi today trying out the python bindings for opencv. You could probably use something like that to do what you want. I will post some pics in a few mins.
