The duce stretch
The arm and claw look good. Looking forward to seeing it work. I would think that either the Jetson or the Pi 4 would be more than ample computing power to control the arm and claw. Let us know how you decide to control it. And maybe even share your code?
Thanks ! I'm going to use the pi 4 I was thinking its going to need a camera mounted on it if its going to work with machine learning so I got a raspberry pi camera for it but would it need something like the pixy cam for it ? Something that it could recognise objects with? Then use that to grab something I'm also adding one more servo at the wrist today I will update the photo .and yes I will certainly share the code now that I have all of the computers on the robot on the forum and can actually type it somewhat clearly so it can look right and I added a Samsung a20 today so I can take better photos of how the arm was constructed which I will also gladly share its real easy and cheap and strong .👍😀
The pixy cam doesn't actually recognize objects as objects it locates areas of color. So if you have a red ball or a blue block or a green toy dragon it can associate their colors with an object name or track the color. Heavy duty analysis of a visual input requires a lot more computing power. The deep neural network programs that apparently can run on the Jetson Nano are closer to what you would call object recognition although I am not sure how much information is returned to allow a robotic arm to manipulate them.
Many years ago I placed a b/w webcam on the first (and superior) toy owi arm just after the elbow joint. The grippers on that model were powerful enough, and their claw large enough, to grasp and lift small wooden blocks.
My idea was to write a block stacking algorithm.
In practice I found that like the human hand it really needed a sense of touch. You can touch just part of an object with just parts of your hand and immediately recognize the object. Something to think about. Object recognition by touch alone.
thanks casey! it seems like this is a difficult task but i think you are right about putting the camera just after the elbow i arrived at about the same point i was trying to get it as close to the claw as possible .i was checking out the links you provided they will be helpful .people just dont realize how hard it is to get something to work on its own .....but now i do lol.