Add to cart is too easy - now what do I _do_ with a robotic arm?
Erlend
Posts: 612
It comes over me sometimes: I surf around sparkfun, mouser, parallax, adafruit,... you know-
This time I hit upon robotshop.com, and almost without noticing I had a large robotic arm kit in the shopping cart, and before I could be stopped (before my wife could pull the plug) I had entered all the creditcard details etc.
The box of numerous gears and cogs, nuts and boldts, servos and boards, arrived a few days ago. After a short assembly-overload-shock I found a youtube showing the relatively simple procedure of how to put the whole thing together. Basically it comprises 'shoulder', 'elbow', and 'hand' joints made from geared down servos, and tied together with aluminum tubes. It has a reach radius of 24", almost like a small human arm.
Really, I did not need this to happen. It is an ok diversion to assemble the thing, but to write code to control it? Instead I should stay focused on the project I am in the middle of, but now I see it coming; I have to learn about kinematics or whatever it is called - converting to/from radians, doing trigonomettrics, etc.
Can the forumians help me this time too? What could be a good robotics-learning experience? How should I approach the math & code part of it?
This is the arm.
Erlend
This time I hit upon robotshop.com, and almost without noticing I had a large robotic arm kit in the shopping cart, and before I could be stopped (before my wife could pull the plug) I had entered all the creditcard details etc.
The box of numerous gears and cogs, nuts and boldts, servos and boards, arrived a few days ago. After a short assembly-overload-shock I found a youtube showing the relatively simple procedure of how to put the whole thing together. Basically it comprises 'shoulder', 'elbow', and 'hand' joints made from geared down servos, and tied together with aluminum tubes. It has a reach radius of 24", almost like a small human arm.
Really, I did not need this to happen. It is an ok diversion to assemble the thing, but to write code to control it? Instead I should stay focused on the project I am in the middle of, but now I see it coming; I have to learn about kinematics or whatever it is called - converting to/from radians, doing trigonomettrics, etc.
Can the forumians help me this time too? What could be a good robotics-learning experience? How should I approach the math & code part of it?
This is the arm.
Erlend
Comments
I've always been fascinated by robotic arms but never felt strong enough in my math to try it (plus it appears to be another bottomless rabbit hole to jump into).
Maybe robotic arms are just beyond my reach! (sorry!)
I must say, that's one nice arm.
Inverse kinematics lets you calculate the angle you need to reach the desired xyz coordinates but if you have a limited number of xyz positions the arm needs to move to, you can just have these angles in a look up table. This way you just experimentally find the angles you need for the desired positions and enter these angles into your table.
The equations in post #25 of my hexapod thread would work to calculate angles from coordinates of the final joint on your arm but the position of the end effector would complicate the math (I think a lot).
I guess a lot depends on what you want the arm to do?
By the way, the code is by far the hardest part on projects like these. I was done putting together my latest hexapod in a few hours but I've been working on the code for several weeks.
Guys like Duane can code IK in their sleep. I'm not good with the math, but it's the secret to making these things really work for you. You can see in the video the arm drawing a rectangle, which at first blush should be an easy thing, right? Well, no. A couple of times it messes up, and has to rethink the path.
The solution is simple. Just dismantle the arm, put it back in it's box and post it to me. Postage across Scandinavia will not be so expensive. I can give you my address via PM if you decide to do the right thing,
The alternative is that you dedicate all your time for many years ahead to the study kinematics and control problems. During which time you will lose your job, your home, your wife and family.
I am ready and willing to save you from that terrible fate.
Oh, I do not have the heart to unload my fate onto your shoulders, you deserve better. Besides, I am already pulled into the vortex of inverse kinematics, there's nothing that can save me now.
/SERIOUS I am thinking a good start could be to have the arm controlled 'manually' by a joystick, then as the next step write code/math for sets of moves ('macros'), then <disappears into rabbit hole>
Erlend
Phew, thank you. I'm already lost in a few worm holes through the space time continuum of my own.
Manual control is a good start. After all it works very well for those guys that make holes in the roads with huge diggers deftly controlled by a few little levers.
Programmed sequences is an obvious next step.
I'm not so attracted to the inverse kinematics thing. How about something a bit more organic? Have the arm follow the motions of your own arm, tracked by a couple of wearable IMU's? How about putting some sensors on it and have it "feel" around for the ground, stuff to pick up and so on.
Can you get it to balance a stick up on end, the inverted pendulum problem?
I like the cyborg'ish idea of having the robotic arm follow my own arm, but if I am to use IMU's they would give out XYZ, so that means IK math is required. To avoid IK my arm will have to be fitted with sensor that gives out the angle of each joint. Shaft encoders, maybe?
I love the idea of building a (tacky) robo-arm which feels its way around. Shouldn't bee too hard.
(starting to slide down the rabbit hole)
Erlend