Syndicate

Builder Info

Name: Atvaark

Activity

kishor: "student" 1 day ago

janefox has added a new robot: Find out the materials on the graduation dresses 1 week ago

Aakash: "I write topics on www.facebook.com/Memighty." 3 weeks ago

calinezul: "I write articles about robots on www.intorobotics.com" 3 weeks ago

Capcharm

Tags:

Category: Misc

Progress: In Progress

Time to Build: 1 year

My goal is to create a cheap and capable robotic arm (Capcharm = CAPable CHeap ARM) using as many recycled parts as possible (mostly from old scanners and printers).

Rough specs:

eventually 4 or 5 axes
extended length about 40cm
in that configuration, able to lift 750 grams
axis-mounted position feedback sensors
hybrid actuators (shoulder: DC motor; elbow: stepper; hand: servos; ...)
hand positioning precision: 5mm
control modes: coupled linear joint motion; linear Cartesian motion

So far, the arm has fully functional shoulder and elbow.

The shoulder uses a DC motor and an axis mounted magnetic encoder for position feedback. The elbow uses a stepper motor. I may be using it above specs as it gets quite hot. I need to add position feedback to the elbow in order to detect missed steps.

Both are controlled through an Arduino. I am getting reasonably good results at controlling the shoulder in position and velocity. The shoulder can follow a trapezoidal velocity profile with little deviation thanks to a PID controller. With open loop position control (generation of a trapezoidal velocity profile, then closed loop velocity control), positioning precision is about 1~2%. I can't wait to add closed loop position control (actually, double loop position+velocity control, as in real industrial robots).

As of now, elbow control is simply servoed to shoulder position. Thus, both joints are proportionally synchronized and shoulder velocity profile is scaled down if necessary, so that the elbow doesn't exceed its maximum velocity.

When I'm done with closed loop joint motion, I'll see what I can do for Cartesian motion. :)

Parts:

Unipolar stepper driver with chopping current limiter

Related Projects

Comments

Electronics has gone from

by Atvaark - 11/11/2012 - 3:02pm

Electronics has gone from breadboard to stripboard : the robot is now standalone and transportable.

New progress: Control desk

by Atvaark - 06/26/2012 - 4:24am

New progress:

Control desk with joystick, LCD, rotary encoder, LEDs (2 controlled + 1 power indicator) and 4 buttons. Connects to the robot through SPI, so the whole thing only takes one additional pin on the Arduino (SPI already used by the magnetic encoder for shoulder angle feedback). 8-bit MOSI (LCD 6 + LEDs 2), 10-bit MISO (joystick 3 + buttons 4 + encoder 2 + unused 1). Cable length: 2m (shielded, no twisted pairs). SPI clock: 4MHz (could probably be 8MHz with proper line termination; currently none). All transmission logic done using standard 74HC shift registers, gates and a tri-state buffer (6 ICs total). Generic enough to be reused on other future Arduino projects. Pictures to come.
Angle feedback on the elbow (potentiometer). Freshly mounted, not tested yet.

My MacBook is currently out

by Atvaark - 05/21/2012 - 5:01pm

My MacBook is currently out of order (stupidly burnt a component when trying to repair the AC adapter) so I'm kind of stuck, at least as far as programming is concerned. I will probably work on mechanics and sensors until I get the replacement part.

For instance, I could add an optical switch for self-calibration, as well as micro switches working as hard limits in case software control fails (already broke a gear when the shoulder went out of control).

A pot for elbow position feedback would be nice, too (even assuming the stepper doesn't miss any step, positioning is imprecise because of chain elasticity).

I just took a quick look at your project, it looks really good. I'll take a deeper look at it when I have better than a phone to browse the web. ;)

cool project! Have you made

by timlines - 05/21/2012 - 3:42pm

cool project! Have you made any more progress?

For the final object to look

by Atvaark - 05/15/2012 - 3:57pm

For the final object to look nice, I bought some mechanical extraction pipes, both rigid and flexible, that will make it look less like a toy crane and more like a sci-fi lab object. :)

Thanks for info. Looking

by wak - 04/20/2012 - 4:26am

Thanks for info. Looking forward to seeing it in action.

-Wak

Hi wak, It's nice to know

by Atvaark - 04/13/2012 - 10:14pm

Hi wak,

It's nice to know that I'm not the only one out there. :)

For the shoulder, I'm using a 10-bit (1024 steps) magnetic encoder that gives me very good accuracy and is completely independent from the direction of motion (no contact or friction). It's an Avago AEAT-6010 and it's quite cheap. I plan to use a simple potentiometer for the elbow, and since I'll be using a stepper motor, I might be able to compensate some of the inaccuracy and noise of the pot by predicting motion and just using the pot for detecting missed steps. I still don't know if I'll reach my 5mm accuracy goal for the end-effector.

Regarding shoulder transmission, I have built a gear box using only spur gears and a chain for the final stage. I will have to build a new one at some point because I made a few mistakes. Gears are too thin and tend to slip on their axis. Reduction is too strong, so motion is slow and the motor would be able to lift more at higher speed. But the weakest point is the small plastic chain that breaks too easily. I already have all the parts to make everything better (larger gears, metal chain) but I'll finish the elbow first.

Hi Atvaark, I'm working on a

by wak - 04/12/2012 - 12:54pm

Hi Atvaark,

I'm working on a similar project, only a bit larger. I'm using HiTec servos and a DC motor for shoulder :)

It may be the servos, but I'm finding the accuracy to be poor. I'm guessing the encoders give a slightly different reading depending on which direction they are turning. This means the end-effector positioning (due to accumulation of joint inaccuracies) is very poor. I imagine you will have a similar problem.

Are you building the shoulder with a worm gear? I'm guessing not. Although inefficient they can lift high loads. I have mine lifting a shoe - I'm going to have it doing the house work very soon ;)

-wak

Welcome to RobotBox!

Features of RobotBox