Mini Delta Robot with background noise to verify actual speed

[douro20]

It may be easy to build a delta robot; the hard part is to get one working.

[BJBlaskowits]

Show off

[jaredknightcom]

Very cool :)



[edotekk]

Which hardware is used (Controller, Motors)?

[MimixMotion]

@edotekk This one uses Panasonic motors and amps with Delta Tau controller. It would have been preferable to use Delta Tau amps, but Panasonic motors use a proprietary feedback, one that is best used with Panasonic amps... which requires Panasonic tuning. Due to the inertial mismatch, the delta would have performed even better if could have been tuned by the Delta Tau controller, but this was good enough.

[JamisonBruch]

@edotekk Delta Tau controllers now can read Panasonic feedback... this will get better/.

[simon3314]

@edotekk This stuffs uses  Java. Is Java the software?

[cliprepository]

This is fantastic. It would be great if you could publish plans so makers around the world could get in on the fun!

[MimixMotion]

@cliprepository The mechanics are so basic that plans aren't needed. Take some 100watt motors, put encoders and brakes on them, plug the encoders and motors into a Delta Tau Brick (controller and amps) and you are off and running. Mechanical challenges are in joints and link lengths. the geometry. This is something I've spent a lot of time optimizing... it isn't required for a home built. The impressive motion routine is courtesy of a Delta Tau guru Sina. Few could replicate that.

[newusername01]

I who be interested to see how fast this thing can type on a keyboard

[technoshamanarchist]

Why didn't you just put a clock there :))

[MimixMotion]

@technoshamanarchist The original version had the sound removed and people were commenting how it was obviously sped up... So I just reposted it with the sound.

[technoshamanarchist]

@MimixMotion I know, I saw the other videos, too. I meant, instead of talking, you could just put a wall clock there for reference :)) Good job :D

[blacklotus808]

Hudson: "Hey, Bishop. Do the thing with the knife."

[MimixMotion]

@blacklotus808

yup... you got it.

[MimixMotion]

@blacklotus808

Looks like AEGMexico liked the routine as well... imitation is the greatest form of flattery...... check it out. lol

[johnnyk427]

@blacklotus808 I was just about to say that.

[blacklotus808]

@johnnyk427

Great minds think alike.

[jimday666]

whoooo, pneumatic parallel robots rule.

[MimixMotion]

@jimday666

Not pneumatically actuated, the pneumatic noise is the vacuum gripper. Actuated with BLDC motors.

[jimday666]

@MimixMotion oh cool.

[Adrianrff]

It still has 3 translational D.O.F. though right?.

Wouldn't those joints make the robot redundant?

Why did you choose ball and socket, a revolute joint on a cylinder joint would have the same efffect right?

Im sorry for the questionnaire but i'm doing my thesis on parallel robots and I didnt really know much about them.

Thank You for replying

[MimixMotion]

Consider these concepts, and investigate where they contribute to PKM architecture. 1-Tolerances in manufacturing. 2-Over/under constrained motion 3- resonant frequencies when material/mechanisms are subjected to particular loading scenarios... then from a practical viewpoint, consider that the god-fathers of PKM's use ball socket in deltas... why?

[Adrianrff]

Ok. Another question arose:

Is it underconstrained?

I mean, each parallelogram bar is linked to 2 B&S joints at each end; what prevents them from rotating about the axis that passes through the centers of the B&S joints (which is parallel to its longitudinal axis) This rotation wouldn't affect the motion of the end effector, I think.

[Adrianrff]

I understand that there can be avdantages in some redundant PKM designs (Enlarging workspaces, solving FK problems etc.). But i cant see clearly if this mechanism is kinematicly redundant due to the S joints.

I dont intend to criticize your work, quite opposite.

Im just trying to clear up some trivial doubts

[JamisonBruch]

To clear up my responses, it appears I've mixed up my personal and Mimix Motion accounts. I am both JamisonBruch and MimixMotion.

[JensOverby]

There is one good reason for choosing balls and sockets. A "revolute joint on a cylinder joint" has the same effect, but at the same time you introduce a lot of backlash in the structure. A ball which is tightened to a socket by a spring has minimum backlash. Backlash is indeed something you need to consider.

[JamisonBruch]

Hi Jens! How are things :) Thanks for picking up my slack in replies. Adrianrff, Jens is a great resource for delta robot analysis. Please check his models out.

The delta is not over constrained, while it is closed chain, the kinematics are not coupled, the geometry is clean, the math is as clean as you can get in PKM's. Any combination of joint angles results in only 2 unique answers, similar to left hand/right hand. Geometry is key... as Jens' work solves/demonstrates beautifully.

[Adrianrff]

Thank you for taking the time in answering my questions.

[JensOverby]

@JamisonBruch Doing allright thanks. However, everyday I'm getting more and more impressed about your mini-delta, because building it must have occupied you for "seasons" - if not years. I realize that now. My own (not so mechanically ambisious) prototype is doing fine - though not on schedule. Just got all servos running coordinated at full speed for a test task. Then I added a bug to the program, and I broke one of the gear sets, so I'm waiting for a new one now. I'll be there soon..

[MimixMotion]

@JensOverby :) Kindred spirits. :) I have 3 of the 2nd generation mini delta in different stages of build, with one running preliminary motion profiles. It uses much cheaper motors/encoders/controls and promises to be much faster.

[Adrianrff]

Thank you for your answers!

[Adrianrff]

Is the Parallelogram linked to the end-effector and to the legs with ball and socket?

[boygenious92]

hello ive got to say that was phenomenal im currently trying to learn programing with pic micro-controllers can you please tell me what hardware/software your using to control this :O ???

[JamisonBruch]

There are two directions one can take if interested in making one of these... high-end hobby hardware and controllers, or low end industrial controls and hardware. I made these with low end hardware and controls... well, the controller is inexpensive but top of the line features. The PMAC from Delta Tau.

[lordzeppo]

I'd like to get that software,any way?

[MimixMotion]

Delta Robot development is still quite new... at least relative to other robot geometries (scara, serial, cartesian etc), so there is high value in mature and tested software. You can pay for it, or research and develop your own. I've spent several years developing an entire automated design system, with geometry, dynamics, kinematics and trajectory generation optimization.

[lordzeppo]

I don't know for you but for me,a robot design pattented in 1987 or from the initial idea of Willard L. Pollard in 1942,this all don't seem still quite new. Secondly,my college is developping those robots for business application.And finally,the only reason why I asked for the code is that I got some spare time and I love to understand and reproduce stuff using some microcontrollers(check my videos). And also, reverse engineer the delta robot wouldn't be too hard.

[MimixMotion]

You are right... simple stuff... all you need is trig and to reverse engineer a simple mechanism.

[lordzeppo]

still,simplicity is sometime way too wonderfull ;p

[MimixMotion]

Quebec? You have some the best researchers of this type of mechanisms... PKM's, in your province. Ilian Bonev, Clemente Gosselin and others literally write the texts for this type of mechanism.

[lordzeppo]

yes we do ;p

just a question on the fly, I saw that the platform is staying parrallel to the base,wouldn't it be feasable to implement an angle to the platform?

[MimixMotion]

Motion has some fundamental laws regarding degrees of freedom, constraints, # of actuators etc. Your questions imply a good aptitude for this type of mechanism, one that is best fostered by the details available in texts. The bible of PKM's will be available shortly, "Fundamentals of Parallel Mechanisms" by Ilian Bonev Phd. Until then, parallemic and published papers are good sources.

[gmark3000]

Actually, I think you guys are missing the big deal. Robot geometries and programming have been common for awhile. The most impressive thing to me is how well this thing is constructed and it's sheer mechanical performance. It's just a damn fine piece of work.

[JamisonBruch]

Thanks!  This one was the fastest of the group. (I've made several different ones for Delta Tau that can be seen on youtube and at conventions.)

New ones coming that are even more optimized for performance.

[plaasjaapie]

I'd love to do one of these and I've run across some pretty inexpensive high resolution controllers by phidget. I haven't been able to find any affordable servos, however, with anything like the ppr ratings that you're using. Any suggestions? :-)

[MimixMotion]

You are coming at this project from a different direction than I did. I try to use the lowest cost industrial motion control, something that still ends up costing way more than a hobbiest can typically afford. The 131kppr is a typical serial encoder resolution, something that is a bit pricey. A good place to start would be Ebay dc gearmotors with encoders (pittman or something).

[plaasjaapie]

I assume you were using some sort of servo. What kind?

[MimixMotion]

I've made several of these, maybe 7 or so, you will see them at conventions, expo's, and around YouTube. Each uses a different motor/feedback/PMAC control system. This particular robot uses a high performance servo @1Nm peak, 24V brake, feedback @131k ppr, Delta Tau Clipper controller.

[MimixMotion]

I see you like steppers. I've designed a couple stepper actuated versions but found a simple servo motor is hard to beat in performance, cost, and risk.

[plaasjaapie]

Brilliant work and great video! I'd like to build one a lot. Few questions...

1) what sort of positioning error is your system getting?

2) what kind of controller are you using?

3) are the maths and/or code that drives the system published anywhere?

Thanks!

[MimixMotion]

I'm glad you like!

1) No time was put into testing the error in these prototypes. I wasn't worried, and it wasn't a problem. It was designed with the proper motor and high resolution feedback, and is controlled by the highest end controller, so error can be calibrated, tuned and compensated out of the equation. No gearing, no backlash or significant friction makes things very easy.

[MimixMotion]

2) Delta Tau PMAC controllers are perfect for atypical mechanism control.

3) Delta Robot kinematics (inverse) might still be difficult to find online for free... but they are easy enough to calculate with simple trig... triangles have been around a long time :)

[onlylunch2]

Wicked. I want one. For what? I don't know. I just want one.

[bigtank2185]

Its awsome, for high speed placing of an object from one spot to onother... j.k.

This is a cool piece of engineering.

[MimixMotion]

"Pick and Place" is a large portion of the automation robotics industry.

ABB's Wodoslawski said that robot sales have been down @75% while their pick and place delta robots have sold more in the first half of 09 than they did in all of 08.

Automation World July 2009, page 15

[bigtank2185]

wow i'd have to say, thats good buisness!

[yatudlabiereicit]

... impressive ... i think i shat myself.

[Lollocide]

Well I believe it, but if you really want to be super-pedantic you could always put a clock in the middle-center of the background, so that the arms occasionally cover the face.

But still, I love the 'I'd like to say software [problem]] because I'm not a software guy' bit.

[MimixMotion]

LOL, very professional I know... the whole conversation is a bit disjointed and informal considering the subject matter.