Go Down

Topic: Delta robot --> possible project realisation (Read 413 times) previous topic - next topic


Hi to all.

After few weeks i started similar topic but i guess it's better to start a new one because i have some new information and the concept has changed. When someone will read this, please be so kind and try to answer each question. Am writing down everything so i can document it for possible future use. In end effect, whole community will gain from it.

SPECS and questions:

1.) Dimension of longer arm will be cca 1 meter and shorter one ( the one attached to motor ) will have length of cca. 30 cm. Everything will be made from aluminum. Weight of whole arm will be between 0.5 -1KG. When i add gearbox to the motor, what kind of power for this motor, should i take into consideration ? The reason am asking this is because i want to use RADDS but i can not find any relevant info on motor power consumption. Can i use some 50W stepper motor for such setup ?.

2.) From my point of view, i should need some stronger motor for one additional thing...
Number of the objects that will be picked from the conveyor track will be between 50 - 60 per minute.
This can not be done with some small stepper motor ( because of the gearbox part ).

3.) Image recognition
First part of the project is to detect colored items and sort them out. For this am planning to use arduino color sensor or maybe some camera. What would be better option ?

4.) I can use raspberry and camera as image detection and processing part. But,i dont know how fast will this this RSP --> arduino part work...Basically, i want to avoid using RSP for anything else accept for image processing and sending signal to arduino...

Best regards !


I do not see this project getting off the ground.

First when you type try not to do things like "cca" and "RADDS" because people (like me) may not know what that means.

"1.) Dimension of longer arm will be cca 1 meter and shorter one ( the one attached to motor ) will have length of cca. 30 cm."

Does this mean you will have Two arms and one is driven by a motor, the other not?

"Everything will be made from aluminum. Weight of whole arm will be between 0.5 -1KG".

This is good you give people something to work with!

"When i add gearbox to the motor, what kind of power for this motor, should i take into consideration ?"

What kind of gearbox and what will the gearbox be driving.  Eg a pump or claw or elbow or sholder joint?

"The reason am asking this is because i want to use RADDS but i can not find any relevant info on motor power consumption. Can i use some 50W stepper motor for such setup ?."

See not knowing what RADDS is leaves me with a blank.  Asking for power consumption at this stage is like me asking what size motor to put in my car after i tell you it will have 4 wheels.

I DO NOT do this to criticize your post but to point out how it is hard for another person to help with the info you provide.

In relation to some other parts like the pick objects from a conveyor belt,

How heavy are the objects? This matters with respect to power.

You say one every 50 to 60 seconds, this means your arm will move from the conveyor to wherever and back again in ONE SECOND?  Do you realise how fast that is and how much power that will take?

Think of this, your arm must travel say 50cm from the conveyor belt and then back, take into account the time for your 'claw' on the arm to grab and release the object and say the object is 250 grams in weight.
This means your arm must move into position on the belt to grab the object, close the claw (lets say the claw takes 200ms to close. This is a made up figure!) Then it must accelerate tis 1 kilo plus the 250 gram object weight to somewhere around 2 meters per second or 7000 kilometres an hour for 50 cm then come to a stop, open the claw then reverse direction bact to the conveyor belt.

Does not seem possable.  What do you think?
I will stop hereasgoing any further is no good.  You might have a look at videos of factories that produce say lollies and small plastic toys and see how they run.  I do not know of a factory that runs different objects on a conveyor belt except recycle centers.



Apr 19, 2018, 08:17 am Last Edit: Apr 19, 2018, 08:20 am by arnix
Hi Daz and thank you for reply. You are absolutely correct about writing part.
I will write additional info and comment when i get back from my job...

This was the main topic:

Btw. instead of clamp we will use vacuum pump and sorting capability should be between 50-60 part per minute.

Ok. gotta go,hear you afternoon.
Best regards !


I think you missed the whole "Arm moving at 7000 kilometers per hour thing"

Try doing it with your own arm and see what happens.



Apr 19, 2018, 04:47 pm Last Edit: Apr 19, 2018, 04:53 pm by arnix
Maybe it's better to describe what we are planning to do ( in future ) and where i would like to start.

This is exact the same stuff that we would like to create in year or two (scroll to 1,5min ), but with different materials and of course, not so perfect. I look at this as onto some guiding goal :-)


This is where am planning to start, but with bigger dimensions:

1.) If we put sw part aside, what do you think, what kind of motors and gearboxes are implemented in robot from the first video ?.



You DO NOT have a budget to build one of those and my statement still stands about the 7000 kph.

Forget the software it will be a while building that arm and take note of your article on the second link.

In robotic path planning, the goal is often to run a trajectory at the highest possible speed. Such a trajectory is constrained by the maximum torque available in each servo axis. Linear cartesian movement of a robot requires coordinated movements of all axes. One axis is not allowed to run ahead or behind the coordinated scheduled trajectory. Such coordinated movements at highest possible speeds, is only possible when the path is planned according to an accurate dynamic model.

More torque, more speed. That is untill the frame flies apart.

Sorry but this is out of my league

Good luck



Great info ! Thank you very much.
Let's forget for a while, that we are planing to do this...

What do you think what kind of motors are used in this setup ?
I keep asking this because i would like to know / understand what kind of controller is used in this setup.
OK, i guess that they use some company made specific one, but maybe there is similariti wit some commercial versions ?

This would help me a lot in defining future steps...


This still reads like a homework assignment. You've been given a problem that can be rearranged into a 1-dimensional problem but with the constraint of "Don't alter the belt" and your own ideas, you've made it into a 4-dimensional problem.

Move the input or the output or run the constrained belt onto a new one that will let you drop the products past a sensor and actuator. The "row of air jets" solution is extremely robust with no moving parts except for the air solenoids.

If this was my assignment, I would get the machine-vision side working first. Even if someone gave you a delta robot, it would not work without knowing where to go. So make it circle the parts on a video screen or something. Obviously plan to have a later digital output to a robot but make the vision work first without trying to build the robot at the same time.

Note that Arduino is completely unsuitable for machine vision so you will have to look elsewhere. The closest thing is the Pixy Arduino camera and that may not be able to discriminate your products.
"The problem is in the code you didn't post."




2 - Never use your own TLA's cca -WTF nothing wrong with circa or about

3 - Add links to things such as "delta robot".

4 - Do your home work first - One Google showed that a delta robot had been made using a 386 (aka an UNO)

5 - Don't talk about "whole community will gain from it." it just makes you look like a pretentious fool, and even more so after 4 above.

Now having said all that yours (when sorted out)   is one of the better questions I've seen in a while.


PS -Wait for part two - I don;t trust my net connection




Your spec should read

You want - a delta (you put in the link) robot with the ability to "see"/"sense" objects with a view to doing somthing with them. (provide link to say the pancake stacking delta bot".

Now we can look at design options. The first thing I see is that we have two problems.

Control of the bot and spotting the object.

Control of the bot (inc the kinematics) by an UNO, proven as possible by our home work.

Can an UNO do the object spotting? Well yes and no. The op must answer this question!

Wait for third part


PS red wine ........



-starship voyager --> captan's log --> supplemental one / two / tree :-) :-)
I have drink 4 beers till now:-) and you still didnt post the third part :-)

-nope,it's not homework project.
I would get the machine-vision side working first
- totally agree ! This is why i want to use raspberry for this part.

Vision part --> raspberry
Stepper controler --> RADDS
Arduino due
RSP send coordinates --> arduino reads it and executes motor movement + vacuum pump.
General update on project :
After today's meeting we have concluded that it would be better to use two smaller robots on the conveyor track. So, if the track is wide 120 cm, first robot will cover 50cm and the second one the same. Gap between two "lines" will be created with shaker and blocker.

With this setup i think i can use RADDS and some standard stepper ?
Like i wrote before, i will try with pixy camera but i think RSP <--> arduino connection would be better choice...

Going to take a beer now :-) 


You want to build two delta robots to each cover half of the conveyor? You know a delta has a large footprint outside of the working volume?

It seems like the bosses are still thinking of having two people, one on each side of the conveyor. That's a poor design for a robot. Robot arms work differently to human arms.
"The problem is in the code you didn't post."


Apr 22, 2018, 04:40 pm Last Edit: Apr 22, 2018, 05:06 pm by arnix
It's still our choice... There was suggestion for design with two robots because that was the only way to get 45+ pickup's per minute ( smaller dimensions, faster speeds, ready made controllers, budget range etc.). Suggestion was to put one robot arm prior to the second one ( one at the begging of first track and second half meter away but on the second track). I have accepted this as suggestion but in the end it will be our choice.

I just want to say, that I really appreciate yours ( and others ) comment and ideas..
Such wish to help is something that is really inspiring.

I will make one smaller test model and then we will see how this works.
If i can get some good pcs/minute range we can build a bigger one. 

Do you agree with HW specs that i want to use ( RSP/radds/arduino ) ?
Btw. do you know where can i find some specification for radds motor support ?
Can i use 5-10A motors ?


Go Up