WHITE PAPER

AN INTRODUCTION
TO COMMON
COLLABORATIVE
ROBOT APPLICATIONS
Published by Universal Robots
TABLE OF CONTENTS

Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 2

Introduction to Collaborative Robots . . . . . . . . . . . . . . . . . . . . . . . . . Page 3

Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 4

Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 4

Pick and Place . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 6

Machine Tending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 8

Packaging and Palletizing . . . . . . . . . . . . . . . . . . . . . . . . . . Page 10

Process Tasks (Gluing, Dispensing or Welding) . . . . . . . . . . . . . . . Page 12

Finishing Tasks (Polishing, Grinding or Deburring) . . . . . . . . . . . . . Page 14

Quality Inspection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 16

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 18

EXECUTIVE SUMMARY
This document is intended to give an o ­ verview how the robot handles the process, plus the
of collaborative robot technology, and how common accessories required to be integrated
it can be integrated into common ­industrial with the robot to successfully complete the task.
­applications to improve m ­ anufacturing The benefits of using collaborative robots in
­processes. Six application areas will be each type of application will also be presented.
­discussed, with an overview of each application,

2
INTRODUCTION TO COLLABORATIVE ROBOTS
The field of collaborative robotics has expanded significantly over the past ten
years, and is now by far the fastest growing segment of the global industrial
robotics market.

A collaborative robot, also known as a cobot, Often a worker’s job will consist only partially
in its most basic definition is a robot which of these repetitive, easy to automate tasks, so
has the ability to safely work directly along- the optimal configuration can be to have the
side human workers to complete a task. At person continue doing the variable parts of the
­Universal Robots (UR), however, we believe the task and make use of the robot as a smart tool
accessibility of the technology through ease of to speed up the repetitive parts to increase
deployment is similarly integral to the cobot his/her overall output. This only really works if
definition. A robot that can operate directly you can stand right next to the robot, passing
alongside and interact with its co-workers does objects back and forth.
open up a huge number of new possibilities
for task automation, but many of these possi- The productivity of an automated solution
bilities could go unfulfilled if the robot system is also inherently easier to monitor than a
is not easy to program, affordable, and flexible ­manual solution, as data on the n­ umber
enough that it can be re-deployed to different of ­cycles completed by a cobot can be
tasks at very short notice. ­conveniently extracted over a network and
incorporated into overall production data.
For this reason, we strive to make UR cobots
safe and collaborative, easy to program and Quality
deploy at an affordable price, to make ­robotic If one in every 20 products has to be scrapped
automation technology truly accessible to due to a lapse in concentration from a tired
everyone. worker, costs can rise significantly. The
­outcome is potentially even worse if the
So why should your company deploy a cobot? person inspecting the output of the process
What benefits could you expect to see? This misses a defect only for it to be noticed by
will inevitably vary from case to case, but here the end consumer.
are some of the common benefits:
The repetitive reliable operation of a
Productivity ­collaborative robot can ensure that these
There are still many things that a human can critical steps in the manufacturing process
do faster than a robot, due to our immense go off without a hitch. From putting on those
dexterity and ability to handle variations in our final labels in perfect orientation every time,
environment. However, for highly repetitive to moving an inspection camera through 15
tasks, involving objects that are predictable in different positions around the product to
both size/shape and the orientation in which make sure everything is just right.
they are presented, a robot that can work
for 24 hours a day without needing breaks Cost
in b
­ etween is likely to result in a significant A UR collaborative robot solution is often
­productivity gain. far more affordable than a traditional robot

3
or fixed automation solution. The ease of the event of the company losing a production
­programming and integration with certified contract or discontinuing a product, the same
accessories will reduce the implementation cobot can easily be moved from one task to
costs, both due to the reduced expertise and another due to its lightweight nature and easy
time required to complete the deployment. reprogramming.

The UR integrated adjustable safety system Understanding the versatility of the Universal
can completely negate the need for any fixed Robots system makes business owners’ decision
safety equipment such as fencing or a­ dditional to automate even more straightforward.
sensing around the robot (subject to risk
­assessment), which can often contribute Health and Safety
roughly a third of the overall project cost. The fact of the matter is that most of the
tasks that are commonly automated with
The human mind is a wondrous thing, and ­collaborative robots are not the ones that
it makes sense to challenge employees with people enjoy doing. They are often ­extremely
interesting tasks wherever possible and put repetitive, monotonous and sometimes
more cost-effective automated solutions into ­dangerous.
the mind-numbing repetitive tasks.
Regardless of whether avoiding a repetitive
Flexibility strain injury or improving morale by moving
There is simply less risk in investing in a to a more varied and interesting task, bringing
flexible, re-deployable piece of automation in a cobot really can improve conditions for
equipment such as a collaborative robot. In employees in the workplace.

APPLICATIONS
The flexibility of the Universal Robots system has led to UR cobots being
­deployed in a huge range of applications and industries around the world.
Here we will discuss 6 common application areas, how to implement a
UR cobot into the application and what accessories would be required.

Glossary I/O Interface – Inputs and Outputs used by the

To save repeating ourselves, we will first go robot to communicate with its s­ urroundings.
through some of the common terms relating Sensors are connected to inputs to ­detect
to collaborative robot applications and what signals representing things such as a p­ roduct
they mean. arriving, or a machine being ready for the
next operation. Outputs typically are used
Workpiece – The object of interest that the to ­actuate devices such as valves, or to
robot will manipulate, generally a product or tell a­ nother machine that an operation is
component. ­complete.

4
Polyscope – Universal Robots’ touchscreen Waypoint – A position in the robot’s
graphical user interface used to program the ­workspace that it will move the tool to.
robots.
Blend Radius – Sometimes the robot does not
Universal Robots+ (or UR+) 1 – Universal Robots+ need to go to the exact position of a waypoint
is a showroom of UR compatible accessories and stop. If the waypoint is just a rough guide
allowing applications to be built with increased as to the path the robot should take, it can
speed/simplicity and reduced cost. get within a certain distance of that point and
then start to curve toward the following point,
End-effector – The tool mounted on the end allowing for reduced cycle time and smoother
of the robot that allows it to interact with the movements.
workpiece, this could be anything from a simple
gripper to a screwdriver or glue dispenser. Vision System – Situations where the
­workpieces to be picked are presented to the
Pneumatic Gripper – A simple gripper p ­ owered robot in a non-standard position or orientation
by compressed air, this is inexpensive but may require a vision system to tell the robot
not necessarily good at handling a range of the precise pick parameters. Vision systems are
­different sized/shaped parts. becoming increasingly affordable and easy to
install, and add flexibility, but also complexity
Vacuum Tool (or Cup) – Objects with a flat to a solution. With that in mind, a m ­ echanical
­surface on the top can be easily picked up guide to force the workpiece into a fixed
with a vacuum tool (or vacuum cup), but ­location where feasible, is a valid and usually
not ­suitable when the product need to be more cost-effective alternative to using vision.
­rotated from the pick position and placed
upside down. A compressed air supply is Photoelectric Sensor – In order for the robot
needed for this method, but many industrial to know when a product is present and ready
­environments have this readily available so it is to be picked, additional sensing is s­ ometimes
not ­necessarily a concern. required. Inexpensive photoelectric or
­photo-sensors (light based) can be installed
Adaptive Gripper – An adaptive gripper, to detect a product arriving on a conveyor or
while more expensive than a vacuum cup or when a tray of products is placed down into a
­pneumatic gripper solution, gives the fl ­ exibility known location, without significantly impacting
of picking assorted sizes of objects and the the overall cost of the project.
ability to limit the force of the grasp. This is a
convenient choice as often the end-effector External Jig – In applications where the
will not need to be replaced when redeploying ­robot is required to place the workpiece
the robot to another application similar task. into a precise location an external jig may
be used, which could either be active or
TCP – Tool Centre Point. The part of the passive. In a passive jig the workpiece will
end-effector that comes into contact with the ­automatically centre itself when placed due to
workpiece, and therefore the point that we are the m
­ echanical properties of the jig, whereas
most interested in controlling the position of. an active jig can close around the work piece
securing it in a fixed location.

1) https://www.universal-robots.com/plus
5
PICK AND PLACE

A pick and place task is one in which the robot

is required to pick up a workpiece and place
“Implementing a UR
into another location and or orientation. Here cobot into a pick and
the handling of the workpiece is the key action place application can
rather than any other process being carried
out on it. Tasks such as machine tending and be done with minimal
packaging/palletizing could also be c­ onsidered cost and disruption to
as pick and place, but they are more ­specific
cases that will be discussed in the later
­production, as it can
­sections. In the simplest instance products will often slot straight into
be presented to the robot in a uniform layout an existing set-up.”
tray/ pallet or on a conveyor in predictable
position, where in more complex cases a vision
system may be required to determine product
orientation.

6
Manual pick and place tasks are often the most Programming
repetitive and mundane tasks in a p ­ roduction Programming of a pick and place application
environment, with the dull n ­ ature of the is generally very straightforward - move to a
task often leading to mistakes or ­decreased pick location, activate the end-effector, then
­efficiency if workers are not ­frequently move to a place location and deactivate the
­rotated. These are generally some of the least end-effector. In the simplest case both pick and
­enjoyable tasks that can also entail a risk of place locations could be fixed, but often one or
­repetitive strain or other injury if the object both positions would need to be adjusted each
is of substantial weight. When selecting a cycle, as they are in a grid or stack o
­ rientation
first task for cobot automation, a pick and (conveniently achieved using palletizing or
place task is an excellent choice as the highly seek Wizard functionality in Polyscope), or
­repetitive nature and simple movements make ­vision guided (achieved using a third party
it very easy to set up. vision plugin from UR+).

Accessories Typical Programming Time – Less than 1 hour

End-effector – Either a vacuum cup based
­solution or a gripper could be suitable, Benefits
­depending on the size and shape of the work- Implementing a UR cobot into a pick and place
piece. If it has a smooth, flat upper surface and application can be done with minimal cost and
will be placed in the same orientation that it disruption to production, as it can often slot
was picked, a vacuum based solution is a good straight into an existing set-up. Companies will
choice. If it has no flat surfaces, or needs to often have to rotate workers on a monotonous
be placed in a different orientation, then an pick and place task every couple of hours as
­adaptive gripper may be required. productivity starts to drop, where as a cobot
allows productivity rates to be maintained
Vision System – If the products arriving for the across multiple shifts.
robot to pick are in a non-standard ­position/
orientation then it may be necessary to
­integrate a simple vision system to detect the
orientation of the part.

Protective Suit – If the machining process

is particularly damp or dirty, additional
­protection for the robot may be required,
in the form of a protective suit, available
from UR+.

7
MACHINE TENDING

Machine tending is another task very w

­ idely As a manual process, machine tending requires
executed by Universal Robots around the workers to stand for hours on end, placing
world, where the machine being tended ­individual parts into a potentially d
­ angerous
could be anything from a CNC to an injection machine. With the addition of a robot,
moulding machine, a laser engraver to a metal ­however, the same employee can handle a
stamping press. far greater number of machines, being able

“The cost of a cobot is usually significantly lower than

the machine it is tending. By allowing the machine to
run 24 hours of back to back cycles, this will likely pay
for the cobot tending it in a very short period.”
8
to supply products in large batches instead of Some I/O interfacing is usually required for
­individually, thus increasing productivity. The the robot to tell the machine when the part
robot will pick a blank unprocessed product is in place and the cycle can begin, then
from a tray, stack, conveyor or some other for the machine to tell the robot that the
feeder configuration, and place it into a fixed ­cycle has finished and it is ready for the next
location in a machine. Once the machine part. For machines like CNC or injection
cycle completes, the robot will take out the ­moulding, ­additional signals will be required to
­completed part and put in another blank. If ­synchronize the handover of the part and to let
the machine cycle is long enough to allow the machine know when the robot is outside
it, one robot can tend multiple machines the machine so the door can be closed.
­simultaneously, accelerating the return on
investment. If the robot is tending multiple machines,
some additional logic will likely be necessary
Due to the small footprint of a UR cobot, it can to decide which machine to tend to next in
often be installed without adjusting f­ actory the case that they do not have identical
­layout, and leaving enough space that an cycle times.
­operator can still access the machines if r­ equired.
Typical Programming Time – Less than 0.5
Accessories days
End-effector – A machine tending application
will often utilize a dual gripper configuration, Benefits
with one gripper picking and placing the blank The cost of a cobot is usually significantly ­lower
and one the completed part. This results in a than the machine it is tending. By allowing
reduction in cycle time, with the robot being the machine to run 24 hours of back to back
able to handle both parts in a single movement ­cycles, this will likely pay for the cobot ­tending
inside the machine. it in a very short period, even faster if the
­cobot is tending multiple machines.
Vision System – If the products arriving for the
robot to pick are in a non-standard ­position/ The small footprint of a UR cobot means
orientation then it may be necessary to that machine layout does not need to be
­integrate a simple vision system to detect the ­adjusted, often on larger machines. The robot
orientation of the part. could even be mounted onto the side of the
­machine, requiring zero footprint.
Programming
Robot movements for a machine tending
application are generally very simple, moving
between ingoing/outgoing product positions
and the fixed machine position.

9
PACKAGING AND PALLETIZING
“The benefits of
automating a packaging
task are similar to
generic pick and place
– increased productivity
over multiple shifts.”

Before any product leaves a factory, it is highly are extremely easy to handle, though a s­ imple
likely that it needs to undergo some form of vision system may be required to detect
packaging ready to be shipped onto its next ­orientation of parts if not uniform. If less rigid
destination. The packaging and palletizing products such as sachets are presented, and
task could involve packaging a product by need to be tightly packed into boxes, extra
placing it into a shrink-wrapping machine, consideration on the handling method is
picking p ­ ackaged products from a conveyor ­required, but still entirely possible.
and c­ ollating them into boxes, or placing these
boxes onto a pallet ready for shipping. For business running high mix low volume
production, rapid product changeover is key,
Such tasks are all extremely repetitive and so the easy programming interface provided
generally involve small payloads making them by Universal Robots is perfect for this, allowing
ideal for automation with Universal Robots. for reconfiguration of an application within a
Rigid products arriving in standard o
­ rientation matter of minutes.

10
Accessories Programming
End-effector – Packaging and palletizing For the most part, setting up the program for
tasks are commonly handled with an array this type of application is very similar to a pick
of v­ acuum cups to pick up and release the and place, with the pick usually from a fixed
­products. In the simplest form, these can position, triggered by a sensor input. The place
be a­ ttached to a single flat plate, but can positions will likely vary for packing into boxes
also be set up so that their positions are or palletizing, with either a horizontal offset
­reconfigurable, allowing a range of different within a layer, or a vertical offset between
sized products to be picked with the same tool. ­layers, both of which are straightforward to
set up within Polyscope. For a more complex
Conveyor Tracking – Synchronizing the ­palletizing pattern, it may make more sense
­movement of the robot with a conveyor to to enter the dimensions and locations of
pick products on the fly is simple with the UR the place positions parametrically instead of
conveyor tracking wizard. Simply connect a ­teaching/offsetting them manually.
position sensing encoder to input channels
in the controller (or via Modbus fieldbus), Typical Programming Time – Less than 0.5
­configure the direction and speed ratio, and days.
the robot is ready to track the movements of a
variable speed conveyor. Note that if this is an Benefits
end of line conveyor, where the products hit a The benefits of automating a packaging task
mechanical stop at the end, it is not necessary are similar to generic pick and place
to track the movement of the conveyor as the – ­increased productivity over multiple shifts.
pick position is fixed. The Universal Robots system comes fully
equipped with interfaces to integrate with
Vision System – If the products arriving for the
­existing p
­ roduction lines out of the box,
robot to pick are in a non-standard ­position/
­without the need to pay for additional add on
orientation then it may be necessary to
features, reducing costs of implementation.
integrate a simple vision system to detect the
orientation of the part.

I/O Interfacing – A few inexpensive light

sensors (photoelectric) connected directly into
the controller will allow the robot to detect the
presence of arriving products and the box into
which they are to be placed.

11
PROCESS TASKS (GLUING, DISPENSING OR WELDING)
In a process task, whether it’s gluing, dispensing
or welding, the key details are the same: the
robot moves a tool through a fixed path while
the tool interacts with the workpiece. In each
of these process tasks, it takes a significant
amount of time to train up a new employee
to be able to control the numerous variables
required to attain an excellent quality finish. If
this control can instead be copied directly from
one robot to another, it becomes a c­ onsiderably
more straightforward process. It is also
­potentially very challenging for a human worker
to dispense a repeatable amount of material
across a fixed path with minimal variation cross
a whole shift, whereas a robot can control the
required variables relatively easily.

A traditional robot welding system for

­example, requires both significant welding
and robot programming expertise to set
up, and is usually far more expensive than a
­welding torch alone. If the difficulty of the
robot p
­ rogramming aspect can be simplified,
anyone who knows how to weld can set up an
­automated robot welding cell.

“Increased productivity
and quality can be
expected with Universal
Robots carrying out
a process task with
a consistently accurate
output over tens of
thousands of cycles
and more.”
12
Accessories Benefits
End-effector – A process tool is required, Increased productivity and quality can be
which could be a welding torch, sealant, glue expected with Universal Robots carrying out a
or solder paste dispenser to name but a few. process task with a consistently accurate o
­ utput
Often the tool does not need to be ­specifically over tens of thousands of cycles and more. The
designed for robotic operation, potentially flexibility and easy programming of the robot
­reducing integration costs. Turning on and off mean updating a program to handle a new
the tool is normally achieved using standard workpiece is possible in a matter of minutes.
digital I/O signals, with the potential addition of
an analogue signal to control deposition rate.

Programming
Polyscope programming interface includes
a process move option, which maintains a
­constant TCP speed, meaning that if the r­ obot
tool is depositing material at a constant rate, the
system achieved constant c­ overage throughout
the programmed path. The s­ implest method of
programming a process task is to define the key
waypoints within a process move, along with
blend radii allowing the robot to curve around
the corners in the path.

If the system is required to deal with a large

number of rapidly changing parts, for which the
CAD models and CAM process paths are r­ eadily
available then it may be more convenient to
­import these paths directly into the robot
­program rather than teaching them m ­ anually.
This can be achieved easily with a number
of different third party software packages,
­outputting programs from the CADCAM data
that can be directly executed on the UR system.

Typical Programming Time – Less than 0.5 days

13
FINISHING TASKS (POLISHING, GRINDING OR DEBURRING)
“As with a process task, A finishing task requires the robot end-effector
to apply a force across the surface of a work-
the possibility for sharp piece to remove a certain amount of material.
productivity and quality Polishing, Grinding and Deburring are different
improvements are in amount, form and location of material to be
immediately clear when removed, but the requirements of the robot
automating a finishing are essentially the same.
task with a UR cobot.” When a person completes a finishing task
with a manual tool, this often requires the
worker to apply a large amount of force to the
workpiece, generating a significant amount of
­vibration which over time can lead to injury
which could be avoided with robotic operation.

14
Finishing tasks often utilize the process move Programming
command mentioned in the processing task The program logic for a finishing task in its
and a robot can either be manually taught simplest form is very straightforward, simply
the path to complete the task, or it can be navigating along a fixed path and activating the
exported from CADCAM data directly to a UR tool at the relevant points.
program.
Typical Programming Time – Less than 0.5 days
Utilising force control can also potentially make
the robot more robust when dealing with parts Benefits
of different dimensions. This can either be As with a process task, the possibility for sharp
achieved with the robots internal force sensing productivity and quality improvements are
capabilities or a wrist mounted external force immediately clear when automating a finishing
torque sensor depending on the sensitivity task with a UR cobot. If done manually, these
required. tasks also require the worker to apply a force
with a hand tool repeatedly over many hours
Accessories potentially leading to repetitive strain i­njuries,
End-effector – A finishing tool is required for so health and safety benefits can also be
this type of application, which could either ­expected.
be designed for general purposes (­manual) or
specifically robotic purposes. While m­ anual
tools are generally cheaper, and can be
­powered on and off by the robot, their fine
controls are generally controlled by push
­buttons, while robot tools include an I/O
­interface for direct control.

Force Torque Sensor – If the task at hand

­requires force control in the sub-Newton range
(phone case polishing for example) then you
may want to consider adding additional s­ ensing
to the robot for finer scale force control.

15
QUALITY INSPECTION

Quality Inspection involves full inspection of “A machine vision system will

a finished product, especially one that is the
result of a precision engineering process,
usually have a faster, more
often requiring high-resolution images to consistent output than a
be c­ aptured from many different angles to person inspecting a product,
­confirm that all of the surfaces and d
­ imensions leading to improved quality
conform to the required specifications. The and productivity.”
cameras capturing these high-resolution
­images are commonly costly, so requiring say
10 cameras to fully inspect a product is not
cheap. If a single camera is mounted on a
robot and moved around the product to all of
the fixed capture positions, however, the costs
of this type of automated inspection system
drop significantly.

16
In order to make the system solution fully Benefits
autonomous the robot may also be equipped A machine vision system will usually have a
to move the parts in and out of the inspection faster, more consistent output than a person
jig, either from a tray or conveyor therefore inspecting a product leading to improved
requiring minimal supervision. ­quality and productivity. As mentioned above,
compared to mounting multiple cameras
Accessories around an inspection cell, a cobot based
Vision Systems – The main accessory involved ­solution can potentially be significantly
in a quality inspection application is the vision ­cheaper while fitting into the same compact
system, including the camera and software to workspace.
process the images. Often the system required
will be a more high-end system than the sort
of system required to simply locate a part
for the robot to pick it, but with i­ncreasingly
­capable vision processing algorithms,
­configuration doesn’t necessarily have to be
super complicated.

End-effector – If the robot is also required to

handle the part in and out of the i­nspection
­location then a gripper should also be
­mounted on the robot alongside the camera to
move the part.

External Jig – Once the robot places a part it

may need to be clamped in place in a jig to fix
the position.

Programming
Setting up this type of application is g­ enerally
very straightforward. In the instance that
the robot does not need to handle the part
­directly, the program will simply consist of
moving to fixed waypoints then t­ riggering
the camera via digital I/O or Ethernet
­communi­cations.

A pick and place operation may also be added

to this in the case that the robot handles the
part, but this is again a very straightforward
process.

Typical Programming Time – Less than 2 hours

(excluding vision system)

17
CONCLUSION
For a newcomer to robotic automation, this paper provides basic insights
into how collaborative robots can be implemented in common tasks and the
­benefits of doing this. The common applications discussed above can be rolled
out in a very short period, especially when utilizing the huge range of plug
& play accessories available in the UR+ showroom.

At Universal Robots, we want to create an By putting automation in the hands of

­ecosystem whereby automation is made ­operators, we allow small companies with low
­accessible for all. Beside content in this white volume production to join larger companies
paper, first-time adopters can learn a lot of the in automating their processes without the
required skills for implementing a UR cobot need for high initial investment and extensive
from Universal Robots Academy 2, which is ­automation experience.
Universal Robots’ e-learning program o ­ ffering
a free, interactive online training course
aimed specifically at newcomers to robotic
­automation. To complement this, UR channel
partners around the world are also capable of
delivering standard UR face-to-face technical
training programs to guide existing technical
staff so they can implement robot ­applications
on their own. Affording existing staff the
opportunity to operate and even program a
collaborative robot system makes for a much
more rewarding experience, and a more
­attractive workplace environment for new staff.

2) https://www.universal-robots.com/academy

LEARN MORE
Please contact [email protected] or visit www.universal-robots.com
to take the first steps in your collaborative automation journey.

18