Industrial and service robotics people are marching toward the two new big pies. We are starting to witness the convergence of industrial and service robotics, both from a technical and commercial standpoint.
Schunk Expert Days - Convergence of Industrial and Service Robotics 2014
I'm very happy to have the opportunity to attend -- and to speak -- at this
event. The fact that the Expert Days exist is an illustration of the topic I am
going to discuss: the convergence of Service and Industrial Robotics.
We started Robotiq in 2008, what a great year to start a company in robotics,
wasn’t it? Actually it was not that bad as we had nothing to sell back then. We
were fresh out of a robotics laboratory, looking at all the markets where we
could apply our know-how. We knew robots but nothing about the robotic
industries and initially we looked very broadly at different markets.
There were predictions that service robotics was about to explode. Talking
about service robotics, you could hear the classic “robotics is where the PC
industry was in the 70’s”. Other people were comparing robotics to consumer
electronics. PC or electronics, whatever, service robotics would be very big.
But how would we make it happen? That, nobody really knew.
The more I learned, the more it seemed like an exciting opportunity! But at the
same time, I had a feeling that no one had clearly figured out HOW service
robotics would become a big thing.
This is a picture from a Lego book I read with my kids. In the book they have
different robots: walking robots, rovers, flying robots and… “real robots”. You
know those yellow robots welding car bodies.
After looking at all the markets, we realized what I would have learned if I had
read that book.
In 2008, real robots were the industrial robots. Even though 2008 and 2009
were disasterous years for industrial robotics, the business model was proven
and robots were being sold. Industrial robotics is where business could be
done from day one, so we designed our products, robot hands and sensors,
for these robots, because they were the ones actually existing in a number
sufficient so we could pay the bills.
In 2008, the two industries, at least in North America, were looking at each
other with a great deal of incomprehension. Service robotics people were
looking at industrial robots as boring, prehistoric technologies. The industrial
people were looking at service robots as gadgets for kids in a quest for a
Today, things have changed. There are some service robotics companies that
have had a great impact and that have figured out a way to great commercial
success. And the possibilities opened the eyes of the industrial robotics people
and have them saying “Hey wait a minute, we are robotics, we want a piece of
In fact, there’s not only one pie, but a few different pies.
The industrial market is a mature market with many established players. It is
still growing at an interesting rate. But there is a whole new pie being defined
right next to it, which will be larger than the current one, and it is the industrial
robots for small and medium factories. And then there are all the applications
outside the factories, they are an even bigger pie.
Industrial and service robotics people are marching toward the two new big
pies. We are starting to witness the convergence of industrial and service
robotics, both from a technical and commercial standpoint.
Before diving into the topic, let’s do a step back. What is industrial robotics,
and what is service robotics?
A very prominent figure of robotics, Joseph Engelberger, who co-founded the
company that sold the first industrial robot, was already struggling with the
definition of a robot.
What a great picture. See that smile? Even though it was an industrial
machine, you can already see the fascination for service robots, and
especially for robots serving drinks. Right from the start, we did not intend to
separate industrial and service robotics, we liked all robots. But manufacturing
and more specifically car manufacturing was the only place we could make the
case for robots.
Some things haven’t changed. Even today the robot videos getting the most
hits on youtube are the ones with a PR2 fetching a beer. You see, a drink, a
robot, 50 years later, same smile. Roboticists just seem to love beer, maybe
that why the German do so well at robotics.
Even if the IFR has a formal definition, it’s difficult to be perfectly clear on the
nature of industrial vs service robots.
Usually, when we talk about an industrial robot, this is what we mean: a robotic
manipulator -- an arm –doing manufacturing tasks (welding, assembly, pick
and place, etc.) in a factory.
And service robots, well, it’s all the rest, split in 2 subsets of service robots.
1- Personal or domestic service robots.
2- Professional service robots, sold to businesses or government
organizations across various industries.
Examples are shown on this slide.
Military (teleoperated on the ground or autonomous airplanes)
Unmanned underwater vehicles
Robots for agriculture, etc,
Most of them are completely new platforms and systems, using different sets
of technological buildings blocks than what it is used in industrial manipulators.
So it is clear that they are service robots.
But what are these?
A mobile robot, ok that is a service robot.
l But it is used in a factory, so should it be called industrial?
l Even though it is used for logistics operations?
And what about this one?
An ABB robot manipulator. ABB makes industrial robots, right?
l But it is not in a factory.
l And it is not doing a manufacturing process.
So is it a service or an industrial robot?
These examples show how the line is being blurred between both industries.
What's confusing is that a combination of attributes are used to define what is
an industrial robot. And some robots have a subset of these attributes, so we
can’t tell if they should be considered industrial or not.
We have platforms doing different applications in different markets.
Here is another representation of what we mean by industrial robots. All the
industrial platforms are based on some technological building blocks.
Then you have on the service side new platforms, based on other
technologies, doing different applications in markets outside the factory.
Here are different things happening right now:
Service robot technologies are making their ways onto new manipulators in
• New platforms doing new applications are making their way into the factories
And the industrial people are also looking to have their platforms used in the
service application and markets.
These four topics will be explored in more details in the rest of this
For the core of industrial robots, the building blocks are made from;
Kinematics, motor drives, safety, manufacturing process knowledge.
All of them use proprietary software and closed controllers.
They are programmed on a teach pendant or using offline tools.
Most of them have for some time added integrated vision to their offering.
They are available at different scales of reach, payload and speed.
They are very mature platforms, they are real workhorses and I am still
impress every time I stand beside one, whether it is super strong or super fast.
Professional service robotics use the following building blocks:
Open Source Software,
Embedded electronics and Connectivity.
They are leveraging open source software and consumer electronics to speed
up development and reduce costs.
They are rapidly evolving platforms.
Some times they are not as mature and reliable as they should be, but that will
come with time and experience, as they build corporate knowledge of what
can go wrong out in the field. Platforms will also open to component providers
that will add value to their solutions.
From a technology standpoint, how is the convergence happening?
First, in the factories you start to see manipulators that include service robot
concepts, such as direct teaching and safe human-robot interaction.
The closest thing to an industrial robot that includes service robot concepts is
probably the Universal Robot that all of you know. Their specifications are not
as impressive as a standard industrial robot. They are not necessarily more
affordable to purchase either, but the cost and complexity of integrating and
reprogramming them is significantly lower. They still have a teach pendant, but
they can be taught by dragging their arm around the desired trajectory. Their
controller runs on Linux. A few thousand of them have already been sold, and
a majority of them, to first time robot users, remember that second pie
overlapping the traditional industrial robot market.
The Kuka Light Weight Robot has been on the market for several years and
new models are coming up. It also includes several service robot building
blocks. It is also safe to be around and can be put in a control mode to offer a
desired force feedback to the operator or to do assembly.
You also have dual-arm platforms that include other service robot concepts. The main
one being advanced perception using several sensors.These guys are very close to
human scale and are made to be put side by side with a human worker without
modifying the work environment.
ABB unveiled Frida some time ago, but are very quiet about when this will become a
product. It is supposedly targeted at tasks such as electronics assembly. It is safe to
be around considering its design and the very low forces it can exert.
Kawada has adapted the technology that it has used in research for a long time to
humanoid robots, to come up with a robotic co-worker called Nexstage. Nextstage
has various sensors, as well as limited force at its joints for safe cohabitation with
humans. Like Frida, it is not programmed by demonstration.
And Baxter from Rethink Robotics includes a lot of service robotics concepts. It does
not limit the human-robot interaction to following an operator's movement. It can also
give comprehensive feedback using its ipad like face. Rethink’s approach was to
develop a whole new set of building blocks to have very cost effective mechanics and
leverage the ever increasing power of electronics and software. They were also
targeting that second pie of first time robot users. In the US, they really created a
huge buzz among this market. What they realized after launch is that maybe they put
too much emphasis on the cost-effective mechanics concept and only recently with
new software releases that improved the control are they getting some traction.
These new manipulator platforms have highly desirable features that
traditional robots don’t have, one of them being lead-through teaching.
We can imagine that industrial robot users will look at the functionalities of
these new robots and wonder why they can’t have them too. Some people,
including ourselves, are working on this.
You have two examples of add-ons here that can be used on traditional
industrial robots to do direct teaching of trajectories.
Being able to do lead-through teaching or having user friendly visual interfaces
should speed up programming and reduce the skill set necessary to integrate
an industrial robot. As robot programming still represents an important barrier
to first time acquisition, this should help a lot of small and medium companies
get their first industrial robot, even if they have a lot of small runs and
changeovers. The great thing here is that you can use the industrial platform's
reliability and maturity.
We see that there are some technological building blocks from service robotics
already in the factories. What are the other ones that could make a difference?
In the US, many key end users want to use the capabilities of ROS to solve industrial
problems. So for me there is little doubt that ROS will make its way into the industry
soon. Large end users in aerospace, automotive and construction equipment are
investing in ROS Industrial. The two main motivations that I hear from them are:
- First, a way to generate robot programs from their CADs and
- Second, a tool to break free from proprietary software, to be able to use standard
programming technologies and port solutions from one platform to another.
Robotics is a small industry compared to many others. We have to recycle
technological building blocks that other industries ruggedized and produce in high
enough volume so that they are very affordable. Examples are sensors like the kinect
or mobile phone platforms that have built in sensing and computing at ridiculously low
I came across a back of the envelope calculation the other day looking at the cost of
an iphone if it would have been built in 1991, considering the computing, sensors and
bandwidth it uses. Do you know how much an iphone would have cost in 1991? More
than 3.5M$.Can you think of any industrial mechanical device that has dropped in
price in a similar fashion in the same amount of time? This is the approach, as I said,
put forward by Rethink robotics. Maybe they pushed it too far, but the idea is still
valid. Robotics has a lot to do with mechanics. Good mechanics is important in a
robot, just like it is in a car. But we have to relax the constraints on mechanics to
reduce the cost and make up the difference by using cheap electronics and clever
There are also new platforms being deployed in the factory environment to do
Think of all the fork lifts that are being driven in factories all over the place.
Moving the raw goods when they come into the factory or the finished ones to
the shipping area seems to be a good fit for robots. This involves a lot of labor
and moving material around is not adding any value to what’s being
So there are things around the manufacturing processes in the factories for
which robots can be used. What's interesting to realize here is that these
problems are really logistics problems. You could eliminate the bloc in the
middle in this image and have the representation of a distribution center.
Distribution and logistics are a large market actively looking for automation
with robots, as the acquisition of Kina Systems by Amazon proves. This
market is changing fast because of the rapid growth of e-commerce and it will
drive a lot of innovation. So we can expect a lot of cross-pollination between
the two industries in the near future. Manufacturing technologies getting into
logistics and vice versa.
Every time a crash in the automotive industry occurs, all the industrial robot
companies say ‘never again!’, and start looking at other applications,
sometimes even outside the factories. It can be a growth vector, but they
should not forget who they are and try to focus on their strengths. Who needs
up-time, speed, repeatability and does not care about cost of integration? It is
definitely not the home or office environment. There might be some
applications where these features are desirable, thinking nuclear or
entertainment. This is interesting, but they are small niches. My bet would be
to look more closely at automation in construction and in the oil and gas
industry. These are two tough places with high labor costs and challenges to
find people where the industrial robot features could be valuable.
New manipulator robots targeting manufacturing tasks will be sold to people
who don’t have a robot right now. But this new pie overlaps with the current
industrial robotics market. Current users of industrial robots are also looking at
those new robots for areas in the factory where current robots would not work.
Then they end up comparing the specs.
Specs are numbers written black on white. It is easy to compare them,
engineers and technicians love to do this, they are trained to do this. It is easy
to compare speed, payload, repeatability and number of cycle ratings. And all
these specs are in favour of the current industrial robots. But how can you
quantify ‘flexibility’ and ‘ease of use’? Do people really understand the cost of
integration and the benefits of easy programming, or repurposing a robot?
This is a challenge for the marketers and salespeople bringing the new
manipulator platforms to the market.
If service robots will put pressure on the industrial ones to add functionalities,
industrial robots will put pressure on the new robots to improve their specs, or
at least to explain them better.
Ok, I talked a lot about how service robotics will influence robotics in the
factories. In fact, I also think that factories are the best labs to mature
technologies that will end up in totally new robotics markets.
Problems are simpler in factories because they are better structured. We
usually work with a finite set of parts in a well defined sequence.
Manipulation is not an easy problem. This is one reason why all those
telepresence robots don’t have an arm even if they’d like one to open doors.
Manipulation is complex, still, it was the first task done by a robot 50 years
ago. It is because they could remove all the uncertainties in the environment.
Simple repetitive movement could do the work.
Before going to fully unstructured environments like homes and offices, to me
it makes sense to break the problem into smaller, simpler parts in the industrial
world. So before Rosie Jetson hits the market, it makes sense to practice in
more structured applications.
Plus, in manufacturing, prices can be justified, because you’re selling to a
company and if you can prove the ROI, it’s already a good starting point. This
will give time for the technology and companies to evolve.
I’ve talked a lot about the technology differences and convergence. How will
this impact the structure of the industry?Let’s look at how traditional industrial
robots end up on the production floor today.
Because they involve advanced, specific programming skills and tooling
customization, there needs to be someone integrating them. This can be an
external integrator, that makes his money charging for engineering hours. The
integrator can also be within the company. Integrators also put together
everything that comes into the space around the robot. For every dollar spent
on a robot, there is usually another 2-8 dollars spent on integration and
Once this relationship is established between the robot manufacturer and the
end user, it usually lasts for a long time. This is central to the robot
manufacturers’ strategy: get a first robot in there. Then the customer has to
invest so much to learn the proprietary tools, that it does not make sense for
him to switch robot suppliers. This is a funny situation: the end user pays to
lock himself to a robot manufacturer. The more he pays, the more reason he
has to stick to his robot color.
Now what happens if a robot is so user friendly that it can be installed and
programmed by anybody on the factory floor?
Well if the production people can do their own programming, you greatly
reduce the cost of integration. Then it starts to make sense for the small shops
to purchase their own robot. They can prove the return on investment and their
current staff can use the robots. And this is what is happening with the new
collaborative robots opening this market.
If you are a systems integrator, does that mean that you’ll go out of business?
No, but it means you’ll have to become master of specific manufacturing
processes and make sure you push you own limits to be able to bring value to
end users in applications that are difficult to automate. Because the simple
things, the end user will do himself.
This situation can be compared with the IT industry. Not so long ago,
everybody needed an IT team to get their systems running. Today, you still
have IT teams, but they work on the bigger problems. The simple ones, offthe-shelf, often cloud based products can now solve them.
One other thing that I can’t wait to see happening in robotics is to have a robot
manufacturer open its platform so third parties can easily develop applications
for it. When I say easily I mean as easy as developing an iphone or android
application. The robot manufacturer that will head this way will have a huge
advantage as many companies will be bringing new value to their customers.
Then the customers will want to stick with this robot manufacturer, not
because they are afraid of the costs of learning another robot, but really
because they get so much value with this robot brand. And if a robot
manufacturer becomes successful with its app store, it could also be a great
business model to generate continuous revenues.
So, who will be the companies capturing these new markets in the factories
and outside the factories?
The industrial robot manufacturers are aware of what is happening and all of
them have taken action to be a player in the new applications and markets.
All of them are looking at using their existing robots outside of the traditional
applications. Some of them are also working on new platforms.
ABB has Frida.
Adept purchased Mobile Robots some time ago and are now deploying mobile
robots in factories.
Fanuc, the largest industrial robot manufacturer, has not unveiled anything yet.
I am curious to see when they will come up with something.
Kuka has the LWR and are working on future generations. They also have
mobile platforms. So they seem to be strongly looking at the emerging
Yaskawa has been working on service robot prototypes such as the one
shown here. They are also partnering with a company to bring to market a
Or will it be other new kids on the block? These guys came out of the gate and
are starting to fill the gap. But looking at the big picture, we are still talking
small numbers, a few thousand, compared to the humongous opportunity. So
they could be outpaced if they don’t fill the space before the real big guys hit
Talking about the real big guys, we know that many big software companies
are interested in robotics. The latest news is coming from Google who was in a
Christmas robotics companies shopping spree at the end of 2013 and who put
together an all-star team of roboticists. If that was not enough, they are
working with Foxconn, the largest electronics contract manufacturer in the
world. Google brings the robotics know-how, Foxconn the manufacturing know
how and this is a market in itself. Just to give you an idea, Flextronics employs
more people than Apple, Dell, HP, Intel, Microsoft, and Sony combined.
I did a presentation on this topic 2 years ago. From the technology side, there
was not a big change. What has changed the most, is the industry, all these
new players with products on the market and these big companies actively
taking action to enter the field.
We’ve seen several examples of robots with impact in this conference, I think
this is just the tip of the iceberg and that we have many exciting years ahead
of us in the robotics industry, both inside and outside the factories.