Universal Robots was founded in 2005. Originally, the first startup specialized in cobots, collaborative robots, didn’t know it, but it would pave the way for a new market. Today, all the biggest industrial robot brands also have their range of cobots. In the meantime, UR has become a company with 1,000 employees and remains focused on its core business, the 6-axis cobots. They have installed 75,000 of them worldwide. We met with Jocelyn Peynet, Country Manager of Universal Robots France, to look back on these 15 years of cobotics and what makes the strength of these gray and blue cobots made in Denmark.
Hicham Dhouibi contributed to this interview.
They are increasingly noticeable in advertisements, movies, or on television sets. Universal Robots’ collaborative robots, recognizable by their gray and sky-blue colors, have found their place in industry.
They burst onto a market that didn’t exist 20 years ago. Since then, the robot capable of working safely with humans, without any cage, has become almost standard technology on production lines of large automobile companies, in factories, SMEs, as well as in smaller workshops where a third arm is welcome. All robot brands, in Europe as well as in China, now have their collaborative version.
6 Axes Forever
In 2005, UR launched its first six-axis cobot, the UR5, with a 5-kilo payload. Since then, UR cobots have increased their payload capacity, to 5 kilos, then 10, 20, up to today’s UR30 for a 30 kg payload. But the brand has never deviated from one rule: always make 6-axis robots, whereas other manufacturers like KUKA, for example, offer 7-axis cobots.
Today, UR cobots, made in Denmark, have a lower price than that of major competitors like Staübli or Fanuc. The starting price for the smallest cobot, the UR3, is between €20,000 and €25,000. For a UR30, it’s below €50,00.
50% of their robots are installed in Europe, 30% in the Americas and 20% in Asia.
Among their major clients, UR now counts Stellantis, BMW, and Safran, as well as luxury brands like Dior, LVMH, and Louis Vuitton.
They have just redesigned their ranges of cobots intended for palletizing and machine-tending, the UR20 and UR30.
We spoke with Jocelyn Peynet, Country Manager of Universal Robots France, to discuss Universal Robots’ very specific positioning, their new cobots, and the ongoing and upcoming innovations.
You’ve redesigned your collaborative robots. They used to be more like tubes, and now, their shape is more aesthetic. Tell us about this new generation of UR cobots.
Jocelyn Peynet: “We started from scratch. Our robots are bigger now. And internally, we’ve completely revamped everything. We changed the motors, reducers, and joints. The new joints are more efficient in terms of speed and torque. They allow our cobots to achieve faster speeds. We’ve already installed 75,000 cobots worldwide. So, we’ve received a lot of feedback from customers with specific needs. They asked for higher speeds, especially in the automotive sector, so we completely reworked the architecture of our robots to make them faster. This resulted in a new generation of cobots like the UR 20 and UR 30. Eventually, even our smallest cobot sizes will evolve. In terms of repeatability, we’re still in the same range, around five-hundredths of a millimeter. We’re still focused on high precision.”
Tell us about the UR30. What type of tasks is it designed for?
Jocelyn Peynet: “The UR30 has a reach of 1300 millimeters and a payload of 30 kilograms. It is particularly suitable for machine tending, i.e., loading/unloading machines. For example, in a machine cycle, the cobot places the raw material in the machine tool, the milling machine or lathe does its job, puts another raw material in, and then gets the finished product. It is also suitable for palletizing. For example, it can load 25-kilogram bags using a suction cup system from a third-party brand, Schmalz.”
Regarding programming, how does it work?
Jocelyn Peynet: “It’s no code. We already have predefined functions for movement, and waypoints. There’s no need to enter lines of code; we enter the movement function, and the robot knows we’re going to indicate a position for it to move to. Then we set the waypoint, just disengage the robot, and move it from point to point. It’s super easy to determine the position. And then we have advanced functions to create subprograms, allowing all roboticists who want to do full programming to customize.
Natively, there are also models. Palletizing, for example, is a model. The movements are already predefined, like a sort of pre-routine. There’s no need to write the program code. It’s already done. You just need to teach it where it will place its first box, and in which direction. And then, find the model for your pallet.”
How does it recognize the pallet?
Jocelyn Peynet: “We input the dimensions and orientation of the boxes, and then it calculates the trajectory.”
You’ve redesigned the cobots. They have increased payload capacities. They can move faster. What are the other innovations we can find in your cobots?
Jocelyn Peynet: “We’re constantly evolving, especially in the software side. What has made our robots strong is the simplicity of integration and programming. We have software that is very user-friendly for many people. But we’re also starting to develop more specialized industry-specific software. We have a version of the software called PolyScope X, which will be available by the end of the year, for machine loading, which is a separate profession. What we continue to do every day is to simplify the integration of robots, simplify their use in many areas like automotive, palletizing, and polishing, to make the robot even simpler for integrators, for specific professions.
You recently partnered with Nvidia on the AI aspect. Can you describe this collaboration and what it could bring?
Jocelyn Peynet: “Our robots have always been open and accessible to developers. We encourage algorithm developers to come to our platform for specific use cases. For example, in quality control, inspecting pans is very difficult for the naked eye. When an operator has been inspecting pans all day, they might miss one or two. Our partners have created vision systems with artificial intelligence algorithms that easily attach to our robots, allowing for these inspections.
What we’re announcing with Nvidia is also about algorithmically controlling the robot’s trajectory. For example, determining the most optimized trajectory for time-saving, by avoiding obstacles and going as fast as possible. AI can help us, especially without having to think and program it ourselves. We’re talking about a potential gain in trajectory optimization of the order of a 1 to 80 ratio. Especially when we’re bin picking, the parts are loose in a bin, there’s a 3D system above that says the part should be picked up like this and like this. And depending on how I tell the robot to pick up the part, we can save a lot of time by orienting the robot correctly. There’s a piece here, the robot is here, it computes and finds the optimized trajectory in terms of acceleration, speed, in terms of the possibility of orienting the robot to avoid wasting time on something unnecessary.”
Has this need also emerged from the automotive industry?
Jocelyn Peynet: “The need for bin picking and the need for high speed are closely linked to the automotive industry because what they want is to go as fast as possible to manufacture parts. What they’re looking for is cost savings and productivity gains. Solutions that locate parts and pick them with a basic trajectory, exist everywhere. However, the trajectory that will save them 80% of cycle time, that’s what we’re increasingly looking for.
Nvidia already has a product that’s somewhat packaged to optimize trajectories. MiR, our sister company, has released a new AMR and they used AI algorithms precisely to optimize trajectories with Nvidia.”
Are you exploring mobile robotics, or putting a robot arm on an AMR?
Jocelyn Peynet: “Yes, indeed, for machine loading. For example, in electronics, we can envision a UR robot arm mounted on a MiR AMR to pick up wafer cassettes, for instance, and load them into a machine. There are several machine tools in Switzerland, particularly in luxury or watchmaking, where a cobot allows for loading and unloading the machine.
Another application is to have one mobile robot instead of three or four stationary ones. If machine cycle times are 3 to 5 minutes, we don’t need to have a robot constantly stationed in front of the machine. Therefore, instead of deploying several cobots in the workshop, there’s one cobot mounted on a mobile robot capable of picking up pieces and transporting them elsewhere in the workshop, rather than just unloading them next to it. This solution is available in our catalog.
Another example is order picking in the automotive industry. Mobile robots equipped with a robotic arm can gather various parts from different bins, place them in a large bin, and deliver them to the operator or a workstation.”
Universal Robots has been around since 2005. Nowadays, there’s quite a bit of competition from traditional robot manufacturers. Even machine tool manufacturers offer cobotic solutions. So, what still sets Universal Robots apart? Is it the open approach with developers? The simplicity? The fact that you stick to your core business of six axes?
Jocelyn Peynet: “There are already answers in what you’re saying. Firstly, we believed in this from the beginning. Those who are just now entering the cobotics market used to say it was slow, useless, and unsafe. But there’s clearly a market. We created a market that didn’t exist before with SMEs, and micro-businesses, who had never considered putting a robot in their workshop.
Simplicity, the fact that I can take the robot by the hand, and teach it the welder’s gesture, that didn’t exist before. So, we made that accessible.
Initially, we weren’t roboticists. However, we created an ecosystem and developed truly diverse applications to meet all needs.
We’re not going to make cobots to lift cars. We stick to human-sized robots. We were asked to go a bit further in terms of payload. We introduced the UR20 and UR30, and it was fully justified.
Our cobots, for example, those with 20 kilograms of payload, or even 30 kilograms, weigh 65 kilograms. Some companies make cobots for a 30-kilogram payload that weigh 150 kilograms. So, there’s still engineering involved. We have 1000 people at UR, and we only make cobots.
What still makes the difference is also the ecosystem of gripper products. We were the first to work with all vision and gripping suppliers. We created a shelf of tools that our customers can use for do-it-yourself projects. Today we have 500 products including magnetic grippers, finger grippers, vision tools and software. They’re also integrated into our programming software. Meaning, the cobot can be super easy to use, but if I spend two days programming a gripper, I’ve lost a lot.”
So, you already have integrated libraries?
Jocelyn Peynet: “Exactly. So, we continue to nourish our ecosystem. We know we’re experts in arms. The rest, we don’t know much about. So we have partners. Other manufacturers don’t have this ecosystem that we brought along.”
Asian companies are entering the market. Also, former automation companies start making robots, like Schneider, and Beckhoff. Others like igus make low-cost cobots using plastic. How do you view this competition?
Jocelyn Peynet: “We see this very positively, especially the big names that are now creating cobots because it means that we didn’t get it wrong twenty years ago. It legitimizes us because, in the beginning, cobots were only Universal Robots, so it may sometimes be seen as anecdotal. Now, indeed, it’s everywhere. So, it means that cobots have an interest; it’s industrial. And what suits me well is that people searching for robots on Google find us first. Today, when we talk about cobots, we talk about UR.
As there are indeed more and more offers, we see that decision times can take a little longer. The clients will form their own opinion. They will test. But they very often come back to us.”
Some manufacturers develop robots dedicated to a specific domain or sector. For example, a robot for the pharmaceutical industry, a robot for clean rooms, different from the one for pharmaceuticals with coatings, screws, and materials dedicated to the sector. Is this something that interests you?
Jocelyn Peynet: “The same cobot that is used at Renault for screwdriving parts of engines is also deployed in a small company that packages chocolate into boxes. We aimed for simplicity and standardization. We maintain consistency by using the same types of joints across robots of different sizes, enabling us to meet delivery times of just a few weeks. Those who offer customized solutions for traditional robots often have lead times extending into months. So, we want to maintain this flexibility with a truly standard solution. Then we adapt the tools, we adapt the robot’s environment. For example, we can put a sock around the robot for more protection. It’s true that, lately, we have been asked more and more for cobots in pharma. So, we are working on a hygienic version. However, this version is not always necessary; sometimes, a simple adaptation like adding a protective covering suffices.”
And what about the chemical industry? This sector often requires robots to have covers or screws made from special materials that are suitable for use in chemical environments.
Jocelyn Peynet: “For now, it would require a lot of effort in R&D for a market that is not yet sufficiently addressed. However, we are evaluating pharma and medical. We have medical machine manufacturers who have already customized the cobot themselves. They changed the covers, and the seals to meet a particular need. These are more like partners since our system is open and quite simple to modify.
Moreover, for explosive environments (painting applications for example), we have partners who are providing services to modify our robots and get ATEX certification.
For our new generation of robots, we have changed the protection index. So far, we were IP54; now we are IP65. That’s why we redesigned the robot from A to Z because bringing these evolutions to a generation that’s already four or five years old would require too much effort. We preferred to start with a new base.”