The market leader in cobots received a great deal of attention for its new large UR 20 cobot, which will not be delivered until the second quarter of 2023, but was nevertheless an eye-catcher at the automatica trade show. And by the way: Without the UR 20, Universal Cobots would not have been able to show a new model - unlike many other competitors. Some of these were also able to score with software. The UR 20, which has been redesigned on the inside, is seen as a blueprint for the revision of the existing models that is probably due in a year or more.
The UR 20 is much faster and the number of installed parts has been reduced by 50%. This further reduces the susceptibility to faults and also the weight. The new design only made the significant increase in speed possible, according to the cobot manufacturer. There was talk of around 30% more speed at the presentation. So much innovation should actually be reflected in the number of patents. But this is apparently not the case. The Chinese manufacturer ELITE ROBOTS has been attacking the market leader like a gnat. Most recently, the Chinese compared their 200 patents with the good dozen that Universal Robots has. ELITE ROBOTS has a lot of patents, but probably no technological USP. The cobots are nevertheless very solid and inexpensive. The ones from Universal Robots are solid and thanks to UR 20 also innovative. Nevertheless, the question arises as to who deserves the honor. The answer to this question may not matter to the UR user, but it is not uninteresting for the industry. After all, the hardware is likely to be offered to other manufacturers as well.
Component manufacturers were the secret stars at automatica
The component manufacturers were well represented at automatica, but nevertheless had a thankless job. They were hardly allowed to name any references and not many visitors are to be expected. After all, the end user does not buy directly, leaving a few dozen robot manufacturers as potential customers. Nevertheless, they scored points with innovations. Nikolai Ensslen, managing partner of Synapticon, was delighted with the exceptionally large number of visitors, despite the low level of end customer loyalty. In conversation, he pointed out the new compactness of its components, which are even more affordable than older servo motors & Co. with greater performance. The motion control specialist, based in Stuttgart, is already represented in Shanghai, Bay-Area in California and Belgrade.
Sensodrive, a startup from the 5-lake region of Upper Bavaria, impressed with highly sensitive sensors. Peter Molnar from Schaeffler was kind enough to explain to me the speed, strength (stiffness), compactness and more of the new Schaeffler components. Appropriately, an interview in the 3/ 2022 issue of "robotik und Produktion". Ralf Moseberg says "With our solutions, robots should achieve higher payloads, longer service life, more reach, smaller installation space, more dynamics, more comfort and sensitivity." Actually, everything that applies to the new UR 20.
The significant improvements in the UR 20 are, I assume, due to component manufacturers - whoever they may be. As mentioned, this may not matter to the user. If my assumption of the purchase is correct, this would not necessarily speak for the innovative power of the Danes. They did not have other innovations. They just concentrate on sales and have long since had user-friendly software. The accompanying circumstances speak in favor of the theory of the purchase: The UR 20 will not be available for another year. If they had developed it themselves, Universal Robots would have had time to build up their own production. So it looks as if a supplier waited for the order and has only since then been setting up its own production. This could again speak in favor of one of the new component suppliers mentioned.
Universal Robots is socially important
The previous statements are in no way to be understood as criticism. Universal Robots has just focused on market development for the benefit of all. This also includes the much praised user software. The strong market presence has made an eco-system possible from which the competitors also profit. If every Cobot manufacturer had a roughly equal market share, there would be far fewer peripherals. Because the accessory manufacturers would have always had to consider many interfaces etc. without knowing whether it would ever be worthwhile. So first for Universal Robots was developed. With success other manufacturers were served.
At the same time, new market participants such as Neura Robotics or Agile Robots were motivated to move into AI or service robotics, since direct competition seemed unpromising for them. (And, of course, their leaders are heavily AI-driven.) Thus, Universal Robots is ultimately of great benefit to society. Robotics is not to be underestimated for society in general: Right after the pandemic began, gripper manufacturer Schmalz was declared systemically important. Because without its grippers, many foods could not be packaged.
Reply from an innovative supplier
Nikolai Ensslen, founder and CEO of Synapticon responded on LinkedIn a few days later:
Dear Guido Bruch, thank you for raising the issue of cobot suppliers in your recent blog entry. This has inspired me to respond to some of your aspects here and to expand the discourse intellectually.
First of all: Unfortunately I am not allowed to solve the riddle of your article either, but you can assume that Synapticon supports (almost) everywhere where new cobots are currently launched and existing products now offer higher performance (i.e. payload, dynamics, precision) and real cobot safety based on certified Safe Motion features. The SOMANET Circulo series is the only offering on the component market that can realize these capabilities. To develop and manufacture all this yourself as a robot manufacturer creates huge effort, significant risk and inhibits innovation where it really counts for robot customers: In software, usability, AI, versatility and real application function.
In this context, I want to suggest an alternative view for your comment that it is less innovative to rely on such a supplier component instead of developing it as a robot manufacturer itself: I say with conviction that the opposite is true: manufacturers who do not waste time developing such basic components have much more capacity and focus for truly relevant innovation at their product level.
No Cobot customer gets up in the morning and thinks: I'll buy this robot because it has such great drives inside. He's not really interested in what's inside the robot. However, the customer has bought a robot for the first and last time if it does not perform as desired or gives up the ghost early - an extremely critical path, especially as a new manufacturer without a historical name. And the potential for a suboptimal result in this respect is very high if you as a manufacturer assign a comparatively small development and production team for e.g. drive controllers, especially in a crazy time like currently (chip crisis). In addition to the constant risk that employees with critical knowledge will eventually be lost, in such a situation there is also a challenge in strategic sourcing that is impossible for such a team to overcome. The EMS service providers are not in a position to take this part off your hands, the margins are too small and the processes too optimized.
Components developed in-house also only mature on their own product and not in hundreds of applications, as is the case with supplier components. And if I as a roboticist want to use the in-house component, I need the component developer(s) on my lap, so to speak, because there is no user-friendly software and no comprehensive operating instructions as with the supplier component. A Synapticon SOMANET controller is as easy to configure as a smartphone, and our documentation even goes well beyond using the product, helping our customers' robotics engineers design and tune a perfect motion control system. With the purchase of the component, you get hypersenior experience and expertise on the team, so to speak - at least at Synapticon, because we actively support our customers with our cumulative experience. We used to develop entire robots ourselves, only later focusing purely on integrated drive technology, and have now supported more than a hundred customers in the development of their robots.
On the subject of development time: You write that it takes longer to get a saleable product with suppliers on board. Have you considered that the component is a mature product in series production? Robot manufacturers can access it overnight, integration takes a fraction of the process of planning and designing an in-house development, the robot is running at full performance within a few days. With a self-developed component, which is still very young and full of errors and unreliability, the commissioning process is often extended by many months and sometimes becomes detective work in order to assign problems to a part of the system. At deadline times, this becomes a test of endurance for the team. There is no question that relying on (good) supplier components accelerates the time-to-market and saves nerves.
For servo drives, for example, you must also bear in mind that it takes at least 5 years (exceptions really excluded 😉 ) to develop and industrialize a powerful drive controller. Integrated encoders, brakes and safe motion not included. This is regularly mercilessly underestimated by some companies and even engineering firms offering themselves. The number of developers is even quite insignificant in this point, you can have too few, but additional people hardly accelerate anything. In our industry (motion control), there is even the opinion among veterans that a drive company needs about 10 years to develop servo drives from scratch and bring them to maturity, which meet the customer and market demands.
Incidentally, buying in a mature component should not be confused with commissioning an external service provider: Here, the disadvantages and the risk are even greater: Hardly anyone in the company understands the externally developed part, and if so, then only superficially. Development service providers also have the general problem that they have to be generalists by principle. This means that there are no service providers who really understand motion control or servo drives in depth. The software stack of an industrial servo drive alone is incredibly extensive and is not available for purchase as a library. Service providers and their customers usually do not understand the difference between the motor control libraries of the chip manufacturers and what is needed for an industrial servo. Accordingly, development projects regularly fail fatally compared to initial expectations. At Synapticon we have many customers who have already failed with in-house development, external development, and/or with the attempt to integrate generic components (which are not designed and optimized for cobots).
The only potential argument against supplier components at our level is cost. Manufacturers like to gloss over the costs, just like the associated internal or external development projects: As a rule, only the material costs are considered, which are of course much lower than the purchase component. When development and certification costs and the overhead involved in industrialization as well as ongoing maintenance are taken into account, they are usually mercilessly underestimated. If this calculation is complete and correct, it shows that an in-house development starts to pay off, if at all, only from a few 100t units, even if the production costs are only 20-50% of the price of the supplier component. The whole described risks and disadvantages that one has with the self-development, however, remain eternal.
In this document we go into some more aspects of the make or buy question for cobots.
Mr. Bruch, I would be very pleased if you would continue to address the topic of suppliers in robotics or cobotics from time to time in the future. These are similarly critical and interesting here as is the case in the automotive industry. Both products are similarly complex, even if the car seems more complex (mechanically) at first glance.
Many greetings, Nikolai Ensslen
In my own account/advertisement
The author of this blog is significantly involved in the AI/robotics project Opdra. He advises on almost all issues related to robotics incl funding/subsidies, but does not go in-depth into the technology. More about him can be found here.