The digital world also includes time-consuming programming. Although many things can now be “taught” or “parameterised”, can ‘t it besimpler? Research is being carried out at the Bern University of Applied Sciences (BFH) and the results are astonishing.

More and more robots are working in the production halls of this world. In 2020, a total of 384,000 new robots were delivered worldwide, which means that there are probably more than three million robots in use, according to the “World Robotics 2021” report by the International Federation of Robotics (IFR). It seems beyond question that conventional industrial robots make up the majority. But the fact that cobots still only have a marginal share of the market is astonishing. Of the 384,000 robots sold, only 22,000 were cobots – that is just under six per cent.

Prof. Dr. Sarah Dégallier Rochat in the BFH laboratory in Biel. Picture: Eugen Albisser

Professor Sarah Rochat of the Bern University of Applied Sciences is not particularly surprised. “With the cobots came a democratisation of robots; but then again, cobots are not as easy to use as one would like.”

A survey conducted by the company Reichelt came to a similar conclusion. Although cobots are indeed on the rise, one of the biggest difficulties for the introduction of robots is that the installation is associated with a high level of effort. For a quarter of the respondents, however, it is critical that when selecting robots, attention is paid to the fact that they can be reprogrammed easily and quickly (see also Box: The cobots are coming!).

Cobots must become even more flexible

The purpose of using collaborative robots is mainly to work alongside humans in manufacturing and industrial environments. Cobots and humans should work together seamlessly. Unlike an industrial robot, this requires that it can also be adapted to new tasks more quickly. Good groundwork has certainly been done by cobot manufacturers. Setting up cobots can often be achieved through a teach function, which is already a considerable simplification. “For simple processes, this can actually work quickly. But as soon as a little adaptivity is necessary, i.e. the situation becomes more complex, then it no longer works,” says Sarah Rochat. And then, first of all, a lot of time is lost and you would have to have a robot expert or an integrator on site. But especially nowadays, when high-mix and small series are on the increase, flexibility is indispensable and companies would also like to hand over more responsibility to robot laymen.

Anticipating and dealing with probabilities

The vision is therefore that a robot cell should be so simply constructed that this adaptivity is achieved more quickly. Cobots should therefore no longer be laboriously programmed. Basically, cobots would have to anticipate many things and would also have to be able to robustly deal with inaccuracies of the sensors or the model. And first and foremost, they should be able to be programmed in the simplest possible way, whereby it should not matter whether there is a Fanuc robot in the cell or a Universal Robots, ABB, Bosch Rexroth or Stäubli brand cobot. And other integrated systems should not have any interface problems either.

This is precisely what is being worked on in the laboratories of the Bern University of Applied Sciences. The demonstrators already set up in the laboratories are each equipped with a robot (Fanuc, Universal Robots, Mecademic) and are equipped with intelligent image processing systems. These also model the workspace on the basis of a priori information such as CAD files or process descriptions.

“But it is also important that this digital twin that we map with this modelling can react to information in real time,” says Sarah Rochat. That means a user should always see immediately what the robot is going to do. Nowadays, that is usually still not the case. If an error happens, you can’t see what the problem might be. An example: the robot does not recognise that two objects are simply too close to each other; it perceives it as a single object. This object is now displayed and the user immediately sees on the screen that the robot only recognises one object. He can now change the illumination of the vision system or adjust the parameters.

From passive to active user

It still sounds as if a touch of expertise is needed, despite everything. “That’s so,” says Sarah Rochat, “in order to exploit the full potential, you need expertise, of course. That’s why we create simple tutorials and this is where you can always go deeper and build that expertise.” But the barrier to entry is low, she says, and you can deepen your knowledge continuously and in the simplest way. Rochat: “There is also a fundamental problem behind this that we want to address. The operator is normally rather passive. If an error happens, he doesn’t solve it because he thinks he can’t. Or because the programming interfaces are so complicated. Or because the programming interfaces are made in such a way that the operator can intervene as little as possible. That’s how you want to prevent errors. Our goal is to give the users more confidence, to make them more secure in dealing with robots. Through these tutorials, they get an idea of how the system works. But the minimum has to be enough to give them confidence.”

Pick Cube and place on red square

But how does simple programming work? “It works via “natural” language, i.e. how an operator would explain something to his colleague,” says Sarah Rochat. You take the appropriate element from the different levels, such as tasks, skills or device properties, and combine them. In the end, it reads like this: Pick Cube and place on red square. Repeat until pallet is full. The words in italics are the parameters inserted by the user.

Everything is already working quite smoothly on the demonstrators that have been set up. They will be tested from January 2023 onwards at a company that has kick-started this project. But it is by no means a project that will only go to this company. A spin-off called Auto-Mate Robotics (formerly “Power Up!”) is already involved, which is pushing the research results further and further maturing the robotic cell so that industry can benefit from it (see also box “Funding from the Gebert Rüf Foundation”).

Fig. 1: Screenshots of the application interface.

Start-up “Auto-Mate Robotics” (formerly “Power Up!”) receives “First Ventures” funding from the Gebert Rüf Foundation

Their idea was convincing: Lucas Renfer, Christian Wyss and Charly Blanc (in the cover photo) receive funding for their start-up Auto-Mate Robotics CHF 150,000 in funding from the Gebert Rüf Foundation. With Auto-Mate Robotics they want to develop a flexible and collaborative automation system. This is to be used in companies that manufacture products in small quantities and with many different variants. In this sector, manual labour is often still used because traditional automation solutions are not financially viable. An adaptable and easily programmable robot system has the potential to reduce production costs and at the same time expand the competences and tasks of the employees.

The cobots are coming – but there are also hurdles

Robots and smart helpers have long since ceased to be a rarity in Swiss companies. Nevertheless, the full potential for use does not yet seem to have been exhausted. Reichelt Elektronik surveyed more than 1500 industrial companies – 200 of them from Switzerland. A full 81 percent agreed that the use of robots helps to increase competitiveness. Almost 52 percent of the companies named production and manufacturing as their field of activity. Concrete tasks include: assembling or loading machines, packing tasks, welding and also palletising, stacking and transporting.

Although robots are becoming increasingly popular, one of the biggest difficulties in introducing robots is that they are expensive to install (47%). A similarly large hurdle (45 %) is maintenance and servicing.

For the implementation of robots to be successful, Swiss companies mainly want flexibility from the smart machines. 73 per cent consider it important that the robot is versatile. Likewise, more than a quarter (27 %) of the respondents consider it critical for the success of the selection of robots that they can be easily and quickly (re)programmed.

This article was published in the trade magazine for industry “Technik & Wissen”, text and images by Eugen Albisser.