A team in Munich is busy perfecting two identical self-learning robots resembling human arms able to accomplish increasingly complex tasks
How Siemens is pushing forward with next-gen robotics
You will not see robots replacing the touch of a human hand to execute complex tasks any time soon.
The dexterity and the problem-solving ability of the human mind controlling that hand means we will probably always have an edge over robots, which, so far, are only good at repetitive, programmed jobs.
The German technology giant Siemens, however, is determined to change that. A team of experts in Munich are busy perfecting two identical self-learning robots resembling human arms and able to execute increasingly complex tasks. The programme is laying the foundations for a bigger role for robots in the industrial sector in the future, the company believes.
Based on the ground floor of an unassuming building nestled amid an equally unassuming cluster of old structures at Siemens’s Perlach offices in Munich – where thousands of researchers are housed – is the company’s robotics and digital machine laboratory. It is a workspace and playground for some of the finest minds in the robotics industry and experts creating digital twins through augmented and virtual reality. The results could have applications ranging from “remote” troubleshooting on oil rigs to creating computer models to build better and more efficient spaces, be it a multibillion-dollar car manufacturing plant or a central transport hub dealing with the flow of tens of thousands of people on a daily basis.
Georg von Wichert, who heads the robotics research group, autonomous systems and controls at Siemens, is in charge of the robotics lab. Being a passionate roboticist, he has followed machine learning and artificial intelligence for years.
For a group of journalists visiting from the Middle East, Mr von Wichert was keen to display the capabilities of the twin robots, which one of his colleagues jokingly said have very complex names: “the right arm and the left arm”.
The robots, equipped with cameras, carry out complex motions as they observe, learn and co-operate to finish tasks such as locating electrical equipment and placing it correctly in a precise position on a rail in the centre of a table.
“All these motions that you see, the robots are not programmed to do that,” Mr von Wichert says. “What the system gets is a description of what needs to be achieved”.
The robots at the Siemens lab understand that they cannot finish some of the tasks on their own. To complete them, they pass electrical equipment pieces, in this case between each other to determine how to fit them in a specific space. These machines have an understanding of themselves and their capabilities; if a part is removed from the task they are undertaking, for instance, they co-operate to locate it and finish the job.
“We are very proud of what we have developed here. You don’t see that even in the big universities,” Mr von Wichert says. “It’s research that is going to take more time, I don’t know how much ... we really see this revitalising the industrial sector.”
At present robots perform repetitive tasks, for example at a car mass-production line. Programming is at the heart, or should that be the head, of an industrial robot. If the same machine is required to do another job, it has to be reprogrammed from scratch, which is no small task. This is among the reasons companies like Siemens are rethinking automation and developing self-learning robots.
What Mr von Wichert and his colleagues are doing in their lab is setting the foundation for an increased role for self-learning robots in the industrial sector.
If today they are only able to weld or paint with precision the chassis and body panels of cars on a production line, they will be able to do more complex jobs higher up in the line in the future, such as engine building or seat installation. But they perhaps will never be able take the human out of the equation completely.
Although we are more prone to making mistakes, our evolution has given us extreme dexterity and flexibility, plus the ability to solve complex problems and anticipate potential ones before they arise – even imagine and implement more efficient ways of doing things – and some jobs will always remain dependent on the human touch.
“It will never be able to completely replace humans,” Mr von Wichert says.
While one side of the Siemens lab is dedicated to the robotics, the other side is the den of number-crunching mathematicians, creating “digital twins” of other things, be it an oil rig, a propeller drive system of a cruise ship, a turbine, a public water distribution network or Frankfurt’s main train station.
These digital twins help to create simulations and models for better space building both in the public and private sector using augmented and virtual reality to create platforms for remote and effective troubleshooting.
Wolfram Klein, from Siemens Corporate Technology, heads a team of experts and PhD students who dedicate their time to creating these simulations, developing software that allows engineers to “virtually” get inside a gas turbine or a ship’s propeller system, locate a problem, find solutions and strategies, discuss it with mechanics sitting thousands of miles away and exchange notes – all by strapping on a virtual reality headset.
Mr Klein’s team can also analyse how to run a turbine at its optimum performance level through augmented reality gear, diagnosing the situation via a heat map of the machine.
Siemens, which ranks among the 10 largest software companies globally, has been aggressively investing in research and development to strengthen its ability to leverage data analytics amid the global economy’s pivot towards digitisation, automation and electrification.
The company’s 2018 R&D budget is already an 8 per cent year-on-year increase from the €5.2 billion (Dh22.31bn) spent a year earlier. About €1.2bn of the total is invested in its digital business. Since 2014, the company has increased R&D spending by 40 per cent.
In terms of core technologies, Siemens plans to increase spending above €500 million next year and will actively continue to look for acquisitions of software businesses, Roland Busch, chief technology officer of Siemens told The National in Munich.
Michael May, the head of the company’s core technology data analytics, artificial intelligence and corporate technology, says the company’s “objective is to have AI helping each and every piece of equipment”.
With the kind of spending that Siemens does on R&D, which might very well result in a further boost in coming years, Siemens is perhaps closer to getting its objectives sooner rather than later.
“This is our dream,” says Mr May.