Roland Siegwart Autonomous Systems Lab
Text: editorial staff “ceo” Magazine/Images: Markus Bertschi/Video: Severin Jakob
Autonomous robots are wending their way into our daily life and changing it in lasting ways. At the fore in the development of these new technologies is the Autonomous Systems Lab at the ETH Zurich.
Roland Siegwart, who heads the laboratory, is convinced that – despite the progress being made – it will still take a number of years before cars can navigate their way through cities all on their own. This is because up to now, robots have had a rough map of the world in their heads, but don’t exactly know how to read it yet.
“The robots need a plan, since they don’t understand our world.”
Roland Siegwart whips out his smartphone and waves it through the air a couple of times. “We’re currently working on a project that will make it possible to do this – you see, just a few swings of your arm – and digitally map an entire room,” explains the professor and head of the Autonomous Systems Lab at Zurich’s Federal Institute of Technology (ETH). We stand in a corridor of the research centre. On one wall hangs an illustration that describes the technology to a tee. “The key is our new sensor. It’s the further evolution of the kind of receptors that are already built in to today’s electronic devices and enable the display to pan along with the direction the device is pointed in,” says Siegwart. In future, the interior of an entire home could be digitally mapped this way. The ETH is working closely with US tech giant Google in this effort.
The so-called Tango Project is just one of many that are being pursued in the ETH lab. Students tinker on new autonomous machines which in the not too distant future will be deployed on land, sea or in the air. Here, new ideas are taking form – from beetle-like robots, to solar aircraft, to droning drones. One room looks like a workshop, another is outfitted with wrestling mats, safety nets and cameras in order to flight test, but not destroy, the self-thinking drones. “Our research realm here is fabulous – it combines mechanical engineering with electrotechnology, informatics, sensor technology, artificial intelligence, and of course man,” gushes the robotics professor.
“My daydream is to see Switzerland become the first country on the planet where autonomous vehicles actually take over certain tasks.”
The overarching objective of the work being done at the Autonomous Systems Lab: to create devices which take the burden off people wherever a job could be physically taxing or damaging to their health, or those that can otherwise be of greatest benefit to the general populace. Siegwart points up examples, such as their applicability in the underground mining of raw materials, in deep-freeze warehouses and even in the agriculture area. “One of the most pressing problems mankind faces is the food supply.” Here, Siegwart can envision a combination of autonomous drones and agricultural robots as means to achieving a marked increase in farm output. The idea: drones overhead would monitor the soil moisture level and seed growth, as well as keep an eye out for pests, while a robot in the field mechanically destroys weeds and provides water and fertiliser as necessary. “The robots relieve the farmer of a cumbersome task and contribute to increased agricultural yield,” Siegwart summarises.
Today, the broad public experience autonomous systems mainly in the area of mobility. Automobiles already have an array of pre-installed support systems that make driving easier: there are cars that can steer themselves into tight parking slots or even stay on cruise control in the same lane while travelling on the motorway. “Autonomous vehicles will fundamentally change the way we go from point A to point B,” says Siegwart. And this is not just a matter of comfort and convenience – the environment will also benefit, as the technology enables vehicles as well as the existing infra-structure to be utilised more efficiently. Not to mention the older generation, who can remain mobile for years to come. “My parents would be pleased as punch if they could use an autonomous vehicle for visits to their grandchildren,” muses the 58-year-old.
But it will take a number of years before the first autonomous automobiles come into daily use. Although those cars have no problem getting around already today in a structured and not all too complex setting, “the robots need a plan, since they don’t understand our world,” explains Siegwart. And precisely that is a prerequisite if one day they are to master the ins and outs of a jammed, multi-lane traffic crossing. In such instances, interaction is called for – it takes eye contact and gestures or signals. Moreover, the systems must be able to recognise unknown objects rapidly and figure out how to deal with them. “In this regard, man is clearly superior to machine,” admits Siegwart.
Other hurdles for self-driving cars in their fast lane to everyday life take the form of the hefty costs of sensors and cameras for navigation, as well as the roadway rules that have yet to be written. “Who should matriculate these vehicles? Who knows how they react in a difficult situation? Here, regulations are necessary,” observes Siegwart. That applies in particular to the case of an unavoidable accident where a system has to make a split-second decision on who or what will be impacted by the collision. Siegwart is confident that the technology will help people even in such instances. “Like airbag systems of today, it will ensure that passengers walk away with as few injuries as possible, if any at all.”
Equally spoken, the robotics field is strewn with questions concerning the ethical aspects. Even the Vatican has taken part in workshops where the limits of the technology were discussed. From a purely scientific standpoint, Siegwart is of the opinion that there are no limits, at least in terms of fundamental research. In fact already today applications are being tested that seem like they came from some science fiction novel: cameras that transmit images to the brain and replace the eye function; or neural impulses that steer autonomous systems.
For companies as well as the overall economy, autonomous systems are a decisive success factor. Yet it need not be that they operate totally on their own. “Especially in Switzerland, which is a global leader in the machine tool industry, these systems are enormously important due to the automation wave that is under way,” says Siegwart. Ever more frequently, robots and humans are working side-by-side on the factory floors.
But in some cases, the machines are in fact supplanting the workers entirely. As was the case in the industrial age, increasing automation leads to a fundamental change in the labour world. There will be losers, of course – but not from one day to the next.
“Robotics is a very complex technology. So its evolution will take place very gradually, thereby giving the affected workers enough time to react to the new circumstances,” Siegwart posits. In any case, he is convinced that autonomous systems will be a win-win for Switzerland and generate many more new jobs than those that go by the wayside. It therefore should come as no surprise that a lively transfer of knowledge is taking place with businesses, a trend that will ultimately benefit the entire country. Siegwart maintains that, already today, Switzerland is one of the three leading robotics nexuses in the world. “My daydream is to see Switzerland become the first country on the planet where autonomous vehicles actually take over certain tasks.”
Roland Siegwart (58) studied mechanical engineering at the ETH in Zurich and earned his doctorate in 1989. Siegwart is director of the Autonomous Systems Lab (ASL) at the ETH’s Institute for Robotics and Intelligent Systems (IRIS). He has been a board member of the NZZ Media Group since 2016 and lives with his family in Schwyz.
The Zurich ETH’s Autonomous Systems Lab (ASL) is a global leader in the research and development of autonomous robots, including drones, aircraft and beetle- or human-like “droids” that can independently carry out complex tasks. The laboratory was founded in 1996 at the EPFL in Lausanne and since 2006 has been part of the Institute of Robotics and Intelligent Systems (IRIS) at the ETH where, almost a decade ago, the very first autonomous drones took flight.