Robots to go

When Marco Hutter initially applied for a Branco Weiss Fellowship, he proposed to develop a new type of arthropod-like leg as a crucial element for a future autonomous robot. The idea had arisen out of curiosity. What if you tried to develop an autonomous system that could walk like a dog? Combining cutting-edge research into hardware design and machine intelligence would enable the robot to perform complicated tasks and climb over exceedingly difficult terrain. Marco’s plan did not just work out – it skyrocketed. Once his approach became feasible, the technology gained a momentum that continues to this day – taking it out of the research lab and into the world outside.

During the past ten years a number of technological leaps have accelerated the development of autonomous systems. There have been massive improvements in the performance of motor/battery and machine learning/sensor technology, while system costs have fallen significantly. For example, the first lidar sensors for perceiving and mapping a robot’s surroundings were the size of a basketball, with the price tag of a luxury car. Today more precise systems the size of a tennis ball or smaller are available at a fraction of the cost. Optimization problems that a few years ago required large computers and had to be laboriously calculated offline can now be solved in milliseconds online and on board the robot. As these robots become powerful and useful for diverse applications, a substantial number of teams developing this kind of autonomous system have emerged around the world.

Without an insight into nature’s solutions, however, the engineers would be operating in the dark. The animal world ultimately shows what’s possible when dealing with the physical limits that exist on earth. Animals are therefore a great source for inspiration as to the direction where developments could go. They achieve impressive mobility and versatility thanks to a unique combination of brain power and embodied intelligence.

The interplay between engineering and the animal research have led to fruitful collaboration. Marco’s team, for example, cooperates with other institutes in the field of biomechanics. There are conferences where specialists in biomechanics and biology meet with robotic or machine learning experts to seek new paths in the design and control of novel intelligent mobile systems. This interdisciplinary exchange inspires new robots and in return supports a better understanding of the biological counterparts.

Most research activities at universities end with demonstrations of the technology in laboratory environments. What is often lacking is the engineering effort to robustify this technology and deploy robots in the real world. From the beginning of his research career, it was Marco’s goal to realize solutions that can be deployed in the most challenging scenarios. In 2016, he established ANYbotics, a company that now employs more than 50 people to commercialize legged robots. There is no lack of dull, dirty, and dangerous tasks people are happy to delegate to machines. Robots work independently in a range of zones which humans would not want to enter. They spend hours collecting data in nuclear facilities, explore territories in mines more than a thousand meters underground, and work diligently on offshore oil and gas plants. They can autonomously explore unknown areas, produce maps with an accuracy of centimeters over several kilometers of terrain, and find or inspect objects. Soon they will be helping disaster relief forces carry out search and rescue tasks, for example in buildings at risk of collapsing after an earthquake.

Marco has always remained sensitive to the ethical aspects of the development of autonomous systems. For example, in a collaboration with Armasuisse and the University of Zurich he contributed to a dossier on this subject, and he partakes in various panel discussions on the opportunities and fears of robots when taking over jobs from humans or being used in areas ranging from security to geriatric care. The outstanding capabilities of the machines that are currently being developed should only be used where they provide the best possible service for people in need.

These days large IT companies are investing huge sums in this field, for example to further automate logistics and to enable autonomous driving. When marketable solutions become available, they will change the way we work and live. Marco sees many more possibilities for applications in the construction industry and agriculture – to make them more efficient, safe, and sustainable. Just as medicine becomes increasingly personalized, robots may soon be looking after individual plants, trees and animals.