Many of us have seen impressive videos of humanoid robots that walk, talk and even seem to think like humans. But can they do something and replace human workers? Can they run faster than even hounds or jump higher than cougars?
Robotics engineers have worked for decades and invested millions of dollars in research trying to create a robot that can walk or run like an animal. Yet many animals are capable of feats that would be impossible for robots today.
Animals are much better at running than robots. The difference in performance arises in the important dimensions of flexibility, reach and strength.
“The wildebeest – an African member of the antelope family that looks like a cow – can migrate thousands of kilometers over rugged terrain; cockroaches can lose their legs but still run fast; and a mountain goat can climb a rock,” said Prof. Max Donelan of the Department of Biomedical Physiology and Kinesiology at Simon Fraser University in British Columbia. “We have no robots capable of such a thing.”
He and colleagues from the University of Washington, the University of Colorado at Boulder and the Georgia Institute of Technology just published a study in the journal Scientific robotics titled “Why Animals Can Outrun Robots”.
To answer why and how robots lag behind animals, they examined various aspects of working robots and compared them to their animal equivalents for a paper published in Science Robotics. The paper finds that by the metrics used by engineers, biological components perform surprisingly poorly compared to manufactured parts. Where animals excel, however, is their integration and control over these components.
To understand how they can move like animals
Researchers study one of five different “subsystems” that combine to create a working robot—power, frame, actuation, sensing, and control—and compare them to their biological equivalents. Until now, it was generally accepted that the superiority of animals over robots was due to the superiority of biological components.
“The way things turned out was that, with few exceptions, the engineered subsystems outperformed the biological equivalents—and sometimes radically outperformed them,” the authors wrote. But also, what’s very, very clear is that if you compare animals to robots at a system-wide level, in terms of locomotion, animals are amazing – and robots are yet to catch up.
Optimistically, for the field of robotics, the researchers note that if you compare the relatively short time robotics has had to develop its technology to the countless generations of animals that have evolved over many millions of years, progress is remarkably fast.
“It will move faster because evolution is undirected,” they added. “Although we can easily adjust the way we design robots and learn something in one robot and download it to any other robot, biology doesn’t have that option o there are ways we can move much faster when we design robots than we can by evolution – but evolution has a huge head start.”
More than just an engineering challenge, practical working robots offer countless potential applications. Whether solving last-mile delivery challenges in a human-made world that is often difficult for wheeled robots to navigate, conducting searches in hazardous environments or working with hazardous materials, the technology has many potential applications.
The researchers hope their study will help guide future developments in robot technology, with an emphasis not on building better hardware, but on understanding how to integrate and control existing hardware. Donelan concludes that “as engineering learns the principles of integration from biology, working robots will become as efficient, flexible and robust as their biological counterparts.”
