Scientists have long sought to build robots that can run, jump and climb as well as animals do.

In a study published this week in the academic journal Science Robotics, a team of Canada- and U.S.-based researchers explored why robots simply aren’t there yet—and offered new insight into how we can close the gap.

Just act natural: Robot designs have made massive strides in recent years. Engineering firm Boston Dynamics—which prompted public backlash years ago with the release of its infamous robot dog—has developed a new two-legged robot that can perform parkour. 

But machines have a long way to go before their movement matches the effortless dexterity of the natural world. Consider how mountain goats climb impossibly steep cliffs, or how caribou traverse hundreds of kilometres of complex terrain during migration. Robots can leap atop obstacles, but they can’t do so with anything like the ease of a 200-pound cougar deliberately placing each paw. 

The sum of their parts: The researchers found that robotics engineers are actually better at building individual parts that outperform those of the natural world—carbon-fibre arms are superior to bone in numerous ways, for example. Where they fall short is in integrating those parts. 

“It seems to be that biology knows how to use worse parts to get better outcomes,” Sam Burden, a professor at the University of Washington and one of the researchers behind the study, told The Logic in an interview.

One of the biggest single shortfalls is energy storage. Lithium-ion batteries have made huge progress in recent years. But they’re nowhere near as efficient as biological systems, which use the body’s metabolism to convert fat into adenosine triphosphate—essentially, the gasoline that fuels biological cells. 

As for sensors, while electronic chips have enormous computing power, animals have many more nerve endings and neurons that make them more agile in response to changes in terrain. 

It’s all connected, man: To continue improving robot movement, engineers will have to get a firmer grasp of what Burden calls “integration principles”—how to build a complete artificial system that functions as well as a skeleton wrapped in sinew powered by neurons.

That’s no easy task. Engineers have always used specific performance metrics to drive technology forward, whether to get more torque out of a drivetrain system or more efficiency out of a jet engine. But when it comes to understanding the movements of the natural world, he said, there are metrics that researchers “don’t even know how to define.”

Asked which creature best represents the chasm between robots and biology, Burden eschewed some of the grander members of the animal kingdom and cited the humble squirrel. The rodents possess a twitchy burst and dexterity that put any robot to shame, he said. Any Canadian who’s tried to keep the critters from raiding their bird feeder would surely agree.