JThis article is an episode of Future Explored, a weekly guide to world-changing technology. You can get stories like this straight to your inbox every Thursday morning by subscribe here.
A robot dog could be your new hiking buddy – if you can keep up with it.
In 2016, scientists from the Swiss research university ETH Zurich introduced the world to their four-legged autonomous robot, ANYmal. The robot dog has now been upgraded with a new controller that helps it traverse difficult terrain without any prior training.
To demonstrate this, ANYmal was tasked with hiking an unfamiliar trail in the Alps – and he reached the summit in four minutes. faster than the average human hiker, without falling or tripping.
Why is this important: As fun as a hike with a autonomous robot dog might be, these robots are not meant to replace pets – they are working dogs, designed to help out with boring, dirty or dangerous jobs.
While some robot dogs can actually To do things – the famous Boston Dynamics Spot has a removable arm that allows it to move objects and turn levers – most are primarily designed to move around and collect data using cameras and sensors.
Robot dogs can also work alongside humans.
They may walk right in front of a first responder during a search and rescue mission in a collapsed building, for example, helping them find survivors faster while searching for areas too unstable for a human to traverse.
Robot dogs could also join soldiers on the battlefield, conducting reconnaissance and surveillance missions that could help the military stay alive (the use of robot dogs as actual weapons is controversial and discouraged by many – but not at all — developers).
Eventually, we might even send robot dogs to places humans can’t yet go, like Mars.
Four-legged space explorers have advantages over wheeled robots, entering areas too dangerous for wheels or treads – including Mars’ underground lava tubes, which some experts say could harbor astronauts, or even harbor evidence of ancient life.
The challenge: We’re just getting to the point where we can reliably use a robot dog to inspect a chemical plant with a pre-determined layout, and it’s much easier than dropping one into uncharted caves on another planet.
To reach this level, we need to make it easier for robot dogs to analyze the unfamiliar terrain in front of them and come up with a plan to traverse it. before they take a single step forward.
“Snow, vegetation and water visually appear as obstacles that the robot cannot walk over.”
The developers of ANYmal
Roboticists call this “exteroceptive perception”, and it’s something humans do all the time – if we go outside and see that it’s snowed the night before, for example, we might walk a little more carefully on the sidewalk in anticipation of ice. .
Giving robot dogs that meaning has been difficult, however. There are an infinite number of environments, and what a bot can “see” with its cameras and sensors may not paint a complete picture.
This “remains a great challenge in robotics”, writes the ETH Zurich team in its new paper, published in Science. “Snow, vegetation and water appear visually as obstacles that the robot cannot walk over – or are completely absent due to high reflectance.”
If a robot is unable to proceed based on its exteroceptive perception, it could use the feedback it gets by physically interacting with the environment.
It’s called “proprioceptive perception”, and it’s also something humans do – imagine how you carefully navigate a dark room or stairwell, using feedback from your feet and your hands to determine if you can keep moving forward.
But just like a person in a dark room, a robot dog needs to move slower when relying on its proprioceptive perception, and we want robots to be able to move quickly, especially in situations like search missions. and rescue, where every minute counts.
“The controller allows ANYmal to tackle rough terrain faster, more efficiently and, above all, more robustly.”
What’s up: To help ANYmal navigate uncharted territory quickly, the ETH Zurich team developed a control system that switches between exteroceptive and proprioceptive systems, depending on how confident it is in what it “sees”.
They trained this controller using computer simulations of different terrains, some with incomplete or missing data. After training, the system could predict when ANYmal should trust what it “saw” and when it should rely more on what it might feel.
“[The controller] allows it to tackle rough terrain faster, more efficiently and, above all, more robustly,” said lead researcher Marco Hutter. said in a press release.
During real-world testing, the ANYmal successfully navigated loose rock, tight tunnels, and snow-covered stairways.
He also completed the hike in the Alps without any failures – despite encountering tree roots, steep slopes and slippery ground – completing the full loop in the time it takes the average person and by reaching the summit four minutes faster.
“When I first checked the terrain, I thought maybe it was too difficult for the robot, but it could handle anything,” said first author Takahiro Miki. says IEEE Spectrum.
Look forward: ANYmal’s new navigation capability is impressive, but the bot still isn’t as nimble as a real dog, and that’s the ultimate goal.
“We think the next level would be somewhere that requires precise movement with careful planning, like stepping stones, or certain obstacles that require more dynamic movement, like jumping over a gap,” Miki told IEEE. .
For now, the researchers will continue to develop their robot dog, preparing it for future deployment on construction sites, in disaster areas and, perhaps one day, on other worlds.
We would love to hear from you! If you have a comment about this article or have a tip for a future Freethink story, please email us at [email protected].