Imagine running on a cement path, then suddenly on dry sand. Just to stay upright you will have to slow down and change the way you run. Likewise, a walking robot would need to change gait to manipulate different surfaces.
Usually we humans and most robots can only change How? ‘Or’ What we will run. But what if we could also change our body shape to run as fast and safely as possible on any surface?
We would like to rely on robots for difficult and dangerous tasks, from inspecting broken nuclear reactors to space exploration. For these tasks, a static body could limit the adaptability of the robot. A shapeshifting body could be the difference between success and failure in these unexpected environments. Better yet, a shape-changing robot could learn the best body shape for different environments and adapt to new environments as it encounters them.
In collaboration with the University of Oslo, we have successfully tested this idea with a four-legged robot that adapts its body to walk on new surfaces as it sees them, performing better than a static body robot. Our research is published in Nature Machine Intelligence.
A quadruped that changes shape
DyRET, the dynamic robot for embedded testing, or “the animal” in Norwegian from its creator, Tønnes Nygaard, was designed to explore the idea of a shape-changing robot. Each of DyRET’s four feet has two telescoping sections, so he can change the length of his thigh or shins. The adjustments are made by motors integrated in the legs and the lengths can be changed automatically while the robot is running.
Motors can change the height of DyRET by approximately 20%, from 60cm to 73cm in height. These 13 cms make a dramatic difference in the robot’s walking. With short legs, DyRET is stable but slow, with a low center of gravity. In its highest mode, DyRET is more unstable when walking but its stride is much longer, which allows it to travel faster and overcome obstacles.
DyRET also has sensors to track what it is walking on. Each of DyRET’s feet has a force sensor that can sense the hardness of the ground. A 3D camera points towards the ground between the front legs of DyRET to estimate the roughness of the ground.
Learn to adapt
When DyRET walks, it constantly detects the environment through its feet and its 3D camera. When the robot detects a change in ground conditions, it can choose the best leg length. But how does the robot know which body shape works best?
We explored two ways for DyRET to learn the best leg setup for different situations: a controlled environment, indoors with known surfaces, and real-world testing outdoors.
In our controlled tests, DyRET walked inside boxes about 5 meters long containing different walking surfaces: sand, gravel and sheets of hard fiber cement. The robot stepped on each material in each of 25 different leg configurations to record the efficiency of its movement. Given this data, we tested the robot’s ability to automatically detect a change in the walking surface inside the boxes and choose the best body shape.
While our controlled experiments have shown that DyRET can successfully adapt his body to surfaces it has walked on before, the real world is a much more variable and unpredictable place. We have shown that this method can be extended to invisible terrain by estimating the best body shape for any surface the robot encounters.
In our outdoor experiments, DyRET used a machine learning model, seeded with knowledge about the best leg configuration for a given combination of hardness and terrain roughness from controlled testing. As the robot walks, it continually predicts the best body shape for the terrain when it encounters it, while also updating its model with measures of its ability to walk. In our experiments, DyRET’s predictions improve as he walks, allowing him to generate efficient movements quickly, even for terrain he has never seen before.
Are shape-shifting robots the future?
DyRET explores the idea of ”embodied cognition” in a robot: that is, the material body of a robot can be used to solve problems in collaboration with its software brain by connecting them closely to the environment . Instead of DyRET’s body being a constraint on its movement, it is itself an adaptive way of solving problems in harsh environments.
This is incredibly beneficial, especially when we cannot predict the exact environmental conditions in advance, making it very difficult to choose just one “right” form of robot. Instead, these robots would adapt to a wide variety of environmental conditions through a change in shape.
Our proof of concept has powerful implications for the future of robotic design, unlocking currently impossible environments that are very difficult and variable. Future shape-changing robots could be used on the seabed or for long-term missions in space.
This article by David Howard, Data61 and Charles Martin, Australian National University is republished from The Conversation under a Creative Commons license. Read the original article.
Published March 16, 2021 – 08:08 UTC