Here at The Optimist Daily, we think it’s important to keep our readers up to date with the latest news on robot progress. These incredible machines may be the key to help carry our future society. Recently, we reported on how scientists taught robots the complicated task of how to open doors, improving the role of autonomous activities the machines can carry out.
This time, scientists from ETH Zurich, have managed to create a robot that can hike! This allows the machine to move over uneven terrain quickly and robustly. The experiment showed that their creation, ANYmal, can complete a virtual hike up Mount Etzel. The hike across the mountain, situated overlooking Lake Zurich, was completed on average four minutes faster than humans could with no mistakes in footing.
Utilizing the environment and touch
The researchers achieved this through machine learning techniques, recruiting visual perception of the environment and touch. Using touch allows for an understanding of the environment, even when visibility is low. Similarly, an experiment we previously wrote about also combines these features to escape a maze. It seems using animal-like senses and taking inspiration from the natural world may be the key to programming the most useful robots.
Using the slippery, steep, obstacle-ridden terrain, ANYmal’s algorithm was trained. “The robot has learned to combine visual perception of its environment with proprioception — its sense of touch — based on direct leg contact. This allows it to tackle rough terrain faster, more efficiently and, above all, more robustly,” stated Marco Hutter, lead researcher on the project.
The results, published in Science Robotics, discuss an exciting future for this technology. In the future, ANYmal may be used to travel over areas that are too dangerous for humans or other robots to pass. The research group plans to expand its testing to different types of terrain. “With this training, the robot is able to master the most difficult natural terrain without having seen it before,” said Hunter.
Source study: Science Robotics – Learning robust perceptive locomotion for quadrupedal robots in the wild