Robots are usually designed for navigating on specific surfaces, which provides them a certain efficiency when in motion. For example, robots that travel on rocky terrain will most often be outfitted with some type of off-road tires or robust legs to keep the robot stable as it navigates. While it's easy for us humans to walk or climb on almost any surface, designing robots that can move from one type of terrain to another is a difficult task to accomplish. We have years of experience and a myriad of senses that allow us to adapt on the fly, while robots rely on programming to move from one point to another.
Researchers from the University of Oslo's Department of Informatics have developed a robot that can adapt to different outdoor environments by adjusting the length of its legs, which enables it to transform into different body shapes for more efficient travel. The Dyret (Dynamic Robot for Embodied Testing) also utilizes AI to learn what leg length works best for specific surface types.
"The robot uses a camera to see how rough the terrain is, and it uses sensors in the legs to feel how hard the walking surface is," states research lead, Tonnes Nygaard. "The robot continuously learns about the environment it's walking on and, combined with the knowledge it gained indoors in the controlled environment, uses this to adapt its body."
During testing, Dyret quickly adapted to walking on grass, even though it only trained on gravel, sand, and concrete and had never encountered grass before. Shorter legs provide the robot greater stability, while longer give the robot increased speed, which was evident during the grass testing phase as the robot designated the terrain as unknown, and opted for stability over speed.
The robot can also learn to adapt if the legs become damaged and recover by limping or adjusting the functional appendages' height. As it stands, Dyret isn't ready to take on any tasks and is still under development. Still, the researchers state the robot could be used for search and rescue operations and agriculture applications in areas with wide-ranging weather conditions.