An unsupported bipedal robot capable of walking down steep slopes and traversing uneven ground has been developed by University of Michigan researchers.
Robots traditionally struggle with slopes and rough terrain, but the team has demonstrated a model that is far better at withstanding these natural conditions while maintaining its stability.
The robot's feedback control algorithms should also be able to help other two-legged robots as well as powered prosthetic legs gain similar capabilities.
"The robot has no feeling in her tiny feet, but she senses the angles of her joints – for instance, her knee angles, hip angles and the rotation angle of her torso," explained Professor Jessy Grizzle who worked on the project. "It's like walking blindfolded and on stilts."
The robot, dubbed ‘Marlo’, is Grizzle's first device that can walk and fall in any direction, known as 3D walking.
The previous robot, named ‘Mabel’, was only capable of moving in two dimensions and was attached to a boom to give sideways stability.
"Getting a robot to walk well in 3D can be a very frustrating process," said mechanical engineering student Xingye Da.
Da developed a way to control the robot with two 2D algorithms, expanding on previous algorithms used to control Mabel’s movements.
"The method could help many researchers speed up the process of achieving stable walking on their robots," he said.
The main controller handles the forward and backward motion and balance, while a second controller handles side-to-side balance.
The team created a library of 15 gaits to handle different walking speeds and ground heights. Each gait is optimised for maximum efficiency, which has the added effect of making the movement seem natural rather than robotic.
Marlo steps blindly, sensing the changes in ground height and adjusts her gait appropriately for the terrain and speed while blending gaits from the library to fit her environment.
The robot has been tested in a number of environments including walking through snow, down a steep hill, and on randomly stacked plywood squares covered in astroturf.
Although impressive, the approach has inherent limitations as the robot cannot make quick turns or sideways movements.
The team is now designing full 3D walking gaits that can theoretically be loaded onto any bi-pedal robot without specific modifications.