New Algorithm Lets Robots Run in the Wild

A new algorithm allows quadruped robots to run, jump and navigate through any terrain, as well as dodge obstacles, in what is hoped to be a pivotal moment in the creation of autonomous search and rescue robots. 

The new system was designed by a team from the University of California San Diego and combines a robot’s sense of sight with its sense of touch. That is, it harnesses a sensing modality called proprioception – referring here to the sensation of the ground beneath the robot’s feet. It is one of the first times these two methods of spatial awareness have been combined to give robots a more fully-rounded understanding of their environment. 

“In one case, it’s like training a blind robot to walk by just touching and feeling the ground. And in the other, the robot plans its leg movements based on sight alone. It is not learning two things at the same time,” said Xiaolong Wang, senior study author. “In our work, we combine proprioception with computer vision to enable a legged robot to move around efficiently and smoothly – while avoiding obstacles – in a variety of challenging environments, not just well-defined ones.”

The team combined data gathered from a camera mounted on the robot’s head with data from sensors on its legs to develop the algorithm, with simulations of a random combination of these data points fed into the robot, and a reinforcement learning model applied so the robot could build an understanding of appropriate action for each environment.

This approach helped the robot to autonomously and rapidly navigate and anticipate changes in its environment, allowing it to dodge obstacles faster on different types of terrains without the help of a human operator.

In tests, the algorithm enabled four-legged robots to walk and run while avoiding both static and moving obstacles. In the trials, the team tested the system’s efficacy in natural surroundings such as sand, gravel and grass, as well as in an office space. In all tests, the robots were able to navigate around obstacles such as trees and fallen branches, as well as people, desks and chairs in the commercial environment.  

Test findings are expected to be presented at this year’s International Conference on Intelligent Robots and Systems in Kyoto, Japan.

Next, the team wants to develop a system to allow robots to climb up and down stairs, as well as walk on stones or cobbles, and even jump over obstacles.