Tuesday, November 21, 2017
Home » Leading Edge Robotics News » Teaching Robots How to Fall Safely

Teaching Robots How to Fall Safely

setting by creating a reduced-gravity environment using a tilted surface similar to an air hockey table outfitted with a leaf blower. Photo courtesy of Georgia Institute of Technology.
A small robot (consisting of a main body and two legs) was placed on a tilted surface similar to an air hockey table, to test algorithms designed to reduce the impact of a fall. Photo courtesy of Georgia Institute of Technology.

Unlike humans and robots, cats are known for their ability to always land on their feet. Now researchers at Georgia Tech’s School of Interactive Computing are studying the way cats reorient themselves during a fall, in an effort to make robots more robust.

According to the paper outlining the research, what makes a cat’s ability so extraordinary is that it can rapidly right itself even without an initial angular velocity or outside forces. Replicating this feat in a robot can’t be done using existing technology; the computations need to be done far too quickly. Georgia Tech researchers worked around this problem by conducting proof of concept testing in a reduced gravity environment: a tilted surface that worked much like an air hockey table.

The researchers hoped to find a control algorithm that can orient a falling robot into a landing pose that will reduce the impact of the fall. They thought, for example, that making the robot roll in a circle might reduce the impact of a fall. While an ideal algorithm for safe falling remains elusive, their work continues.

Leave a Reply

Your email address will not be published. Required fields are marked *