Home → News → The Secret to Small Drone Obstacle Avoidance Is to... → Full Text

The Secret to Small Drone Obstacle Avoidance Is to Just Crash Into Stuff

By IEEE Spectrum

November 3, 2016

[article image]


Researchers at the University of Pennsylvania's (UPenn) GRASP Lab are working to develop autonomous quadcopters that can fly through windows without running into them.

However, some researchers think encouraging small aerial robots to run into objects may be a simpler solution for improving navigation and obstacle avoidance.

They believe the penalty due to collisions is small at these scales, and sensors and controllers are not precise enough to guarantee collision-free trajectories. The researchers say the focus should be on not completely destroying the system when it runs into something.

The GRASP team, led by professor Vijay Kumar, built a fleet of lightweight pico quads, each featuring a self-righting roll cage made from a heat-cured yarn consisting of 12,000 strands of carbon fiber. The UPenn researchers say the pico quads are controlled by a system that does not consider the position of other robots or obstacles, and does not contain special actions to command as a consequence of a collisions. Instead, the system tries to stabilize the pico quads as best it can while directing them toward a goal position.

The researchers found this strategy works impressively well, even when the robot has no knowledge of any obstacles or other pico quads that might be in its way.

From IEEE Spectrum
View Full Article

 

Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA

0 Comments

No entries found