Pigeons Give Robots A Bird’s-eye View

Pigeons Give Robots A Bird’s-eye View

A pigeon gets ready to fly through an obstacle course, a high-speed camera attached to his head. Photo by Talia Moore.

Researchers from Harvard University and the Massachusetts Institute of Technology are attaching high-speed cameras to pigeons to gain insight into how the birds navigate through dense forests and around other obstacles. They believe this information may lead to new developments in robotics and auto-pilot technologies.

“Most bird navigation studies have looked at the larger context of migration,” said Harvard’s Huai-Ti Lin. “We’re interested in short-range path finding, how to get from point A to point B, in the context of the forest environment or obstacles in an urban environment.”

Attaching a camera to the birds, and filming them from several other angles, enables the researchers to reconstruct both what the birds see and how they move. Pigeons were selected as study subjects because they are easily trained, are very maneuverable, and their 300-degree panoramic vision allows them to assess obstacles on either side.

The researchers tracked pigeons as they flew through an indoor track with artificial trees as obstacles, then conducted an optical analysis of the film to identify what information was prominent from the bird’s-eye view. Among other things, they found that the birds most often chose the straightest route through the obstacles, and they exited the course heading in the same direction as they entered, despite making numerous twists and turns.

“These birds don’t seem to plan very far ahead, which is surprising because they could see far ahead,” said Lin. “Depth perception is tricky for birds with that kind of vision. If they try to plan too far ahead, they don’t have enough information to tell them which way to go. Some information is only available to them as they are approaching obstacles.”

The researchers believe that flying animals use a set of simple rules to determine how to maneuver around obstacles, and those rules can be applied in a variety of engineering technologies. Lin’s collaborators at MIT are using the pigeon data to develop an obstacle avoidance algorithm for flying robots that may be developed in the future. “If there are simple rules of thumb in biology that get you through obstacles without causing much damage,” Lin said, “then that could be a good rule to apply in robotics,” including unmanned aircraft that must navigate through nonuniform environments.

The next step for Lin and his colleagues is to determine if the obstacle avoidance rules birds use over a short course are also used over longer routes, at higher speeds, and through denser forest environments. The next course they build will include obstacles that move in real time, forcing the birds to make last second changes to their route.

 
Discussion

No discussion as of yet.

Join the discussion

To ensure a respectful dialogue, please refrain from posting content that is unlawful, harassing, discriminatory, libelous, obscene, or inflammatory. Northern Woodlands assumes no responsibility or liability arising from forum postings and reserves the right to edit all postings. Thanks for joining the discussion.

Please help us reduce spam by spelling out the answer to this math question
three plus three adds up to (3 characters required)