In order to help find trapped victims in collapsed buildings or other cluttered spaces, scientists have created a new type of AI hummingbird robot that can reach places where drones can’t.
Researchers from Purdue University have created flying robots that behave similar to hummingbirds and are trained by machine learning algorithms based on different techniques the bird uses naturally every day.
After learning from a simulation, the robot bird will know how to move around on its own just like a normal hummingbird. AI, along with flexible flapping wings, allows the robot to teach itself new tricks. For instance, even though the robot can’t see it, it senses by touching surfaces. Every touch alters an electrical current that could be tracked, as per Science Daily.
“The robot can essentially create a map without seeing its surroundings. This could be helpful in a situation when the robot might be searching for victims in a dark place - and it means one less sensor to add when we do give the robot the ability to see,” said researcher Xinyan Deng.
As compared to drones, the hummingbird robot uses entirely different aerodynamics due to their flexible wings. The robots have 3D-printed bodies, wings made of carbon fiber, and laser-cut membranes. The 12g robot is able to lift more than its own weight – up to 27g. The robot uses only two motors and can control each wings independently.
The robotic hummingbirds will not only help with search-and-rescue missions, but will also let biologists study hummingbirds more reliably in their natural habitat through the senses of a realistic robot. “We learned from biology to build the robot, and now biological discoveries can happen with extra help from robots,” Deng said.
Video Courtesy: Purdue Bio-Robotics Lab/YouTube