Don’t be misled by the fog, eerie lights, or fake bats—Worcester Polytechnic Institute’s robotics lab isn’t throwing a Halloween party.
Instead, the lab serves as a testing site for miniature drones designed to carry out search and rescue operations in dark, smoky, or stormy environments.
“We know that during earthquakes or tsunamis, power lines often fail—frequently at night—so waiting until morning to rescue survivors isn’t an option,” explained Nitin Sanket, assistant professor of robotics engineering. “That led us to look at nature. Could any creature handle this?”
Bat-Inspired Drones in Rescue Missions
Sanket and his students turned to bats, inspired by their remarkable echolocation skills. With funding from the National Science Foundation, they are creating small, cost-effective, and energy-efficient aerial robots capable of operating in conditions where conventional drones cannot.
Last month, emergency teams in Pakistan deployed drones to locate people stranded on rooftops amid severe flooding. In August, a drone helped rescuers locate a man in California who had been trapped for two days behind a waterfall. Earlier in July, drones assisted in mapping a safe path to reach three miners who had been trapped underground in Canada for over 60 hours.
Although drones are increasingly used in search and rescue, Sanket and other researchers aim to go beyond the current model of manually controlled, single robots. The next major advancement, according to Ryan Williams, an associate professor at Virginia Tech, is creating aerial robots that can operate in swarms and autonomously decide where to search.
“That kind of deployment—fully autonomous drones—barely exists,” he said.
Using Data-Driven Drones to Enhance Search and Rescue
Williams addressed this challenge in a recent project, programming drones to coordinate their search patterns with human rescuers. His team also analyzed historical data from thousands of missing-person cases to build a model predicting how a lost person might behave in the wilderness.
“We then used that model to guide our drones, focusing their searches on areas with the highest likelihood of finding someone,” he explained.
At WPI, Sanket’s project tackles other drawbacks of current drones, such as their size and sensory limitations.
“Most existing robots are large, heavy, costly, and can’t function in all environments,” he said.
In contrast, his drone is small enough to fit in the palm of a hand, built primarily from low-cost hobby materials, and capable of operating in complete darkness. It uses a tiny ultrasonic sensor—similar to those in automatic faucets—to mimic a bat’s echolocation, sending out high-frequency pulses and detecting obstacles through the returning echoes.
Drone Navigates in Darkness and Obstacles
In a recent demonstration, a student remotely launched the drone in a brightly lit room and then again with only a dim red light. As it neared a clear Plexiglas wall, the drone consistently stopped and reversed, successfully navigating even in darkness with fog and artificial snow in the air.
“Right now, search and rescue robots mostly work during daylight,” Sanket explained. “The challenge is that these operations are tedious, hazardous, and often take place in the dark.”
However, the development process wasn’t without hurdles. The team discovered that the noise from the drone’s propellers disrupted the ultrasonic signals, so they designed 3D-printed shells to reduce the interference. They also applied artificial intelligence to help the drone filter and interpret the sound signals effectively.
Drones Still Far from Matching Bats’ Precision
There’s still a significant gap between the drones and real bats, which can finely control their muscles to focus on specific echoes and detect objects as tiny as a human hair from several meters away.
“Bats are incredible,” Sanket said. “We’re far from replicating what nature can do, but the aim is that, eventually, our drones will reach that level and be practical for use in real-world deployments.”
Read the original article on: Tech Xplore
Read more: As AI Develops Further, it Could Become More Self-Serving
