Researchers at Worcester Polytechnic Institute (WPI), led by Nitin J. Sanket showed that palm-sized drones with ultrasound and AI can navigate fog, smoke, and other tough conditions using minimal power—perfect for search-and-rescue.
The bat-inspired breakthrough, reported in Science Robotics, indicates that ultrasound could offer a lightweight, low-cost alternative to traditional drone navigation systems, which often struggle in challenging environments.
“Bats weighing less than two paper clips can navigate dark, damp, and dusty caves by emitting short chirps and processing faint echoes with very few neurons,” said Nitin J. Sanket, assistant professor in WPI’s Department of Robotics Engineering.
“By developing an ultrasound system that uses only two tiny sensors and minimal computation, small drones can better sense their environment, make decisions, and operate autonomously for longer periods in cluttered, hazardous areas where current drones often fail.”
Nature-Inspired Robotics and Navigation Challenges
Sanket’s research centers on nature-inspired robotics, drawing ideas from creatures like bees and bats. Autonomous drones usually rely on sensors, cameras, controllers, power sources, and advanced algorithms to understand their environment and navigate effectively.
Certain robots gather environmental data using radio waves or light pulses, but lidar (light detection and ranging) and radar systems tend to be bulky, energy-demanding, and expensive.
Darkness, adverse weather, and ambient noise can disrupt light-based navigation systems. Propeller sounds further complicate a drone’s task of distinguishing useful echoes, consuming additional time and energy for processing.
Sanket’s team modified a 6-inch-wide, X-shaped quadrotor with ultrasound sensors and an acoustic shield to reduce propeller noise. They used deep learning to help the drone interpret faint ultrasound echoes, like a bat’s brain.
The roughly 1-pound drone was tested outdoors in a forested area and indoors in a lab filled with obstacles like transparent plastic and metal poles.
Drone Performance in Challenging Indoor Conditions
Some indoor tests used complete darkness or fog and snow on the course. The drone could navigate the course autonomously for about five minutes per flight on a single battery charge.
Across 180 tests, the researchers found the drone succeeded 72% to 100% of the time in navigating challenging courses. It struggled more with thin objects, like metal poles or slender tree branches, which reflected ultrasound signals weakly.
Sanket said the next step is smaller, lighter drones with longer flight and faster speeds using low-power ultrasound.
“In a real search-and-rescue operation, even a few extra seconds of flight can make the difference between life and death for a survivor,” he said.
Read the original article on: Tech Xplore
Read more: The wristband allows users to control a robotic hand using their own movements