Researchers at Worcester Polytechnic Institute (WPI) are using ultrasound sensors and a type of artificial intelligence (AI) to enable palm-sized aerial robots to navigate with limited power and computation through fog, smoke and other difficult conditions amid search-and-rescue operations.
Taking inspiration from bats, the team suggests that ultrasound may be a possible solution to navigation technologies that add weight to a drone or that falter in poor conditions.
Source: WPI
"Bats that weigh less than two paper clips can accurately navigate in dark, damp, and dusty caves by sending out short chirps and listening to the weak echoes with a limited number of neurons," the researchers explained. "By creating an ultrasound-based system that needs just two tiny sensors and little computation, we can open up opportunities for small aerial robots to perceive their surroundings, make decisions, and independently operate longer in cluttered, hazardous places where current aerial robots struggle."
Typically, autonomous aerial robots use sensors, controllers, cameras, a power source and algorithms to understand their surroundings and to make navigational decisions. While some robots gather information about a landscape by analyzing radio waves or light pulses, technology based on lidar and radar tend to be heavy, power intensive and expensive. Further, darkness, bad weather and noise can interfere with light-based perception systems.
As such, the WPI team customized an X-shaped aerial quadrotor drone about 6 inches wide and weighing 1 lb with ultrasound sensors and a physical barrier dubbed an acoustic shield that dampens propeller noise. The team also used the AI technique deep learning to train the robot's computer to analyze weak ultrasound echo patterns much like the way a bat brain processes sound to decipher echoes.
The robot was tested both outdoors in a wooded area and indoors in a laboratory outfitted with obstacles like transparent plastic or metal poles.
Meanwhile some indoor tests were conducted in the dark with black obstacles, while others featured fog or snow blown onto the obstacle course.
The robot reportedly had a success rate of 72% to 100% in navigating through challenging courses during 180 tests. However, the robot struggled with dodging thin objects, like metal poles and slender tree branches.
An article detailing the robot, “Milliwatt ultrasound for navigation in visually degraded environments on palm-sized aerial robots,” appears in the journal Science Robotics.
