The use of drones to reach dangerous or unstable areas continues to grow. While research has come a long way in how these robots fly, such as mimicking insect movement, or the ability to move objects, giving them more biological senses may have once seemed a bit far-fetched. Since detecting chemicals, gas leaks, explosives and other potentially disastrous situations is a key benefit of their use, University of Washington researchers set out to give their drones a sense of smell.
Knowing that most sensors can’t quickly process specific smells, the researchers turned to the Manduca sexta hawkmoth, commonly known as the Carolina Sphinx moth, for inspiration. When the moth picks up a scent, the particular odor attaches to proteins on its antenna, which in turn activates neurons to detects certain chemicals.
“Cells in a moth antenna amplify chemical signals,” said Thomas Daniel, UW professor of biology. “The moths do it really efficiently—one scent molecule can trigger lots of cellular responses, and that’s the trick. This process is super-efficient, specific and fast.”
The research, led by Melanie Anderson, mechanical engineering doctoral student, set out to incorporate the ability of a moth antennae into an autonomous drone. The result was Smellicopter, which navigates toward smells via a live moth antenna. After cold anesthetizing a moth, the team removed the insect’s antenna, which stays functional for up to four hours. Wires were then added to both sides of the antenna, enabling connectivity to electrical circuit.
“A lot like a heart monitor, which measures the electrical voltage that is produced by the heart when it beats, we measure the electrical signal produced by the antenna when it smells odor,” Anderson said. “And very similarly, the antenna will produce these spike-shaped pulses in response to patches of odor.”
Using human-made sensors for comparison, the team tested the antenna’s response to a floral scent and ethanol. Puffs of odors were released from one side of a wind tunnel for both the antenna and sensors to process. The antenna proved to be faster and required less time for recovery between each odor release.
Once the antenna proved reliable, it was time to give the antenna a platform to showcase its abilities. The researchers attached it to a commercially available, customizable handheld quadcopter. Two plastic fins were added to the copter to ensure the drone was oriented upwind.
The team developed a cast-and-surge protocol to direct how the drone flew. They programmed it to mimic a moth’s natural ability to seek out smells. It starts by navigating left. If nothing is detected after it travels a certain distance, it then moves to the right covering the same distance. Its flying pattern only changes if an odor is detected.
“So, if Smellicopter was casting left and now there’s an obstacle on the left, it’ll switch to casting right,” Anderson said. “And if Smellicopter smells an odor but there’s an obstacle in front of it, it’s going to continue casting left or right until it’s able to surge forward when there’s not an obstacle in its path.”
During testing, Smellicopter proved it could avoid obstacles. The drone is outfitted with four infrared sensors that measure the area around it 10 times each second. If something is detected within 20 centimeters, the drone shifts directions and continues its normal cast-and-surge routine. Instead of GPS, cameras are used for surveying. The process is similar to how insects see. The team believes that makes it an ideal option for small or tight underground or indoor areas that large drones would be unable to access.
Although Smellicopter proved to be quite moth-like and preferred flying toward more interesting smells, such as florals, the team hopes to address that with future research. Their goal is to somehow train the drone to process specific smells, such as carbon dioxide or chemicals released after an explosion.