Researchers in Tokyo have created a groundbreaking way for robots to sense wind direction, inspired by the natural abilities of birds and insects.
Using flexible, hummingbird-like wings equipped with tiny sensors, they achieved an impressive 99% accuracy in detecting wind flow.
This innovation could transform how flying robots navigate challenging conditions.
Birds and insects naturally have special receptors on their wings that help them adjust their flight by sensing wind, body movements, and environmental changes.
Inspired by these natural abilities, scientists at the Institute of Science Tokyo, led by Associate Professor Hiroto Tanaka, designed flexible robotic wings that mimic those of hummingbirds.
Their study, published in Advanced Intelligent Systems, explains how these wings can sense wind with remarkable precision.
The researchers attached seven small, affordable strain sensors to flexible wings made from tapered shafts and film, mimicking the structure of real wings.
These wings were connected to a simple motorized flapping system, which moved the wings back and forth 12 times per second. They tested the system in a wind tunnel under gentle winds (0.8 m/s) from different angles, including no wind.
To analyze the data from the wing sensors, the team used a convolutional neural network (CNN), a type of artificial intelligence.
This allowed the robot to “learn” how to identify wind directions. The results were remarkable: with just one flapping cycle, the system achieved 99.5% accuracy in recognizing wind directions. Even with much shorter data, about 20% of a flapping cycle, the system was still 85.2% accurate.
Interestingly, even a single sensor placed on the wing could provide highly accurate readings, ranging from 95.2% to 98.8% for one flapping cycle.
However, using multiple sensors increased the reliability of the system, especially in shorter flapping cycles. Tests also showed that the flexible, shafted design of the wings enhanced the sensors’ performance.
This technology is a major breakthrough for small flying robots, which typically can’t carry bulky wind-sensing equipment. By adding lightweight strain sensors to their wings, these robots can gain an advanced ability to sense wind and adjust their flight, just like birds and insects do.
“This study shows that birds and insects may use similar wing strain sensing to control their flight in different wind conditions,” says Tanaka. “With this approach, we can create smarter, more adaptable flying robots.”
The discovery is not just a leap forward for robotics but also deepens our understanding of how nature solves complex challenges.
