According to foreign media reports, Purdue researchers have recently gained some inspiration through the sensors of spiders, bats, birds and other animals to help improve the ability of self-driving cars and drones. These new sensors are associated with the nerve endings of the organism and special neurons called mechanoreceptors. These only need to be in the form of hair, cilia or feathers to detect and process the information necessary for insect/animal survival.
Andres Arrieta, assistant professor of mechanical engineering at Purdue University, said: "Intelligent systems can already collect large amounts of data, and this speed is growing faster than data that traditional computing can handle." His lab applies natural principles to structures. Designed from robots to aircraft wings. He also said: "Nature does not need to collect every piece of data, it will automatically filter out the required data."
Many biomechanical sensors filter data based on thresholds (such as changes in pressure or temperature) (ie, the information they receive from the environment). For example, a spider's furry mechanical sensor is located on its leg. When the spider web vibrates at a frequency associated with the prey or spouse, the mechanical sensor detects it, produces a reflection in the spider and then reacts very quickly. Mechanical sensors do not detect lower frequencies, such as dust on the net, because it is not important for spider survival.
The idea is to integrate similar sensors, ultrasonic sensor directly into the housing of an automated machine, such as an airplane wing or a car body. In a paper published by the researchers on the ACS Nano, researchers have shown that mechanical sensors inspired by spider hair can be customized to detect predetermined forces. For example, in real life, these forces will be associated with an object that the autonomous machine needs to avoid.
But the sensors they developed not only can be perceived and filtered at very fast speeds, they can be calculated, and power is not needed. Professor Arrieta said: "There is no difference between hardware and software. They are all interrelated. Sensors are designed to interpret data and collect and filter data."
In nature, once a particular force level activates a mechanoreceptor associated with a hairy mechanical sensor, these mechanoreceptors calculate information by switching from one state to another. Researchers at Purdue University worked with Nanyang Technology University in Singapore and ETH Zürich to design the same sensors and use these switch states to interpret these signals.
According to Professor Arrieta, these artificial mechanical sensors are capable of sensing, filtering and calculating very quickly due to their high hardness. The sensor material is designed to quickly change shape under the action of external forces. The change in shape causes the conductive particles inside the material to approach each other, allowing current to pass through the sensor and carry the signal.
Professor Arrieta said: "With machine learning algorithms, we can train these sensors,ultrasonic sensors from the main work with minimal energy consumption, and there are no obstacles in manufacturing sensors of various sizes."