Posted in | Automotive Sensors

Scientists Design Animal-Inspired Sensors for Self-Driving Cars to See Better

Imagine if self-driving cars and drones possessed the tingling “spidey senses” of Spider-Man. They might truly be able to spot and avoid objects better, says Andres Arrieta, an assistant professor of mechanical engineering at Purdue University, as they would process sensory data very fast.

In nature, “spidey-senses” are activated by a force associated with an approaching object. Researchers are giving autonomous machines the same ability through sensors that change shape when prompted by a predetermined level of force. (ETH Zürich images/Hortense Le Ferrand)

Improved sensing capabilities would render it possible for drones to steer through risky environments and for cars to avoid accidents caused by human fault. Present-day high-tech sensor technology does not process data sufficiently quick– but nature does.

Plus scientists would not have to engineer a radioactive spider to bestow autonomous machines with superhero sensing capabilities.

In its place, Purdue scientists have constructed sensors inspired by spiders, birds, bats, and other animals, whose genuine spidey senses are nerve endings connected to special neurons called mechanoreceptors.

The nerve endings – mechanosensors – only sense and process information vital to an animal’s survival. They can be in the form of cilia, hair, or feathers.

“There is already an explosion of data that intelligent systems can collect – and this rate is increasing faster than what conventional computing would be able to process,” said Arrieta, whose lab applies principles of nature to engineer structures, spanning from robots to aircraft wings.

“Nature doesn’t have to collect every piece of data; it filters out what it needs,” he said.

A number of biological mechanosensors filter data – the information they obtain from an environment – based on a threshold, such as variations in temperature or pressure.

The hairy mechanosensors of a spider, for instance, are situated on its legs. When a spider’s web vibrates at a frequency related to prey or a mate, the mechanosensors sense it, producing a reflex in the spider that then reacts very swiftly. The mechanosensors would not detect a lower frequency, like dust on the web, as it is insignificant to the spider’s survival.

The idea would be to incorporate similar sensors directly into the shell of an autonomous machine, such as the body of a car or an airplane wing. The scientists illustrated in a paper published in ACS Nano that engineered mechanosensors inspired by the spider’s hairs could be modified to detect pre-established forces. In the real world, these forces would be related to a certain object that an autonomous machine has to avoid.

But the sensors they built do not merely sense and filter at a very rapid rate – they also compute, and without necessitating a power supply.

There’s no distinction between hardware and software in nature; it’s all interconnected. A sensor is meant to interpret data, as well as collect and filter it.

Andres Arrieta, Assistant Professor, Mechanical Engineering, Purdue University

In nature, once a specific level of force triggers the mechanoreceptors related to the hairy mechanosensor, these mechanoreceptors compute information by swapping from one state to another.

Purdue scientists, in partnership with Nanyang Technology University in Singapore and ETH Zürich, engineered their sensors to act in the same way, and to use these on/off states to understand signals. An intelligent machine would then react based on what these sensors compute.

These artificial mechanosensors can sense, filter, and compute very rapidly as they are rigid, Arrieta said. The sensor material is designed to quickly change shape when triggered by an external force. Shape changing causes conductive particles inside the material to move closer to each other, which then permits electricity to pass via the sensor and convey a signal. This signal notifies how the autonomous system should react.

With the help of machine learning algorithms, we could train these sensors to function autonomously with minimum energy consumption. There are also no barriers to manufacturing these sensors to be in a variety of sizes.

Andres Arrieta, Assistant Professor, Mechanical Engineering, Purdue University

This work received financial support from ETH Zürich and Purdue University, and lines up with Purdue's Giant Leaps celebration, recognizing the university’s international advancements made in algorithms, AI, and automation as part of Purdue’s 150th anniversary. This is one of the four themes of the yearlong celebration’s Ideas Festival, programmed to highlight Purdue as an intellectual center solving everyday issues.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback