Pitt | Swanson Engineering
Powering AI in Sensors with Energy Harvested from Nature
Jingtong Hu and collaborators received an NSF award to enable AI in sensors that are powered by energy-harvesting technology
Experts rely on remote sensors fastened beneath a bridge that continuously detect vibrations and produce data for structural health monitoring. Jingtong Hu plans to apply artificial intelligence to extend the lifetime of sensors and devices deployed in remote areas. Photo credit: Shutterstock.

In a city with 446 bridges, how do you effectively monitor the health of the structures that help residents navigate above Pittsburgh’s three rivers? 

Experts rely on remote sensors fastened beneath a bridge that continuously detect vibrations and produce data for structural health monitoring. These kinds of wireless, battery-operated devices are often placed in hard-to-reach areas, complicating maintenance. Researchers from the University of Pittsburgh and the University of Notre Dame want to apply artificial intelligence to extend the lifetime of sensors and devices deployed in remote areas. They received a $500,000 award from the National Science Foundation to support their work. 

“One of the major challenges with these sensors is battery replacement. Many times, it is costly, inconvenient, or even infeasible to replace or charge these batteries after deployment,” said Jingtong Hu, lead researcher on the study and associate professor of electrical and computer engineering at Pitt’s Swanson School of Engineering.

Hu and the research team want to develop a way to save power on the remote sensor device by leveraging energy-harvesting technology, which sources power from the environment, such as solar, thermal, or wind.

They plan to add a second, small sensor that can trigger a more robust device, thus saving energy and allowing users to change the battery less frequently. The smaller sensor -- powered by energy harvested from the environment -- will run unattended, and with the help of AI, it can be trained to recognize patterns and signal the larger device to turn on during a specific event.

“The main device is programmed to do all of the legwork,” explained Hu. “The smaller sensor is the watchdog that can monitor the environment and wake up the larger sensor when necessary.”

These devices have many applications, including monitoring and predicting natural disasters. Sensor technology is currently used to observe gases emitted by active volcanoes in some of the most remote parts of the planet. This requires researchers to take long, arduous hikes to reach the location -- all while wearing protective equipment to prevent damage to the skin and lungs from the extreme heat and corrosive gases. 

With Hu’s improvements, the researchers may be able to make trips such as these – whether to the tops of volcanoes or under bridge trusses –  less frequently. If successful, the project may ultimately allow these devices to be powered by the environment to help protect the environment.

He will collaborate with Yiyu Shi, associate professor of computer science and engineering at the University of Notre Dame.

“One of the main challenges of running AI algorithms with energy harvested from the environment is that the energy from the environment is intermittent,” Hu explained. “Much like a laptop, if the sensor loses power, you lose the data, so we want to help AI algorithms reach an accurate decision, even with intermittent power.

“By applying AI, we hope to increase the lifespan of unattended sensors and make them more reliable and useful,” he said.

# # #

9/30/2020

Contact: Leah Russell