- Researchers at the University of Glasgow have patented a system that tracks the behavior of elderly or at-risk individuals in their homes, without requiring those people to wear RFID tags
- With RFID interrogator on one wall, and tags arrayed on the opposite wall, the system uses machine learning and AI to identify when someone is in the room, disrupting the transmission, and how they are positioned or moving
Most common RFID-based tracking systems consist of readers capturing and identifying the presence of RFID tags, able to be applied to people’s badges or wristbands for systems such as access control.
University of Glasgow is turning that approach on its head, by identifying not when a tag is read, but when that tag read is interrupted and in what way. With that data, the university’s department of engineering is using RFID technology to track and understand human movements and positions for assisted living. Mohammad Zakir Khan, researcher at the school and author of the white paper, said the team has dubbed the technology the Transparent RFID Tag Wall (TRT-Wall).
With two walls facing each other in a five-meter-wide room, the Human Activity Recognition (HAR) system leverages an Impinj R700 reader with a circularly polarized antenna. It interrogates 15 tags applied to the opposite wall in a three by five grid, each placed 30 centimeters apart.
If an individual walks between the walls, they will disrupt the transmission, even if only subtly, and how that disruption is experienced helps identify the speed a person is walking, whether they are sitting. standing or if they fell. In fact, the system is designed to detect actions such as arm swings and knee bends.
Keeping An Eye on At-Risk People
“The TRT-Wall system revolutionizes RFID by enabling non-intrusive, contactless human activity recognition without requiring tags to be worn,” said Qammer H. Abbasi, University of Glasgow professor of Applied EM and Sensing, and project lead. “This approach overcomes the limitations of wearable and camera-based systems, making it an invaluable solution for privacy-sensitive, real-time monitoring needs.”
The researchers’ goal was to find a non-invasive, handsfree, wireless system to track people for safety purposes, such as identifying if they are in trouble or if their activity changes. Khan points out that using RFID tags or sensors attached to a person’s clothing or body would be unrealistic.
“We wanted something that would be totally different from the typical experience of using the RFID tags,” he said, calling the alternative that his team developed a contactless version since individuals don’t need to wear tags.
Instead, the system relies on return signal strength indicators (RSSI) when the reader receives its response transmission for the tags on the opposite wall. Khan said it is intended to track five states within a room: a person sitting, standing, walking in two directions and no-activity at all.
Disrupting RFID Transmissions
The human body is filled with water, and water can alter an RFID transmission. Often the transmission is not entirely blocked, but the sensitivity is reduced.
With that in mind, the group set the system to prompt about ten responses from each tag every second. If that number of responses is received, no one is in the room. When someone does enter to the room, as they pass in front of a tag it may respond to the reader only three or five times per second.
“On the basis of blockage, I can easily identify where the person is,” Khan said, according to the specific tag whose transmission was compromised. The second level of understanding is managed in the software to properly identify what kind of activity the occupant of the room is performing.
Reducing Tag Response
The tags are designed to provide both horizontal and vertical positions—if a tag mounted against the wall near the floor is not responding properly, the system can determine that someone is sitting in front of the wall, rather than standing. The placement of reader antenna also influences read response.
The speed at which those changes in read performance take place can indicate an individual’s behavior, differentiating if they fell, are sitting, or bending over to pick something up. Additionally, Khan said “we can easily identify direction of walking with a person forward moving or backward.”
The RFID tagged walls have been tested at 2.5 meters distance from the reader, 3.54 meters and up to 5 meters. While the tags can technically sense up to 9.8 meters in optimal conditions, Khan said, “our testing was conducted in a more realistic, furnished environment with tables and chairs to create lateral obstructions. This better simulates real-world conditions, unlike an empty room setup.”
Beam Steering to Focus Data
The software leverages machine learning and artificial intelligence (AI) to understand changes in read rates. The system can also enhance the relevant data being received through a process of beam steering of RFID transmission from the reader.
If beam steering technology is used (which the team has not yet tested), more sensitive data could be collected when an individual is detected in the room. For example, a person who is sitting in the corner, could train the system to understand where the individual is, and the reader antennas could then steer toward that area to gain more precise data.
“With the help of this feature we can easily steer signals to a specific position where we want interesting data.”
Identifying Individual People
In separate research, the team is developing sensors to not only detect when a person enters a room, but who that person is based on physical data such as heart rate or breathing rate, leveraging IR sensors and AI to identify which pattern is typical for which person. Potentially RFID technology could be designed to identify an individual’s size or gait as well, and even if that person has changed their gait, such as developing a limp.
Khan added that differentiating the presence of something such as pets can be accounted for.
“Animals such as a dog or a cat could be filtered out of the system based on their size and movement,” he said.
Thus far testing has centered only on identifying the presence of a single person, with Khan adding, “multi person detection is still a challenge.”
Commercial Applications in Office, Parking Management
Beyond healthcare and safety, the TRT-Wall could be used for occupancy monitoring, and tracking or localization use cases, Khan stated. For instance, the researchers have tested using the system in more public locations such as an office conference room. Signal changes within a conference room could indicate in which chairs people tend to sit. That data could provide benefits for facilities planning.
Similarly, the RFID system could be deployed in a parking lot to identify which spaces are occupied by vehicles. While people can block or alter RF transmission, so can metal objects. This application has been tested in a Glasgow parking lot.
“Each application benefits from TRT-Wall’s privacy-preserving, passive monitoring capabilities, positioning it as a powerful tool for smart, connected environments across diverse environments,” Abbasi said.
The TRT-Wall project began in April, 2023, and Khan said there is a patent pending on the development.