Newswise — A new technology developed by engineers at the University of California San Diego has the potential to make augmented reality (XR) experiences smoother and more seamless. The technology consists of an asset localization system that uses wireless signals to track physical objects with centimeter-level accuracy in real-time and then creates a virtual representation of those objects. Uses of this technology range from enhancing virtual gaming experiences to improving safety in the workplace.
A team led by Dinesh Bharadya, a professor in the Department of Electrical and Computer Engineering at the UC San Diego Jacobs School of Engineering, presented the technology ACM Conference on Embedded Networked Sensor Systems (SenSys 2023) It was held in Istanbul, Turkey.
Existing localization methods face significant limitations. For example, many XR applications use cameras to localize objects, whether through virtual reality (VR) devices, augmented reality (AR) glasses or smartphone cameras, said study co-first author Aditya Arun, who has a Ph.D. in electrical and computer engineering. .D. student in Bharadya's lab.
“However, these camera-based methods are unreliable in highly dynamic scenarios with visual obstacles, rapidly changing environments, or poor lighting conditions,” Arun said. Meanwhile, wireless technologies such as WiFi and Bluetooth Low Energy (BLE) often fail to provide the required accuracy, and ultra-wide-band (UWB) technology involves complex setup and configuration.
The new asset localization system, developed by Bharadia's team at UC San Diego in collaboration with Shunsuke Saruwatari at Osaka University, Japan, overcomes these limitations by providing precise, real-time localization of objects with centimeter-level accuracy, even in dynamic and poorly lit conditions. environment. The system is also packaged in an easy-to-use and compact module measuring one meter that can be plugged into electronic devices such as TVs or soundbars with minimal setup.
The researchers built their system using the power of wireless signals up to 6 GHz. “Unlike camera-based methods, these wireless signals are less prone to visual blockage and continue to function even in non-line-of-sight conditions,” Arun said.
The system uses wireless signals to locate battery-powered UWB tags attached to objects. It consists of two main components. One is a UWB tag that transmits a beacon signal for localization. Another component is the localization module, which is equipped with six UWB receivers that are time and phase synchronized to receive the beacon signal. As this signal travels, it reaches each receiver at a slightly different phase and time. The system combines these differences intelligently to accurately measure the location of the tag in 2D space.
In tests, the researchers used their system to play a life-size chess game using everyday objects. They re-arranged the shelves with UWB tags, which they turned into virtual chessboards. As the pieces moved across the table, the system could seamlessly track their movements in real-time with centimeter-level accuracy.
“We found that our system is reaching 90e percent accuracy in dynamic scenarios and performs at least eight times better than state-of-the-art localization systems,” said Arun.
The team is currently improving the system. Next steps include improving the PCB design to make the system more robust, reducing the number of receivers to improve power efficiency, and adding antennas along the vertical axis to support full 3D localization.
The title of the paper: “XRLoc: Accurate UWB localization for XR deployment, co-authored by Aditya Arun*, Sureel Shah, and Dinesh Bharadya, UC San Diego; and Shunsuke Saruwatari*, Osaka University, Japan.
*These authors contributed equally to this paper.