Abstract
Conventional augmented reality (AR) systems often struggle to accurately and contextually place user interface (UI) elements, leading to a non-immersive or intrusive user experience. This disclosure presents a method for personalized UI placement in AR through reinforcement learning. The technique utilizes an agent trained with deep reinforcement learning to determine the optimal position and orientation of virtual UI elements. Input features, including environmental depth maps, eye-tracking data, and application context, are processed to inform placement decisions. The agent learns from both implicit user interactions, like gaze aversion or physical interference, and explicit user corrections. This process allows the UI placement to adapt over time to individual user behavior and environmental context, thereby improving the usability and realism of the AR experience.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Shin, Dongeek, "Adaptive Placement of Augmented Reality User Interface Elements Using Reinforcement Learning", Technical Disclosure Commons, (December 15, 2025)
https://www.tdcommons.org/dpubs_series/9033