D2.3 – Adaptive and personalized user interfaces

This deliverable presents the outcomes of Task 2.4, “Dynamic UI toolkit for human-centric and adaptive XR experience,” covering activities conducted between M1 and M26, towards the development of human-centred, context-aware, and adaptive Extended Reality (XR) environments for industrial applications.Task 2.4 focuses on designing and implementing Adaptive User Interfaces (AUIs) that automatically adjust graphical content, layout, and interaction methods based on user characteristics, environmental conditions, and task needs. This document presents the fourth phase of the INDUX-R Human-Centered Design (HCD) methodology, detailing the work done to develop adaptive and personalized interfaces that meet user requirements outlined in WP1. The activities are organized around three main areas: implementing cognitive load assessment tools using linguistic and paralinguistic analysis, developing a model-based Reinforcement Learning (RL) Decision Maker (DM) for real-time GUI adaptation, and creating adaptive GUIs with multiple Levels of Detail (LoD) tailored to the needs of the INDUX-R pilot applications. Following the HCD approach, end-users are involved throughout the development process to ensure human-in-the-loop adaptation and usability-focused XR interfaces.The resulting INDUX-R Adaptive Framework integrates ontology-based context modelling, combinatorial optimisation (CO) strategies, RL techniques, and real-time rendering mechanisms within the Unity 3D engine. Together, these components enable intelligent decision-making regarding what information should be displayed, when it should appear, and how it should be positioned within the user’s field of view. Ten AUIs categories have been developed across the two pilot applications, five for Use Case (UC)1 and five for UC2, reflecting use-case-specific operational requirements and user-defined functionalities. Iterative prototyping with expert-based heuristic evaluations and early-stage user testing demonstrates how AUIs improve interaction quality and task performance across two INDUX-R UCs: UC1, focusing on collaborative omniconference environments, and UC2, addressing industrial inspection and training workflows. Insights from the evaluations led to adjustments in the interface, optimisation of gesture-based interaction, and improvements in adaptive visualisation strategies.This document marks the completion of Task 2.4 and provides its outputs to the development of the INDUX-R adaptive applications.