About LAPIS projects

Scientific projects

ExQuMe - Explaining Quantum Mechanics SAFER - Smart AI- and IoT-based Fire Extinguisher CPM Smart Predictive Analytics for Reliable Turbine Assets AERONAUT - Advancing research capacities for the development of autonomous marine-aerial systems for novel Adriatic applications Nonverbal Communication-Based Adaptive Learning Model for Interactive AI Agents Virtual - transfer - Technology Transfer of an Affective Virtual Being for Immersive Interaction AMICORC Project ACRON - A New Concept of Applied Cognitive Robotics in Clinical Neuroscience Mixed reality applications

Student projects

Robotic plotter for printing photos in the vertical plane A mechanism for directing the virtual agent's gaze based on sound and visual input A robotic manipulator for printing photos Computational model for classification and counting of push-ups Data transmission between two computers using blockchain technology Artificial intelligence and virtual reality in creating interactive 3D environments Automated label quality control using computer vision and a collaborative robot Control of a robot using real-time hand gesture recognition Real-time driver state detection A vision system for controlling an industrial robot with hand movements

Nonverbal communication based adaptive learning model for interactive ai agents

Duration: 2025 - 2030

Programme: Croatian Science Foundation (HRZZ), DOK-2023-10, Career Development of Early-Stage Researchers - Training of New PhD Holders

Coordinator: UNIZAG FSB, LAPIS, DOK-NPOO-2023-10-1790

This project develops an agentic AI model and cognitive architecture enabling computational agents to learn and adapt in real time through multimodal, nonverbal human interaction.

The approach combines online learning, reinforcement learning (RL), and agent-based modelling (ABM) to support continuous updating of internal representations, context-aware reasoning, and adaptive decision-making within a dynamic information space. By maintaining a shared and evolving common ground, the system enables mutual understanding and robust interaction in changing environments.

Building on the AMICORC project, the work advances the PLEA platform with real-time adaptive capabilities, positioning it as a next-generation interactive, learning AI agent.