Research

The Embedded AI Systems Lab studies how intelligent systems behave in the real world where constraints such as hardware limitations, embodiment, and human interaction fundamentally shape AI performance and perception.

Our work sits at the intersection of AI systems, Human-Computer Interaction (HCI), Virtual Reality (VR), and embedded computing, with an emphasis on building, deploying, and evaluating systems beyond idealized settings.

AI + Human Experience (HCI & VR)

AI Agents in Immersive Environments

We investigate how AI-driven agents such as LLM-based companions influence user experience in virtual environments. Our work explores questions of trust, engagement, perceived safety, and meaning-making, particularly in multi-phase and dynamic scenarios such as VR simulations.

AI for Training and Education in VR

We develop VR-based educational systems augmented with AI assistants, mostly focusing on educational domains. These systems aim to make complex concepts more accessible through interactive, adaptive instruction.

AI in Therapeutic and Behavioral Contexts

In collaboration with the Psychology Department, we design and evaluate VR exposure therapy environments enhanced with AI agents, studying how adaptive AI influences emotional response and therapeutic outcomes.

Embedded and Resource-Constrained AI

Efficient AI on Edge and Embedded Systems

We study how to deploy modern AI models under strict resource constraints (latency, energy, memory). Our work includes techniques such as quantization, pruning, and system-level optimization for real-world embedded platforms.

Hardware-Aware AI and Approximation

We explore how approximation at the hardware and model level affects not only performance and efficiency, but also downstream behavior, raising new questions about robustness, fairness, and user perception.

Cross-Cutting Themes

Across all projects, we are particularly interested in:

  • How constraints (hardware, environment, embodiment) shape AI behavior
  • How humans perceive, trust, and interact with AI systems
  • Bridging the gap between theoretical AI performance and real-world deployment

For more information about our projects or to get involved, please contact Dr. Spantidi.