Politecnico di Torino logo
Staff

Stefano Calzolari

Stefano Calzolari's picture

Ph.D. candidate in Ingegneria Informatica E Dei Sistemi , 39th cycle (2023-2026)
Department of Control and Computer Engineering (DAUIN)

Adjunct lecturer/Adjunct instructor
Department of Control and Computer Engineering (DAUIN)

Profile

PhD

Research topic

Fondamenti per la Credibilità degli Agenti Virtuali

Tutors

Keywords

Computer graphics and Multimedia
Data science, Computer vision and AI
Software engineering and Mobile computing

Biography

I obtained a MSc degree in Cinema and Media Engineering (LM-32) in April 2023 and started my PhD in Computer and Control Engineering in November of the same year. My research group is the Computer Graphics & Vision Group (CG&VG), operating in the fields of Augmented, Mixed and Virtual Reality, Human-Computer Interaction (HMI), User eXperience (UX), Computer Vision, Machine Learning and Artificial Intelligence.

My research focuses on Virtual Humans — artificial entities that looks and act like real human in XR and virtual 3D real-time applications. Virtual Humans are deployed across a wide range of domains including education, training, healthcare, and interactive entertainment. In many of these contexts, their effectiveness is tightly coupled with their believability: the degree to which users perceive them as coherent, responsive, and credible agents rather than scripted automatons.

The central theme of my research is believability as a systemic property of Virtual Humans — not reducible to any single expressive channel, but emerging from the coherent integration of emotional expressivity, cognitive behavior, and interaction capabilities. This framing has evolved over the course of the PhD: early work focused on individual expressive modalities, while subsequent developments — particularly the advent of Large Language Models and agentic AI architectures — have enabled a more integrated approach to agent design.

The research is structured around three interconnected objectives:
Objective 1 — Believable Emotional Expressivity. Studying and evaluating methods for the credible simulation of emotions in virtual agents, including facial expressions via the Facial Action Coding System (FACS) and body language through Text-To-Motion generation techniques. The perceived quality of these modalities is assessed through user studies measuring believability and authenticity.
Objective 2 — Integrated Framework for Believable Virtual Humans in XR. Designing and implementing an agent architecture (A.D.A.M.O.) that integrates multimodal perception, LLM/VLM-based reasoning, and affective-cognitive state modeling into a coherent decision loop. This reduces dependency on manual scripting and increases the agent's generalizability across situated interaction scenarios — both agent-user and agent-environment — in eXtended Reality contexts.
Objective 3 — Believability as an Evaluation Construct. Exploring believability as a multidimensional construct and developing evaluation protocols that combine UX questionnaires, behavioral interaction data (interaction frequency, user trajectories, eye-tracking), and task-based metrics.

The methodology alternates between development and empirical evaluation phases. On the development side, components targeting believability — including emotional expressivity and situated interaction capabilities — are designed and implemented. On the evaluation side, user studies are conducted in XR environments using presence scales, believability questionnaires, and behavioral data collected during interaction.

Teaching

Teachings

Master of Science

MostraNascondi A.A. passati

Bachelor of Science

  • Informatica. A.A. 2025/26, INGEGNERIA AEROSPAZIALE. Collaboratore del corso
  • Informatica. A.A. 2024/25, INGEGNERIA AEROSPAZIALE. Collaboratore del corso
MostraNascondi A.A. passati

Research

Research groups

Publications

Latest publications View all publications in Porto@Iris