Marco Pardini

Research Groups & Affiliations

Cognitive Humanoid Robotics

Working directly with advanced humanoid platforms, focusing on hyper-realistic control, collision avoidance, and facial retargeting in real-time to create believable interactions.

Abel & Seth

Abel Close-up

Seth Close-up

Abel: Real-time collision avoidance with MoveIt

Seth: Real-time facial retargeting during conversation

Bringing Humanoids to Life

[Placeholder text] In this section, we explore the intricate process of mapping human facial expressions directly onto humanoid platforms like Abel and Seth in real-time. This involves extracting nuanced facial landmarks and translating them into actuator commands that preserve the emotional intent of the user.

[Placeholder text] Additionally, real-time collision avoidance is a critical component for safe human-robot interaction. By utilizing frameworks such as MoveIt, the robots are capable of dynamically replanning their trajectories to avoid obstacles and moving humans, ensuring both safety and fluid motion during tasks.

Affective Computing & Bio-Signals

The BRAVE Project aims to support individuals with Social Anxiety Disorder (SAD). We capture biosignals (ECG, EDA) in immersive VR scenarios to extract features and detect anxiety levels in real-time, bridging the gap toward emotionally-aware cognitive architectures.

Immersive VR scenario with biometric sensor array

Real-time interface: ECG & EDA to anxiety features

Real-time anxiety level detection during the VR experience

Measuring Emotional Responses

[Placeholder text] By integrating electrocardiography (ECG) and electrodermal activity (EDA) sensors, we can reliably monitor physiological changes that correspond to stress and anxiety in real-time. This continuous stream of biometric data is analyzed to provide immediate feedback on the patient's emotional state.

[Placeholder text] The immersive Virtual Reality environments are carefully designed to simulate social situations that might trigger anxiety. This safe, controlled exposure allows researchers to gather highly accurate data and eventually adapt the virtual scenario dynamically based on the user's detected stress levels.

Unreal Engine & MetaHumans

Utilizing Unreal Engine to craft ultra-realistic MetaHumans for simulation, interactive Holobox displays, and digital twin recreations from raw point cloud data.

3-angle split render required for holographic illusion

The final result projected inside the physical Holobox

Holobox Integration

[Placeholder text] Displaying MetaHumans in a physical Holobox requires a complex multi-camera setup within Unreal Engine. The scene must be rendered from multiple specific angles simultaneously to construct the convincing illusion of a 3D holographic presence in the real world.

[Placeholder text] These digital avatars, such as AOI and ADA, are fully interactive. By utilizing real-time rendering capabilities, the avatars can respond to audience inputs with minimal latency, making the holographic experience feel entirely seamless and alive.

Point clouds converted to static meshes

Detailed hair grooming in UE

Philip Mazzei: Fully animated MetaHuman

Thomas Jefferson: Fully animated MetaHuman

Digital Recreations: Mazzei & Jefferson

[Placeholder text] The process of reviving historical figures begins with high-density point clouds. These point clouds are meticulously converted into workable static meshes directly within Unreal Engine, forming the foundational geometry for the characters.

[Placeholder text] Once the base meshes are established, advanced grooming tools are employed to recreate hair, eyebrows, and other fine details. The final step involves rigging these intricate models to the MetaHuman framework, allowing for full body and facial animations that bring historical personalities back to life.

Education

Selected Projects