Observe
Learn meaningful signals from video, audio, gaze, motion, and other modalities across people, time, and viewpoints.
- Multimodal perception
- Gaze and human motion
- Behavioural datasets
From perceiving behaviour to understanding it in context
Human behaviour is dynamic and shaped by personal, interpersonal, situational, and social factors. My work connects multimodal perception, structured knowledge, and explicit reasoning to capture that complexity.
Reason
Join statistical learning with behavioural science knowledge and human-readable concepts for context-aware inference.
- Hybrid and neuro-symbolic AI
- Contextual reasoning
- Explainable inference
Interact
Translate understanding into adaptive, socially appropriate behaviour for embodied and virtual intelligent systems.
- Social robotics
- Virtual agents and XR
- Responsible social AI
ECASIS
Active · 2025 to 2030Royal Academy of Engineering
Engineering Context-Aware Socially Intelligent Systems with Hybrid AI
ECASIS develops a holistic computational framework for social AI. It augments sensory data with behavioural science knowledge, combines machine learning with explicit reasoning, and jointly models the factors that shape behaviour.
01 Perceive multimodal behaviour02 Reason with knowledge03 Interact responsibly