Principal Investigator: Mariella Dimiccoli, ELLIS Unit Barcelona member.
Email: mdimiccoli@iri.upc.edu
Web: www.iri.upc.edu
Brief Theme Description:
Current Foundation Models (FMs) excel at high-dimensional statistical pattern matching, yet they often lack an internal representation of the causal dynamics governing the physical (and social) world. To move toward truly autonomous (and socially intelligent) agents, this research theme focuses on the development of World Models specifically tailored for human behaviour anticipation. Unlike static distributions, these models should represent the spatiotemporal transition functions of human states—predicting how a person will move or interact with objects(/persons) based on video observations. Such capabilities are a cornerstone of human cognition and are essential for the next generation of collaborative and assistive robotics.
This role is ideal for researchers with expertise in machine learning or computational modelling who are eager to advance conceptual innovation toward practical industrial deployment.
Qualifications
- PhD in Computer Science, Machine Learning, Applied Mathematics, or related fields
- Experience in Computer Vision
- Proven problem-solving skills and experience working with complex data systems
Available Infrastructures: Fully equipped Robot Perception and Manipulation Laboratory, see https://www.iri.upc.edu/research/perception#facilities.
Possible Secondments: Other ELLIS Units in Europe, i.e. the University of Freiburg, Germany.
Keywords: World models, Causality, Human Behaviour Anticipation..
