Frontier stage
Keynote

Coding robots with social awareness using a 3D motion dataset of real people

In person
  • Date
    9 July 2025
    Timeframe
    15:45 - 16:05 CEST
    Duration
    20 minutes
    Share this session

    Examples of the use of AI and Machine learning (ML) in the programming of socially aware robots capable of interacting and collaborating closely with people. By encoding pedestrian etiquette into robot behaviors, more intuitive and safer interactions are achieved, not only with direct users but also with bystanders. The development of what are often called ‘physically intuitive robots’ necessitates incorporating data that reflects social awareness, moving beyond mere optimization. A carefully curated and structured motion capture dataset of real people in authentic environments offers a robust way to capture this socially aware information. Motion capture serves as a valuable foundation for both Physical AI models and heuristic control algorithms developed through ML. Its lightweight nature, sub-millimeter 3D precision, and high frame rate contribute to the safety and subtlety of robot behaviors in real-world, close-proximity human interactions – scenarios that robots might otherwise struggle to navigate.