Keynote
In person
LeadersGoldDiscovery

Embodied intelligence for safety and well-being

  • Date
    7 July 2026
    Timeframe
    11:30 - 11:45
    Duration
    15 minutes
    • Days
      Hours
      Min
      Sec

    Artificial intelligence has made remarkable progress through large-scale data and simulation, yet most systems remain fundamentally disembodied, learning to mimic rather than to truly understand the physical world. This limitation becomes critical when robots are expected to operate safely alongside humans, where interaction is not optional but essential.

    In this talk, I argue that safe and beneficial AI requires a shift toward embodied intelligence: systems that learn through physical interaction, sensing, and adaptation. Touch, in particular, plays a central role. It provides the feedback needed to build causal models of the world, enabling robots to act robustly in uncertain, dynamic environments.

    I will present our work on equipping robots with large-area tactile sensing (artificial skin), spatial awareness, soft compliant bodies, and sensorimotor control strategies that allow safe contact-rich behavior. Rather than avoiding interaction, these systems leverage touch to navigate clutter, manipulate objects, and engage safely with people.

    Finally, I will discuss applications in domains such as occupational therapy training, where safe physical interaction is key. By designing robots that can feel and adapt, we move toward AI systems that are not only more capable, but fundamentally aligned with human safety and well-being.

    Are you sure you want to remove this speaker?