KM Talks — Probabilistic Human Intent Recognition

At the United Kingdom National Nuclear Laboratory (KM Talks series) I presented our probabilistic framework for recognising operator intent during mobile manipulation. The talk focused on how dual-...

At the United Kingdom National Nuclear Laboratory (KM Talks series) I presented our probabilistic framework for recognising operator intent during mobile manipulation. The talk focused on how dual-phase inference lets robots anticipate navigational goals and manipulation targets fast enough for collaborative tasks in hazardous settings.

Talk Highlights

  • Positioned human intent prediction as the bridge between high-level mission goals and low-level robot execution in nuclear decommissioning.
  • Shared experimental data from Isaac Sim trials that benchmarked GUIDER against navigation and manipulation baselines.
  • Discussed how uncertainty quantification informs when the robot should request clarification or revert autonomy to the human.

Questions from the Audience

  • Strategies for scaling the approach from simulation to remote operations in shielded facilities.
  • How to fuse operator biometrics or gaze tracking without overloading the communication channel.
  • Opportunities to pair intent prediction with mixed-reality visualisation for remote supervision.

Event Details

  • Date: 25 September 2025
  • Location: United Kingdom (virtual + on-site hybrid)
  • Host: United Kingdom National Nuclear Laboratory — KM Talks series.

Further Reading

  • Browse the full research summary on intent prediction in the GUIDER post.