Skip to main content

Interaction Lab

The Interaction Lab pioneers advanced interactive AI using deep expertise in Conversational AI, Natural Language Processing, Multimodal Machine Learning, and Human-Robot Interaction. We focus on building safe, robust, and adaptive agents that collaborate with humans across various modalities (speech, vision, gesture, and language). A critical area of our research addresses the bias, fairness, and safety of Generative AI models used in language processing, and in robotics and embodied AI systems.

Our techniques have demonstrated real-world impact in:

  • Conversational Assistants (e.g., Alexa).
  • Healthcare and assistive technologies.
  • Human-Robot Collaboration.

Our evaluation methods using real user trials and simulations has led to notable achievements, including reaching the Finals of Amazon Alexa Challenges (2017, 2018) and the Alexa Prize Simbot competition (2023). We have worked with organisations and companies such as UNICEF, the NHS, PAL robotics, BMW, Honda, and Amazon.

We work on: ​​​​

  • Conversational Artificial Intelligence (AI)
  • Natural Language Processing (NLP)
  • Multimodal Interfaces
  • Human-Robot Interaction (HRI)
  • Spoken Dialogue Systems
  • Natural Language Understanding (NLU)
  • Natural Language Generation (NLG)
  • Data-to-text, summarisation, Question Answering, Open Domain Dialogue
  • Ethical approaches to NLP and conversational AI
  • Trustworthy Interaction
  • Geospatial intelligence (e.g. location based services)
  • AR / VR / MR
  • Grounded and situated interaction, embodied AI