King's College London

Recently there has been a great interest in developing robots and artificially intelligent systems that interact with humans in a more naturalistic ways. In these projects we draw on our studies of workplaces and settings like museums and galleries to inform the design, analysis and deployment of such systems. In particular we draw on studies of talk and social interaction to suggest the embodied capabilities in which these technologies might better ‘interact’ with humans. We develop quasi-naturalistic experiments to assess these new forms of human-system interaction, for example by simulating cases where in museums and galleries robots act as museum guides.

These analyses have been undertaken with researchers in computer science and robotics and engineering and have sought to accomplish a range of activities that include

• Robots that navigate around a space and can be used to identify objects in that space for remote participants,
• Robots that produce features of embodied conduct (e.g. body, head and gaze orientation) when interacting with human participants;
• Systems that plan routes and navigate around remote locations.

As well as informing the design and development of robots and artificial intelligence systems such as planners, we also seek to support innovation in the way in which these innovative technologies are assessed. By serving as challenging settings as in a way as ‘breaching experiments’ these 'technical interventions’ can inform our understanding everyday work and social interaction.

We currently are undertaking studies related to the THuMP and Trusted Autonomous Systems projects that explore explanation in planning systems and trust and teamwork in robot assisted surgery and the abilities of autonomous vehicles to access, explain and understand data related to failure or accidents.