Schedule

Possibilities exist for robots to greatly improve the quality of our lives and to contribute positively to the safety, creative potential, and atmosphere of public spaces. But as this trend develops, the risk emerges of robots transforming public spaces and social interactions in undesirable ways. This presentation considers how public policy could simultaneously enhance opportunities created by the presence of robots in public spaces and reduce the risks of undesirable outcomes. It draws on previous public policy approaches to harnessing and regulating disruptive technology. Consideration is given to how policy designers might work with robot developers and other stakeholders to guide policies on robots in public spaces. Topics noted include: (1) safety, (2) privacy and ethics, (3) productivity, (4) aesthetics, (5) co-creation, (6) equitable access, and (7) systemic innovation.

When deploying robots into public spaces, or places were they are expected to operate or support untrained and not specifically briefed people, robots usually come with some initial set of (pre-programmed) behaviours, that a designer has envisioned to work well in the setting. However, it is in the nature of such deployments that the behaviour is far from optimal, whatever the robots specific task may be. So, we’d want robots to learn from their experience and improve their task delivery or behaviour in-situ, as part of long-term deployments. In this talk, I’ll present some learning and adaptation models we have deployed onto robots in the public area of a care home and a museum. In total, these robots have been operating for 4 years completely autonomously, and we can show that they indeed get better on the job by exploiting (sometimes noisy) reinforcement signals and carefully trading off exploration vs exploitation of their behaviour models.