Ingmar Posner Seminar

Abstract: Recent progress in AI technology has been breathtaking. Deep learning has played a central role. However, many of the advances have played to the strengths of virtual environments: infinite training data is available, risk-free exploration is possible, acting is essentially free. In contrast, we require our robots to robustly operate in real-time, to learn from a limited amount of data, take mission- and sometimes safety-critical decisions and increasingly even display a knack for creative problem solving. Cognitive science suggests that, while humans are faced with similar complexity, there are a number of mechanisms which allow us to successfully act and interact in the real world. One prominent example is Dual Process Theory, popularised by Daniel Kahneman’s book Thinking Fast and Slow. While AI and robotics researchers have drawn inspiration from the cognitive sciences pretty much from the outset, I posit that recent advances in machine learning have, for the first time, enabled meaningful parallels to be drawn between AI technology and components identified by Dual Process Theory. This allows us to cast progress in robotics in a new light and leads to a rich set of problem-driven -- yet fundamental -- AI research challenges. Against the backdrop of Dual Process Theory in this talk I will give an overview of recent work by the Applied AI Lab aiming to accelerate the advent of robust, versatile and safe robots of significant value to the public domain.