Over the past two decades, computer architecture research has moved from general-purpose computing toward domain-specific accelerator architectures. We believe the field is on the cusp of a new revolution — human-centric architectures — driven by emerging platforms such as Augmented Reality, Virtual Reality, and autonomous machines that all intimately interact with humans: they continuously capture and interpret visual data from humans and generate visual data for humans to consume. These computing platforms must be designed, from the ground up, with principled considerations of human cognition.
Our research is centered around building human-centric visual computing systems, both to obtain unprecedented efficiency guided by human cognition and to augment human cognition through computing technologies. The approach we take is to bridge the conventional computing domain with imaging (i.e., image sensing) and human perception, the two fundamental components that connect computing with humans. The tenet of our work is that a computing problem that seems challenging may become significantly easier when one considers how computing interacts with imaging and human perception in an end-to-end system.