Artificial Intelligence (AI) is fundamentally transforming society, and, consequently, the role of computers and electronics. The fundamental purpose of AI applications is to extract information from input data, and they therefore benefit from executing close to their data source. Most contemporary AI applications however do Cloud AI, i.e., they execute in the data centers of the cloud and are far removed from the data they analyze. By moving to Edge AI – in which the system that captures the input data also executes the AI application – we can significantly improve efficiency and, in turn, enable innovative and futuristic applications. Edge AI is technologically challenging because (i) AI methods often require extensive computation, and (ii) the computational capabilities of edge hardware platforms are limited.

Moving key AI applications from the cloud to the edge thus requires technological innovation across a broad front to deliver sufficient computing performance while respecting stringent energy and power consumption requirements and minimizing system cost. The Norwegian Center for Intelligent Computers and Electronics (NICE) will address this challenge by holistically optimizing the fundamental subsystems of Edge AI platforms, thereby providing next-level cost efficiency and, in turn, enabling its industry partners to take leading roles in emerging Edge AI markets. The Research Pillars (RPs) of NICE are long-term research efforts to improve the efficiency of the fundamental Edge AI subsystems, i.e., the sensing, processing, and communication hardware as well as its system software, through prototyping, simulation, and mathematical modeling. To link research to innovation, NICE uses the insights of the RPs to build demonstrators that address real-world industrial and societal challenges in the Applied Partner Themes (APTs).