Keynote Speakers

Digital Healthcare harnesses the power of sensors, information and communications technology, signal and data processing, analytics, and machine learning for informed and better decision-making in healthcare. With the emergence of Internet of Things (IoT), wearable and wireless sensing, real-time machine learning algorithms digital healthcare is expected to make a significant impact in the day-to-day lives of many people. It also paves way for telemedicine and mobile health applications. Specific case studies related to mental health and wellness research will also be covered.

A digital twin is a digital replication of a living or non-living physical entity. By bridging the physical and the virtual worlds, data is transmitted seamlessly allowing the virtual entity to exist simultaneously with the physical entity. A digital twin uses AI to facilitate the means to monitor, understand, and optimize the functions of the physical entity and provides continuous feedback to improve quality of life and wellbeing of citizens. A Metaverse is a universe of persistent Digital Twins In this research program, we explore the design and development of frameworks, methodologies, and methods regarding the convergence of multimedia technologies (AR/VR, AI, IoT, Big Data, Cybersecurity and 5G) towards the Metaverse. We will discuss metaverse sensory data fusion and streaming techniques and will explore open questions and possible R&D venues.

The proliferation of low-cost sensors means that they find their way into more and more applications. Their role in industrial, medical, military, aerospace, and transportation are all areas that continue to see dramatic expansion. In some of our work for the Department of Defense, we have merged sensing with the associated processing of those signals in artificial intelligence (AI) driven applications. This evolution of sensors combined with edge computing has created a new normal, which will drive significant efforts to harness the power of AI as close to the point of sensing as possible. This architecture paves the way for further advances in autonomous systems with greater potential for independence.

The rapid progression of high-bandwidth, high-resolution data acquisition methods such as LIDAR and high-definition video streaming, in combination with the advent of ubiquitous AI, has placed significant demands on existing edge hardware. The problem here has three contributing factors – the high volume of data being acquired, the performance demands of AI workloads, and resource constraints of the edge computers. In the past, AI tasks were often offloaded to the cloud or the server away from the edge, to be executed on computing platforms with more resources. However, the latency of data transmission and the volume of data traffic makes it essential for the sensor data to be processed at the edge. Given this landscape, there is a strong motivation to design and develop domain-specific computing accelerators at the edge. Customized accelerators accomplish two objectives – a) they execute AI workloads at the edge, resulting in improved performance since data no longer must be transmitted to a non-edge infrastructure and b) they operate within the design constraints of the edge environment. Existing research indicates that performance to power consumption ratio improves by at least an order of magnitude when customized accelerators are tightly integrated with data acquisition systems. Our preliminary efforts explored the efficacy of commercial off-the-shelf edge devices such as the Nvidia Jetson AGX Orin; ongoing work involves the design of accelerated computing structures targeted at AI models suitable for edge deployment.