AI meets IoT - overview
We try to improve human environments and business processes by bringing artificial intelligence (AI) into the environment with the help of the Internet of Things (IoT). What does this mean? Let us look at a few projects we are working on.
Indoor location is an important context variable for enabling a variety of very useful services, such as seat utilization estimation in workplaces and location-aware mobile services. GPS-based positioning does not work well inside buildings. While one could intrument the indoor environment with sensors and cameras for this purpose, we take a different approach. We look at software-based approaches to localization that exploit existing IoT infrastructure to infer location. For this we develop machine learning approaches that infer location based on information from networking infrastructure (WiFi, ethernet), energy data and user devices (smart phones, laptops). We then apply these approaches in end-to-end solutions for helping business optimize their space utilization and helping occupants access location-optimized services on their devices.
Modeling the human
One of the key anchors for making IoT and AI work for the human, is for the computing system to understand the physical appearance of the human. We work on image processing and deep learning algorithms and systems that can extract relevant attributes from images or videos of humans, for example facial features, neck features, and full body measurements. These human understanding models are then employed in solutions such as a cognitive mirror for fashion retail to empower humans to exploit AI and IoT in their natural environments to improve their experience.