Sensing and Robotics Science Co-Pilot: Developing Technology for Next-Gen Field Surveys

Scientific exploration is rapidly changing, with an emphasis on real-time, precise data derived from field studies. There's a growing trend towards tools that allow us to engage not only with digital representations of the physical world, but also with higher-level insights derived from the data. This talk will delve into the world of AI-powered sensing devices that integrate seamlessly into the physical environment. We will explore the diversity of embedded ML algorithms that have been developed over the years, driven by the massive 6 orders of magnitude difference in energy efficiency of the underlying compute and sensing hardware. Presented in the context of case studies and applications, we'll discuss how these tech breakthroughs can optimize data collection, increase data accuracy and scale, and improve timeliness of data insights. We'll conclude the talk touching on recent generative AI models that offer a potential step-change in how domain users interact with advanced technology. Generalizable real-world exploration tools based on advanced sensing, robotics, and AI technologies may well become a key component of the scientific toolkit in near future.