Strange Loop

Exploring deep space and deep ocean with AI guidance

There are many challenges associated with data and autonomous decision making in day-to-day life. These are amplified when we explore extreme environments such as the ocean floor, the surface of Mars, or other distant bodies. Scientists and engineers must balance the desire for science return and actionable insights with safety, latency, transmission volume, and environmental factors. Key questions include: What data should be collected? What data is useful for decision making? What data should be returned? What operations should be autonomous?

In some cases, the answer to these questions is simple. Large scale ocean floor exploration, for example, cannot be completed without autonomous/semi-autonomous underwater vehicles. Mars exploration also requires autonomy because the signal latency is between 3 and 22 minutes depending on the orbital positions. Martian rovers move slowly but waiting minutes for each command is not possible. Additional planned missions to distant bodies including Europa, a moon of Jupiter, require advances in instrument awareness and decision making.

In this talk I will discuss several data acquisition techniques including Raman spectroscopy and Laser Induced Breakdown Spectroscopy (LIBS), what we can learn from these techniques, and why they require ML/AI advancement. I will also discuss the role of remote sensing on autonomy for maximizing information return as well as remaining challenges in ML/AI related to extreme environment exploration.

Daniel Van Hoesen

Impossible Sensing

I am a data scientist at Impossible Sensing LLC which is headquartered on Cherokee street in St. Louis. My background is in condensed matter physics with an emphasis on phase transitions. At Impossible Sensing, I apply ML tools to space and other remote sensing data related to spectroscopy and autonomous decision making. Our focus is to understand extreme environments from micro to macro scales to aid scientific advancements including the search for life, resource identification, and to aid in environmental sustainability.