No foolin’- let’s keep up with Astronomical Journal, for April (vol. 169, #4)- both Open Access:
Tang, Y. Jiang, Y. Feng, Y. et al. Transformer-based Approach for Accurate Asteroid Spectra Taxonomy and Albedo Estimation 201 adb710
Wilson, T. C. Silber, E. A. Colston, T. A. et al. Bolide Infrasound Signal Morphology and Yield Estimates: A Case Study of Two Events Detected by a Dense Acoustic Sensor Netw… 223 adbb70
As I’ve mentioned, there’s a lot of hype to Artificial Intelligence… and some reality. Without letting our guard down, consider asteroid characterization. The problems are well-defined, and fairly bounded; there are only so many asteroid types, and basically three of them encompass all the rest as subtypes. Meanwhile, asteroid albedos are known (beforehand) to range from not even four percent, to ~42% at the other extreme. Also consider that, once you know either the spectral type or albedo, the constraints on the ‘other’ (the unknown variable) narrow down greatly. Tang et al. take this rather clear, clean problem and tackle it with (nowadays common) AI algorithms.
The other part of scientific study is gauging the output or result. Meteors are hard to wrangle, being high-altitude and spontaneous and brief. Meteor study, then, is often more like stamp collecting, less like physics. But enough stamps will get you somewhere. If it takes two observing posts to do triangulation, then a network of such posts truly give something to work with. Eventually, a meteor will turn out to be a ‘mega-meteor’ (bolide), giving listening posts at longer ranges enough signal to work with. Wilson et al. have two such bolides, and gauge the accuracy of a meteor sensor network.