Sky Hub, a new scientific nonprofit, is developing a free, public database with open-source software to collect data and observe unknown origin events occurring in the skies. The pioneering effort is experiencing significant challenges because characterizing unidentified objects by specific features introduces implicit biases. Therefore, the Sky Hub team is attempting to overcome the challenges by developing a sky-bound taxonomy.
The intent behind building the taxonomy is to enable the internal machine learning algorithm to perform automated data sorting. Taxonomic tools and decision trees for identifying unknown objects are relatively commonplace in optical machine learning systems. However, as no such project has been undertaken for such a comprehensive range of objects flying in the sky, building the taxonomy is also a significant challenge for the Sky Hub team.
Nevertheless, the Sky Hub team has developed an initial taxonomy and decision schema. First, using a machine learning approach, the team proposes to characterize familiar objects. Anything left will be by default categorized as the unknown. By becoming better at understanding and categorizing these known objects, it is expected that the open-source software used in Sky Hub will gain a robust and compelling set of unrecognized events.
Furthermore, the initial taxonomy will include the objects and events commonly expected to be captured by Sky Hub units deployed in the field. These will be used to silo known events, based on comparing the observed object's characteristics versus known objects' performance and features. In that case, it will be categorized and filed away.
To develop this taxonomy, it will be necessary to feed the system enough data to assimilate factors such as commercial and military aircraft, animals, bugs, weather phenomena, meteors, satellites, and other common aerial phenomena. The Sky Hub team assumes that taken collectively with the velocity, acceleration ranges, and displacement vector sums, those for a bird will be significantly different from those for an airplane.
Similar analysis methods will be performed for things like the shape of the objects in view, their apparent size and the coloration and light patterns on the objects, and other data points. Secondary data sets such as known flight paths, weather conditions/patterns of interest, and satellite/meteor activity can also be used to refine the set of possible cases which go into the unknown bucket. Over time, other data sets can be included.
In this manner, the Sky Hub team expects to build the sky-bound taxonomy and employ it to train and re-train internal machine learning systems to overcome the aforementioned challenges.
Click here to read the original article published by Sky Hub.
Please give your feedback on this article or share a similar story for publishing by clicking here.
