Adaptive cyber-physical system for efficient monitoring of unstructured environments

A non-transitory computer-readable storage medium comprising executable instructions that, when executed by a processing system including at least one processor, perform operations comprising:

• analyzing at least one initial image of a plurality of first images of an area to automatically identify a plurality of target areas, each of which deviates from a determination of an average of the plurality of first images and each of which represents an anomalous place within the area that is identified as an agronomically anomalous target area, the agronomically anomalous target areas collectively forming a set of agronomically anomalous target areas, and wherein the plurality of first images of the area are captured at a first spatial resolution;
• selecting one or more of the agronomically anomalous target areas of the set to form a subset of one or more agronomically anomalous target areas, wherein a number of agronomically anomalous target areas in the set is larger than is in the subset, wherein each agronomically anomalous target area in the subset is selected as a result of that agronomically anomalous target area being larger than at least one agronomically anomalous target area of the set that is excluded from the subset;
• and
• obtaining at least one second image of the one or more agronomically anomalous target areas of the subset that is captured by a camera of an unmanned vehicle as the unmanned vehicle moves along a path, wherein the path is generated according to the selecting of the subset, wherein the path directs the unmanned vehicle to the one or more agronomically anomalous target areas of the subset, and wherein the at least one second image is at a second spatial resolution which is higher than the first spatial resolution.

A method comprising:

• obtaining, by a processing system including at least one processor of an unmanned vehicle, a predetermined path, wherein the predetermined path is determined according to:
• an analysis of at least one initial image of a plurality of first images to automatically identify a plurality of target areas, each of which deviates from a determination of an average of the plurality of first images and each of which identifies an agronomically anomalous target area, wherein the agronomically anomalous target areas collectively form a set of agronomically anomalous target areas, and wherein the plurality of first images are captured at a first spatial resolution;
• and
• a selection of one or more of the agronomically anomalous target areas of the set to form a subset of one or more agronomically anomalous target areas, wherein a number of agronomically anomalous target areas in the set is larger than is in the subset, wherein each agronomically anomalous target area in the subset is selected as a result of that agronomically anomalous target area being larger than at least one agronomically anomalous target area of the set that is excluded from the subset;
• and
• capturing, by the processing system via a camera of the unmanned vehicle, at least one second image of the one or more agronomically anomalous target areas of the subset as the unmanned vehicle is caused to travel along the predetermined path, wherein the at least one second image is at a second spatial resolution which is higher than the first spatial resolution;
• wherein the analysis of the at least one initial image comprises transforming the at least one initial image from an RGB colorspace to an alternative colorspace;
• and
• wherein machine learning algorithms are trained to automatically identify specific causes of anomalies.