Abstract: Implementations are described herein for optimizing soil sensing spatial selection. In various implementations, object recognition processing may be performed on one or more images captured by a vision sensor carried by a vehicle as the vehicle travels through the agricultural field. Based on the object recognition processing, one or more plants depicted in the one or more images may be identified, and one or more locations of the one or more identified plants relative to the vehicle may be determined. one or more soil sensing targets may be selected on the ground at one or more selected distances from the one or more locations of the one or more identified plants relative to the vehicle. One or more manipulators carried by the vehicle may be operated to perform soil sensing at the one or more soil sensing targets.