FIELD: computer engineering.
SUBSTANCE: group of inventions relates to a method and a system for planning the movement of an unmanned vehicle. Vehicle is equipped with multiple sensors. Sensors generate measurement data. Data characterize surrounding space of unmanned vehicle. Car is connected to the processor. Processor is capable of executing a plurality of machine learning models (MLM) for detecting objects in the surrounding space of an unmanned vehicle. MLMs are connected in series to each other. Output of the first MLM from the plurality of MLM serves as the input for the second MLM, which follows the first model in order. Each MLM is trained to detect objects in the surrounding space of the unmanned vehicle on the basis of measurement data characterizing it. Subsequent MLM is capable of detecting objects with a corresponding value of the metric of accuracy of detecting objects. Value of the object detection accuracy metric exceeds any of the values associated with the MLM preceding this next MLM.
EFFECT: enabling increase in the efficiency and accuracy of planning the movement of an unmanned vehicle in conditions of limited computational resources.
20 cl, 11 dwg
