Using radar data for automatic generation of machine learning training data and localization

A non-transitory machine-readable storage medium having computer-executable instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

• identifying, by a processor, a set of radar data captured by at least one autonomous vehicle when the at least one autonomous vehicle was positioned in a lane of a roadway, and respective ground truth localization data of the at least one autonomous vehicle;
• filtering, by the processor, one or more objects identified via the set of radar data in accordance with an attribute of the one or more objects by:
• determining at least one of the one or more objects is moving based on the set of radar data;
• and
• filtering the set of radar data into a set of filtered radar data by excluding the at least one of the one or more objects from the set of radar data;
• generating, by the processor, a plurality of map layers to be included within a high definition map, based on the set of filtered radar data and the respective ground truth localization data of the at least one autonomous vehicle, generating the plurality of map layers further comprising segregating different characteristics of the one or more objects into separate layers in the plurality of map layers, wherein each map layer of the plurality of map layers is stored in a separate dataset;
• identifying, by the processor, a second set of radar data captured by a second autonomous vehicle when the second autonomous vehicle was positioned in the lane of the roadway;
• and
• localizing, by the processor, the second autonomous vehicle by executing a matching protocol to match an object within the second set of radar data with an object within the plurality of map layers.

A system comprising a processor configured to:

• identify, by a processor, a set of radar data captured by at least one autonomous vehicle when the at least one autonomous vehicle was positioned in a lane of a roadway, and respective ground truth localization data of the at least one autonomous vehicle;
• filter, by the processor, one or more objects identified via the set of radar data in accordance with an attribute of the one or more objects by:
• determining at least one first object of the one or more objects is moving based on the set of radar data;
• and
• filtering the set of radar data into a set of filtered radar data by excluding the at least one first object of the one or more objects from the set of radar data;
• generate, by the processor, a plurality of map layers to be included within a high definition map, based on the set of filtered radar data and the respective ground truth localization data of the at least one autonomous vehicle, generating the plurality of map layers further comprising segregating different characteristics of the one or more objects into separate layers in the plurality of map layers, wherein each map layer of the plurality of map layers is stored in a separate dataset;
• identify, by the processor, a second set of radar data captured by a second autonomous vehicle when the second autonomous vehicle was positioned in the lane of the roadway;
• and
• localize, by the processor, the second autonomous vehicle by executing a matching protocol to match an object within the second set of radar data with an object within the plurality of map layers.