Landmark-based localization methods and architectures for an autonomous vehicle

A method for localization of an autonomous vehicle (AV), the method comprising:

• by one or more primary control system processors of a primary control system:
• determining a first predicted location of a landmark in an environment of the AV,
• wherein determining the first predicted location of the landmark is based on a first sensor data instance of first sensor data that is generated by one or more first sensors of the AV;
• determining a second predicted location of the landmark in the environment of the AV,
• wherein determining the second predicted location of the landmark is based on a determined local pose instance of a local pose of the AV and a stored location, of the landmark, in a stored mapping of the environment,
• wherein the determined local pose instance is generated based on a second sensor data instance of second sensor data that is generated by one or more second sensors of the AV, and
• wherein the second sensor data instance used to determine the local pose instance temporally corresponds to the first sensor data instance used to determine the first predicted location of the landmark;
• generating a correction instance based on comparing the first predicted location of the landmark and the second predicted location of the landmark;
• and
• transmitting the correction instance to a secondary control system;
• by one or more secondary control system processors of the secondary control system:
• receiving the correction instance;
• generating an additional local pose instance of the AV based on the correction instance and based on an additional second sensor data instance of the second sensor data;
• and
• causing the AV to be controlled based on the additional local pose instance.

A system for localization of an autonomous vehicle (AV), the system comprising:

• at least one processor;
• and
• at least one memory storing instructions that are operable, when executed by the at least one processor, cause the at least one processor to:
• generate a pose-based predicted location of a landmark relative to the AV,
• wherein the pose-based predicted location of the landmark relative to the AV is generated based on:
• a local pose instance that defines a location of the AV within a mapping of an environment, and
• a stored location of the landmark within the mapping of the environment, and
• wherein the local pose instance is generated without direct utilization of any LIDAR data from a LIDAR sensor of the AV;
• generate a LIDAR-based predicted location of the landmark relative to the AV,
• wherein generating the LIDAR-based predicted location of the landmark relative to the AV is based on identifying, in an instance of LIDAR data that temporally corresponds to the local pose instance, one or more features that are indicative of the landmark;
• generate a correction instance based on comparing the pose-based predicted location and the LIDAR-based predicted location;
• using the correction instance in generating an additional local pose instance, the additional local pose instance being generated without direct utilization of any LIDAR data from the LIDAR sensor;
• and
• cause the AV to be controlled based on the additional local pose instance.