Change detection criteria for updating sensor-based reference maps

A non-transitory computer-readable storage medium comprising instructions that, when executed, cause a processor of a vehicle system to:

• receive, based on information from a sensor device, an indication that a plurality of registered objects are detected in proximity to the vehicle;
• determine, based on the indication, differences between features of the plurality of registered objects and features of a sensor-based reference map, the features of the sensor-based reference map comprising map locations that correspond to coordinate locations of the plurality of registered objects;
• execute a machine-learned model to determine whether the differences between the features of the plurality of registered objects and features of the sensor-based reference map justify updating the sensor-based reference map by answering contextual questions about the plurality of registered objects and providing rationales justifying answers to the contextual questions;
• responsive to determining that the differences between the features of the plurality of registered objects and features of the sensor-based reference map justify updating the sensor-based reference map, cause the sensor-based reference map to be updated to reduce the differences;
• and
• cause the vehicle to operate in an autonomous mode that relies on the sensor-based reference map for navigating the vehicle in proximity to the coordinate locations of the plurality of registered objects.

A system, the system comprising:

• a processor configured to:
• receive, based on information from a sensor device, an indication that a plurality of registered objects are detected in proximity to the vehicle;
• determine, based on the indication, differences between features of the plurality of registered objects and features of a sensor-based reference map, the features of the sensor-based reference map comprising map locations that correspond to coordinate locations of the plurality of registered objects;
• execute a machine-learned model to determine whether the differences between the features of the plurality of registered objects and features of the sensor-based reference map justify updating the sensor-based reference map by answering contextual questions about the plurality of registered objects and providing rationales justifying answers to the contextual questions;
• responsive to determining that the differences between the features of the plurality of registered objects and features of the sensor-based reference map justify updating the sensor-based reference map, cause the sensor-based reference map to be updated to reduce the differences;
• and
• cause the vehicle to operate in an autonomous mode that relies on the sensor-based reference map for navigating the vehicle in proximity to the coordinate locations of the plurality of registered objects.