Determining portions of a roadway model requiring updating

A computer-implemented method for determining which portions of a roadway model representing a real-world roadway require updating, the method comprising:

• maintaining a roadway model of the real-world roadway, the roadway model being sufficient to enable autonomous driverless operation of a self-driving road vehicle;
• receiving, from a plurality of self-driving road vehicles on the roadway, from at least one sensor carried by each self-driving road vehicle, sensor data associated with known positions on the roadway, the sensor data representing at least one feature of the roadway corresponding to a model feature in the roadway model;
• using the sensor data to determine discrepancy data;
• aggregating the discrepancy data into aggregated discrepancy data, wherein the aggregated discrepancy data comprises, for at least some portions of the roadway, discrepancy data from a plurality of individual self-driving road vehicles;
• using the aggregated discrepancy data to identify, as the portions of the roadway model which require updating, those portions of the roadway model corresponding to portions of the roadway for which the aggregated discrepancy data exceeds a threshold;
• and
• responsive to identifying the portions of the roadway model which require updating, doing at least one of:
• dispatching one or more survey vehicles to the portions of the roadway for which the aggregated discrepancy data exceeds the threshold;
• and
• modifying an existing plan for updating the roadway model to include the portions of the roadway for which the aggregated discrepancy data exceeds the threshold.