Apparatus and method of automatically detecting dynamic object recognition errors in autonomous vehicles

An apparatus for automatically detecting a dynamic object recognition error in an autonomous vehicle, the apparatus comprising:

• a processor;
• a sensor controlled by the processor to obtain sensor data by frame units;
• a parser executed by the processor to parse the sensor data to generate raw data;
• a dynamic object recognition model executed by the processor to analyze the raw data to output a dynamic object detection result;
• an error detector executed by the processor and configured to, when the dynamic object detection result satisfies an error detection condition, determine that detection of a dynamic object recognition error succeeds;
• and
• a non-volatile memory controlled by the processor and configured to, when the detection of the dynamic object recognition error succeeds, storing the raw data and the dynamic object detection result,
• wherein the error detection condition comprises a condition for determining whether a ratio of at least one specific dynamic object to all dynamic objects, detected from the raw data and the dynamic object detection result in M (where M is a natural number of 2 or more) number of frames, is greater than or equal to a reference ratio and a second condition for determining whether a frame ratio of frames including the at least one specific dynamic object, having a confidence score which is less than or equal to a reference confidence score in the M number of frames, is greater than or equal to 50%.