Object tracking based on unused sensor data

A method comprising:

• receiving sensor data collected by one or more sensors of a vehicle, processing the sensor data to identify a plurality of object detections, classify each of the plurality of object detections as a strong object detection or a weak object detection, generating a primary track of the objects using the strong object detections, wherein unused sensor data comprises the weak object detections;
• validating the weak object detections in the unused sensor data based on one or more parameters associated with the objects;
• comparing the weak object detections in the unused sensor data with the strong object detections generating the primary track for duplicates, and remove duplicate weak object detections from the unused sensor data;
• determining kinematics of the objects detected in the unused sensor data based on a geometry of the objects;
• generating a secondary track of the objects from the unused sensor data;
• calculating a probability that one or more of the objects are associated with the secondary track;
• and
• adjusting localization of the vehicle based on one or more of the primary track, the secondary track, and the probability.

A non-transitory computer-readable medium having stored thereon instructions which, when executed by one or more processors, cause the one or more processors to:

• receive sensor data collected by one or more sensors configured to detect objects proximal to a vehicle (AV), process the sensor data to identify a plurality of object detections, classify each of the plurality of object detections as a strong object detection or a weak object detection, generate a primary track of the objects using the strong object detections, and wherein unused sensor data comprises the weak object detections;
• validate the weak object detections in the unused sensor data based on one or more parameters associated with the objects;
• compare the weak object detections in the unused sensor data with the strong object detections generating the primary track for duplicates, and remove duplicate weak object detections from the unused sensor data;
• determine kinematics of the objects associated with the weak object detections in the unused sensor data based on a geometry of the objects;
• generate a secondary track of the objects from the unused sensor data;
• calculate a probability that one or more of the objects are associated with the secondary track;
• and
• adjust localization of the vehicle based on one or more of the primary track, the secondary track, and the probability.