Secondary sensor data-based soft constraint optimization for pose graph and mapping refinement

A system comprising:

• one or more processors;
• and
• a memory storing processor-executable instructions that, when executed by the one or more processors, cause the system to perform operations comprising:
• receiving a pose graph determined based at least in part on first sensor data, the pose graph indicating a series of one or more of positions or orientations as a series of pose estimates associated with a vehicle in an environment;
• receiving a first candidate location associated with a second sensor data type and a first pose of the series;
• receiving a second candidate location associated with the second sensor data type and a second pose of the series;
• determining, based at least in part on the first candidate location and the second candidate location, a proposed location based at least in part on determining that the first candidate location and the second candidate location are within a first threshold distance and that a first object detection associated with the first candidate location and a second object detection associated with the second candidate location are associated with a same object classification;
• altering, as an adjusted pose, at least one of the first pose, the second pose, or the proposed location to reduce a distance between the first candidate location, the second candidate location, and the proposed location,
• determining to disassociate the second candidate location from the first candidate location and the proposed location based at least in part on determining that the second candidate location is failing to merge to at least one of the first candidate location or the proposed location;
• determining, based at least in part on the adjusted pose, map data associated with the environment;
• and
• transmitting the map data to the vehicle configured to traverse the environment based at least in part on the map data.

A non-transitory computer-readable medium storing processor-executable instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

• receiving a pose graph determined based at least in part on first sensor data, the pose graph indicating a series of one or more of positions or orientations as a series of pose estimates associated with a vehicle in an environment;
• receiving a first candidate location associated with a second sensor data type and a first pose of the series;
• receiving a second candidate location associated with the second sensor data type and a second pose of the series;
• determining, based at least in part on the first candidate location and the second candidate location, a proposed location;
• altering, as an adjusted pose, at least one of the first pose, the second pose, or the proposed location to reduce a distance between the first candidate location, the second candidate location, and the proposed location,
• determining to disassociate the second candidate location from the first candidate location and the proposed location based at least in part on determining that the second candidate location is failing to merge to at least one of the first candidate location or the proposed location;
• and
• determining, based at least in part on the adjusted pose, map data associated with the environment.