Using geometric features and history information to detect features such as car exhaust in point maps

An autonomous vehicle system comprising:

• a light detection and ranging (LIDAR) device;
• and
• a controller configured to:
• generate a point map of a scanning zone of the LIDAR device, the point map comprising a plurality of points;
• generate a density profile of the point map, the density profile indicative of density of one or more regions in the scanning zone characterized at one or more distances;
• generate an elevation profile of the point map, the elevation profile indicative of elevation of one or more features in the scanning zone characterized at one or more distances;
• and
• based on the density profile and the elevation profile, classifying regions of the scanning zone as including a surface of the vehicle or an exhaust gas from the vehicle.

A non-transitory computer readable medium storing instructions that, when executed by one or more processors in a computing device, cause that computing device to perform functions, the functions comprising:

• receiving information indicative of a light detection and ranging (LIDAR) scan of an environment including a vehicle, the LIDAR scan characterized by a scanning zone;
• generating a point map of the scanning zone based on the received information, the point map comprising a plurality of points;
• generating a density profile of the point map, the density profile indicative of density of one or more regions in the scanning zone characterized at one or more distances;
• generating an elevation profile of the point map, the elevation profile indicative of elevation of the one or more features in the scanning zone characterized at one or more distances;
• and
• based on the density profile and the elevation profile, classifying regions of the scanning zone as including a surface of the vehicle or an exhaust gas from the vehicle.