Methods and systems for detecting road curbs

A system comprising:

• at least one processor;
• and
• a memory having stored thereon instructions that, upon execution by the at least one processor, cause the system to perform functions comprising:
• receiving a plurality of point clouds collected in an incremental order as a vehicle navigates a path, wherein a given point cloud comprises data points representative of the environment of the vehicle at a given timepoint and has associated position information indicative of a position of the vehicle at the given timepoint;
• based on respective associated position information of the plurality of point clouds, processing the plurality of point clouds into a dense point cloud representation;
• identifying, using the dense point cloud representation, given data points having a distribution indicative of a physical structure of a potential road curb;
• determining features using the given data points of the dense point cloud representation for determination of whether the given data points correspond to one or more road curbs in the environment, wherein the features include:
• an elevation profile indicating a height and a slope of data points of the distribution relative to a ground level of the environment;
• an angle of incidence of a collection of the given data points;
• providing the features to a classification system that is configured to output an estimate of whether the features are representative of a road curb;
• and
• based on an output of the classification system, determining whether the given data points represent one or more road curbs in the environment of the vehicle.

A non-transitory computer readable medium having stored thereon instructions that, upon execution by a computing device, cause the computing device to perform functions comprising:

• receiving a plurality of point clouds collected in an incremental order as a vehicle navigates a path, wherein a given point cloud comprises data points representative of the environment of the vehicle at a given timepoint and has associated position information indicative of a position of the vehicle at the given timepoint;
• based on respective associated position information of the plurality of point clouds, processing the plurality of point clouds into a dense point cloud representation;
• identifying, using the dense point cloud representation, given data points having a distribution indicative of a physical structure of a potential road curb;
• determining features using the given data points of the dense point cloud representation for determination of whether the given data points correspond to one or more road curbs in the environment, wherein the features include:
• an elevation profile indicating a height and a slope of data points of the distribution relative to a ground level of the environment;
• and
• an angle of incidence of a collection of the given data points;
• providing the features to a classification system that is configured to output an estimate of whether the features are representative of a road curb;
• and
• based on an output of the classification system, determining whether the given data points represent one or more road curbs in the environment of the vehicle.