Systems and methods for deriving planned paths for vehicles using path priors
Disclosed herein is technology for deriving a planned path for a vehicle that is operating in a geographic area. In an example embodiment, a computing system associated with the vehicle may function to (i) identify a set of path priors to use as a basis for deriving the planned path for the vehicle in the geographic area, (ii) sample path priors in the identified set of path priors and thereby producing a set of points that is representative of the identified set of path priors, (iii) fit a curve to the set of points produced by the sampling, and (iv) derive the planned path for the vehicle based on the fitted curve.
1. A method comprising:
while a vehicle is operating in a geographic area, identifying a set of path priors to use as a basis for deriving a planned path for the vehicle in the geographic area, wherein the set of path priors corresponds to paths traversed in the geographic area prior to the vehicle operating in the geographic area and include road segments filtered according to trajectory poses;
sampling the road segments individually within the set of path priors to produce a set of points that are representative of the set of path priors, and that includes increasing a number of samples according to a complexity of road geometry associated with the road segments that is associated with reduced safety;
fitting a curve to the set of points;
deriving the planned path for the vehicle based on the curve; and
controlling the vehicle according to the planned path.
2. The method of claim 1 , wherein identifying the set of path priors to use as a basis for deriving the planned path for the vehicle in the geographic area comprises:
selecting a sequence of the road segments that the planned path is to traverse;
identifying a respective subset of path priors associated with each road segment in the sequence of the road segments; and
combining respective subsets of path priors associated with the road segments in the sequence into the set of path priors.
3. The method of claim 1 , wherein fitting the curve to the set of points comprises:
fitting the curve to the set of points to achieve a balance between minimizing a deviation from the set of points and minimizing a curvature of the curve.
4. The method of claim 1 , wherein the curve comprises a cubic smoothing spline.
5. The method of claim 1 , further comprising:
before sampling path priors in the set of path priors, filtering the set of path priors based on a comparison between contextual information related to the path priors in the set of path priors and contextual information related to operation of the vehicle, and wherein sampling the path priors in the set of path priors comprises sampling remaining path priors that remain in the set of path priors after the filtering.
6. The method of claim 5 , wherein filtering the set of path priors comprises:
obtaining data related to the operation of the vehicle for at least one given contextual variable;
for each path prior in the set of path priors:
accessing metadata for the at least one given contextual variable,
comparing the metadata to the data related to the operation of the vehicle for the at least one given contextual variable, and
identifying a path prior for removal if there is a mismatch between the metadata and the data related to the operation of the vehicle for the at least one given contextual variable; and
removing each path prior identified for removal from the set of path priors.
7. The method of claim 6 , wherein the at least one given contextual variable relates to one of time-of-day, date, weather conditions, traffic conditions, school-zone state, or geometry of an applicable sequence of the road segments.
8. The method of claim 1 , wherein sampling the road segments individually within the set of path priors comprises:
using a sampling scheme that involves dynamically selecting a sampling rate to apply to each path prior in the set of path priors based on contextual information related to operation of the vehicle.
9. The method of claim 8 , wherein the contextual information related to the operation of the vehicle comprises one or more of time-of-day, date, weather conditions, traffic conditions, school-zone state, or geometry of an applicable sequence of the road segments.
10. The method of claim 1 , wherein sampling the road segments individually within the set of path priors comprises:
using a sampling scheme that involves dynamically selecting a sampling rate to apply to each respective path prior in the set of path priors based on information related to a respective path prior that comprises one or more of a geometry of a road segment that corresponds to the respective path prior, contextual information for the respective path prior, or a confidence level associated with the respective path prior.
11. The method of claim 1 , wherein identifying the set of path priors comprises defining path priors, and wherein the path priors are encoded into map data for the geographic area.
12. A non-transitory computer-readable medium comprising executable program instructions for:
while a vehicle is operating in a geographic area, identifying a set of path priors to use as a basis for deriving a planned path for the vehicle in the geographic area, wherein the set of path priors corresponds to paths traversed in the geographic area prior to the vehicle operating in the geographic area and include road segments filtered according to trajectory poses;
sampling the road segments individually within the set of path priors to produce a set of points that are representative of the set of path priors, and that includes increasing a number of samples according to a complexity of road geometry associated with the road segments that is associated with reduced safety;
fitting a curve to the set of points;
deriving the planned path for the vehicle based on the curve; and
controlling the vehicle according to the planned path.
13. The non-transitory computer-readable medium of claim 12 , wherein identifying the set of path priors to use as the basis for deriving the planned path for the vehicle in the geographic area comprises:
selecting a sequence of the road segments that the planned path is to traverse;
identifying a respective subset of path priors associated with each road segment in the sequence of the road segments; and
combining respective subsets of path priors associated with the road segments in the sequence into the set of path priors.
14. The non-transitory computer-readable medium of claim 12 , wherein fitting the curve to the set of points comprises:
fitting the curve to the set of points to achieve a balance between minimizing a deviation from the set of points and minimizing a curvature of the curve.
15. The non-transitory computer-readable medium of claim 14 , wherein the curve comprises a cubic smoothing spline.
16. The non-transitory computer-readable medium of claim 12 , further comprising executable program instructions for:
before sampling path priors in the set of path priors, filtering the set of path priors based on a comparison between contextual information related to the path priors in the set of path priors and contextual information related to operation of the vehicle in the geographic area, and wherein sampling the path priors in the set of path priors comprises sampling remaining path priors in the set of path priors after the filtering.
17. The non-transitory computer-readable medium of claim 16 , wherein filtering the set of path priors comprises:
obtaining data related to the operation of the vehicle for at least one given contextual variable;
for each path prior in the set of path priors:
accessing metadata for the at least one given contextual variable,
comparing the metadata to the data related to the operation of the vehicle for the at least one given contextual variable, and
identifying a path prior for removal if there is a mismatch between the metadata and the data related to the operation of the vehicle for the at least one given contextual variable; and
removing each path prior identified for removal from the set of path priors.
18. The non-transitory computer-readable medium of claim 17 , wherein the at least one given contextual variable relates to one of time-of-day, date, weather conditions, traffic conditions, school-zone state, or a geometry of an applicable sequence of the road segments.
19. The non-transitory computer-readable medium of claim 12 , wherein sampling the road segments individually within the set of path priors comprises:
using a sampling scheme that involves dynamically selecting a sampling rate to apply to each respective path prior in the set of path priors based on contextual information related to operation of the vehicle.
20. A computing system comprising:
at least one processor;
a non-transitory computer-readable medium; and
program instructions stored on the non-transitory computer-readable medium that are executable by the at least one processor such that the computing system is capable of:
while a vehicle is operating in a geographic area, identifying a set of path priors to use as a basis for deriving a planned path for the vehicle in the geographic area, wherein the set of path priors corresponds to paths traversed in the geographic area prior to the vehicle operating in the geographic area and include road segments filtered according to trajectory poses;
sampling the road segments individually within the set of path priors to produce a set of points that are representative of the set of path priors, and that includes increasing a number of samples according to a complexity of road geometry associated with the road segments that is associated with reduced safety;
fitting a curve to the set of points;
deriving the planned path for the vehicle based on the curve; and
controlling the vehicle according to the planned path.