Weather and road surface type-based navigation directions

1. A method for generating a set of navigation directions according to road surface types of road segments, the method comprising:

training, by the one or more processors, a machine learning model for identifying road surface types of road segments using (i) a plurality of images of road segments, and (ii) indications of road surface types for each of the plurality of images of road segments;

receiving, at one or more processors, a request for navigation directions for a vehicle from a starting location to a destination location;

identifying, by the one or more processors, a set of candidate routes for navigating from the starting location to the destination location;

for each road segment within each candidate route in the set of candidate routes, determining, by the one or more processors, a road surface type for the road segment, wherein the road surface type for at least one of the road segments is determined by:

applying visual features of the road segment to the trained machine learning model;

selecting, by the one or more processors, a route from the set of candidate routes based at least in part on the road surface types of the road segments within each candidate route; and

providing, by the one or more processors, a set of navigation directions for presentation on a client device for navigating from the starting location to the destination location via the selected route.

2. The method of claim 1 , further comprising:

obtaining, by the one or more processors, weather conditions in a geographic area at least partially including the candidate routes; and

wherein selecting the route from the set of candidate routes comprises selecting, by the one or more processors, the route further based on the weather conditions in the geographic area.

3. The method of claim 2 , wherein:

obtaining, by the one or more processors, weather conditions in the geographic area comprises obtaining weather conditions in the geographic area for a predefined period of time before receiving the request.

4. The method of claim 2 , further comprising:

for each road segment within each candidate route in the set of candidate routes, determining, by the one or more processors, whether the road segment includes an area where precipitation is likely to accumulate;

wherein selecting, by the one or more processors, the route from the set of candidate routes comprises selecting the route further based on which road segments are determined to include an area where precipitation is likely to accumulate.

5. The method of claim 1 , further comprising:

obtaining, by the one or more processors, information on a type of the vehicle; and

wherein selecting the route from the set of candidate routes comprises selecting, by the one or more processors, the route further based on the vehicle type.

6. The method of claim 2 , further comprising, after providing the set of navigation directions for presentation on the client device:

obtaining, by the one or more processors, weather conditions in a second geographic area at least partially including the candidate routes;

selecting, by the one or more processors, an updated route from a current location of the vehicle to the destination location based at least on (i) the road surface types of the road segments within each candidate route and (ii) the weather conditions in the second geographic area; and

providing, by the one or more processors, a set of updated navigation directions for presentation on the client device for navigating from the current location of the vehicle to the destination location via the selected updated route.

7. The method of claim 1 , wherein selecting, by the one or more processors, a route from the set of candidate routes based at least in part on the road surface types of the road segments within each candidate route includes:

for each candidate route:

assigning, by the one or more processors, a score to each road segment within the candidate route based at least in part on the road surface type of the road segment;

combining, by the one or more processors, the scores for each road segment within the candidate route to generate an overall score; and

selecting, by the one or more processors, the route from the set of candidate routes according to the overall score for each respective candidate route.

8. The method of claim 7 , wherein assigning a score to each road segment includes:

assigning, by the one or more processors, a score to each road segment within the candidate route based at least in part on the road surface type of the road segment and on the weather conditions in the geographic area.

9. The method of claim 7 , wherein assigning a score to each road segment includes:

assigning, by the one or more processors, a score to each road segment within the candidate route based on an estimated amount of time for traversing the road segment; and

adjusting, by the one or more processors, the score for each road segment based on the road surface type for the road segment.

10. The method of claim 1 , further comprising:

obtaining, by the one or more processors, an image of a geographic area including the starting location and the destination location;

identifying, by the one or more processors, a set of visual features from the image of the geographic area; and

applying, by the one or more processors, the set of visual features to the machine learning model to identify one or more road surface types and corresponding locations of the road surface types within the geographic area.

11. The method of claim 10 , wherein the set of navigation directions are provided within a map display of the geographic area, the map display represented with vector graphics data, and further comprising:

generating, by the one or more processors, a raster image of the geographic area annotated with identified road surface types at the corresponding locations;

converting, by the one or more processors, the corresponding locations of the road surface types within the raster image to vector graphics data; and

comparing, by the one or more processors, the converted locations represented with vector graphics data to the map display to identify a road segment corresponding to each road surface type within the geographic area.

12. The method of claim 10 , wherein the determined road surface type is at least one of: dirt, asphalt, concrete, grass, gravel, brick, cobblestone, or a bridge underpass.

13. A computing device for generating a set of navigation directions according to road surface types of road segments, the computing device comprising:

one or more processors; and

a non-transitory computer-readable memory coupled to the one or more processors and storing instructions thereon that, when executed by the one or more processors, cause the computing device to:

train a machine learning model for identifying road surface types of road segments using (i) a plurality of images of road segments, and (ii) indications of road surface types for each of the plurality of images of road segments;

receive a request for navigation directions for a vehicle from a starting location to a destination location;

identify a set of candidate routes for navigating from the starting location to the destination location;

for each road segment within each candidate route in the set of candidate routes, determine a road surface type for the road segment, wherein the road surface type for at least one of the road segments is determined by:

applying visual features of the road segment to the trained machine learning model;

select a route from the set of candidate routes based at least in part on the road surface types of the road segments within each candidate route; and

provide a set of navigation directions for navigating from the starting location to the destination location via the selected route.

14. The computing device of claim 13 , wherein the instructions further cause the computing device to:

obtain weather conditions in a geographic area at least partially including the candidate routes; and

select the route from the set of candidate routes based on (i) the road surface types of the road segments within each candidate route and (ii) the weather conditions in the geographic area.