Systems and methods for modeling pedestrian activity

1. A method comprising:

receiving images of a plurality of road portions;

receiving data relating to pedestrian activity at each of the road portions;

for each of the road portions, determining whether a number of pedestrians that cross a street within the road portion outside of a crosswalk during a predetermined period of time is greater than a predetermined threshold;

labeling each image of a road portion for which the number of pedestrians that cross the street during the predetermined period of time exceeds the predetermined threshold as containing a constructive crosswalk;

labeling each image of a road portion for which the number of pedestrians that cross the street during the predetermined period of time does not exceed the predetermined threshold as not containing a constructive crosswalk;

using the labeled images to train a neural network to receive an input of a road portion and determine a likelihood that the road portion contains a constructive crosswalk where pedestrians are likely to cross a street within the road portion outside of a crosswalk;

capturing an image of a sample road portion with one or more vehicle sensors of an autonomous vehicle;

inputting the image into the trained neural network;

determining a likelihood that one or more locations of the road portion comprise a constructive crosswalk based at least in part on an output of the trained neural network; and

controlling operation of the autonomous vehicle based on the determined likelihood that the one or more locations of the road portion comprise a constructive crosswalk.

2. The method of claim 1 , further comprising determining a pedestrian profile associated with each of the road portions labeled as a constructive crosswalk, the pedestrian profile comprising a most probable path for pedestrians to cross the street at the location of the constructive crosswalk.

3. The method of claim 1 , further comprising determining a pedestrian profile associated with each of the road portions labeled as a constructive crosswalk, the pedestrian profile comprising an average velocity at which pedestrians cross the street at the location of the constructive crosswalk.

4. The method of claim 1 , further comprising determining a pedestrian profile associated with each of the road portions labeled as a constructive crosswalk, the pedestrian profile comprising an average density at which pedestrians cross the street at the location of the constructive crosswalk.

5. The method of claim 1 , further comprising determining a pedestrian profile associated with each of the road portions labeled as a constructive crosswalk, the pedestrian profile comprising a likelihood of pedestrians ceding a right of way to vehicles when crossing the street at the location of the constructive crosswalk.

6. A computing device comprising:

one or more processors;

one or more memory modules; and

machine readable instructions stored in the one or more memory modules that, when executed by the one or more processors, cause the computing device to:

receive data relating to pedestrian activity at each of the road portions;

for each of the road portions, determine whether a number of pedestrians that cross a street within the road portion outside of a crosswalk during a predetermined period of time is greater than a predetermined threshold;

label each image of a road portion for which the number of pedestrians that cross the street during the predetermined period of time exceeds the predetermined threshold as containing a constructive crosswalk;

label each image of a road portion for which the number of pedestrians that cross the street during the predetermined period of time does not exceed the predetermined threshold as not containing a constructive crosswalk;

use the labeled images to train a neural network to receive an input of a road portion and determine a likelihood that the road portion contains a constructive crosswalk where pedestrians are likely to cross a street within the road portion outside of a crosswalk;

capturing an image of a sample road portion with one or more vehicle sensors of an autonomous vehicle;

input the image into the trained neural network;

determine a likelihood that one or more locations of the road portion comprise a constructive crosswalk based at least in part on an output of the trained neural network; and

control operation of the autonomous vehicle based on the determined likelihood that the one or more locations of the road portion comprise a constructive crosswalk.

7. The computing device of claim 6 , wherein the one or more processors further cause the computing device to determine a pedestrian profile associated with each of the road portions labeled as a constructive crosswalk, the pedestrian profile comprising a most probable path for pedestrians to cross the street at the location of the constructive crosswalk.

8. The computing device of claim 6 , wherein the one or more processors further cause the computing device to determine a pedestrian profile associated with each of the road portions labeled as a constructive crosswalk, the pedestrian profile comprising an average velocity at which pedestrians cross the street at the location of the constructive crosswalk.

9. The computing device of claim 6 , wherein the one or more processors further cause the computing device to determine a pedestrian profile associated with each of the road portions labeled as a constructive crosswalk, the pedestrian profile comprising an average density at which pedestrians cross the street at the location of the constructive crosswalk.

10. The computing device of claim 6 , wherein the one or more processors further cause the computing device to determine a pedestrian profile associated with each of the road portions labeled as a constructive crosswalk, the pedestrian profile comprising a likelihood of pedestrians ceding a right of way to vehicles when crossing the street at the location of the constructive crosswalk.