Sidewalk detection for pedestrian behavior modeling
Aspects of the disclosure relate to detecting sidewalks adjacent to roads. In this regard, a set of potential sidewalk areas adjacent to one or more roads in a vehicle's vicinity may be determined based on map data. Topology data for the set of potential sidewalk areas may be generated based on sensor data received from a perception system of the vehicle. The set of potential sidewalks may be filtered to remove areas unlikely to include a sidewalk. The vehicle may be operated based on the filtered set of potential sidewalk areas, which may include taking precautionary measures when within a predetermined distance from any of the filtered set of potential sidewalks.
1. A method for determining sidewalks, the method comprising:
identifying, by one or more computing devices based on map data corresponding to one or more roads in a vicinity of a vehicle, a set of potential sidewalk areas adjacent to the one or more roads;
generating, by the one or more computing devices, topology data for the set of potential sidewalk areas based on sensor data received from a perception system of the vehicle;
detecting, by the one or more computing devices, one or more objects in one or more of the potential sidewalk areas that does not provide at least a threshold clearance in the one or more potential sidewalk areas through which a pedestrian can pass;
filtering, by the one or more computing devices, the set of potential sidewalk areas to remove those potential sidewalk areas that include the one or more objects; and
operating, by the one or more computing devices, the vehicle in an autonomous driving mode based on the filtered set of potential sidewalk areas.
2. The method of claim 1 , wherein the set of potential sidewalk areas are areas within a predetermined distance from the one or more roads.
3. The method of claim 1 , further comprising:
determining, by the one or more computing devices based on the topology data, one or more slopes for each potential sidewalk area of the set of potential sidewalk areas, wherein the filtering is based on whether the one or more slopes meeting a threshold slope.
4. The method of claim 1 , further comprising:
applying, by the one or more computing devices, a morphological filter to the topology data of the set of potential sidewalk areas by using a three dimensional pedestrian model, wherein the filtering is based on results of the morphological filter.
5. The method of claim 1 , further comprising:
determining, by the one or more computing devices based on the sensor data, visual features for each potential sidewalk area of the set of potential sidewalk areas, wherein the filtering is based on whether each potential sidewalk area of the set of potential sidewalk areas includes visual features corresponding to a material for constructing sidewalks.
6. The method of claim 1 , further comprising:
applying, by the one or more computing devices, an image filter to the sensor data of the set of potential sidewalk areas, the image filter including one or more criteria corresponding to visual features of one or more materials for constructing sidewalks, wherein the filtering is based on the results of the image filter.
7. The method of claim 6 , wherein the image filter includes a color filter.
8. The method of claim 6 , wherein the image filter includes a threshold brightness intensity level.
9. The method of claim 6 , wherein the one or more criteria of the image filter are adjusted based on an intensity and a color of an ambient light.
10. The method of claim 6 , wherein the one or more criteria of the image filter are adjusted based on a weather condition.
11. The method of claim 1 , further comprising:
determining, by the one or more computing devices, whether one or more remaining areas in the filtered set of potential sidewalk areas form a connected pathway;
determining, by the one or more computing devices, whether the connected pathway runs parallel to an adjacent road; and
categorizing, by the one or more computing devices based on the determination that the connected pathway runs parallel to an adjacent road, the connected pathway as a sidewalk.
12. The method of claim 1 , wherein operating the vehicle based on the filtered set of potential sidewalk areas includes controlling the vehicle to drive below a threshold speed when the vehicle is within a predetermined distance from the filtered set of potential sidewalk areas.
13. The method of claim 1 , wherein operating the vehicle based on the filtered set of potential sidewalk areas includes controlling the vehicle to maintain a threshold clearance distance from the filtered set of potential sidewalk areas.
14. The method of claim 1 , further comprising:
using, by the one or more computing devices, the filtered set of potential sidewalk areas as input for a pedestrian behavior model, wherein the pedestrian behavior model is configured to predict a trajectory of a detected pedestrian based on the filtered set of potential sidewalk areas.
15. The method of claim 1 , further comprising:
labeling the filtered set of potential sidewalk areas in a set of sensor data; and
sending the set of sensor data to a remote server for training a pedestrian behavior model to predict trajectories of detected pedestrians based on locations of sidewalks.
16. The method of claim 1 , further comprising:
receiving, by the one or more computing devices, historical data including past detections of pedestrians on the set of potential sidewalk areas, wherein the filtering is based on the past detections of pedestrians not meeting a threshold number.
17. A system, comprising:
one or more computing devices configured to:
identify, based on map data corresponding to one or more roads in a vicinity of a vehicle, a set of potential sidewalk areas adjacent to the one or more roads;
generate topology data for the set of potential sidewalk areas based on sensor data received from a perception system of the vehicle;
detect one or more objects in one or more of the potential sidewalk areas that does not provide at least a threshold clearance in the one or more of the potential sidewalk areas through which a pedestrian can pass;
filter the set of potential sidewalk areas to remove those potential sidewalk areas that include the one or more objects; and
operate the vehicle in an autonomous driving mode based on the filtered set of potential sidewalk areas.
18. The system of claim 17 , wherein the one or more computing devices are further configured to:
apply a morphological filter to the topology data of the set of potential sidewalk areas, the morphological filter using a three dimensional pedestrian model, wherein the filtering is based on results of the morphological filter.
19. The system of claim 17 , wherein the one or more computing devices are further configured to:
apply an image filter to the sensor data of the set of potential sidewalk areas, the image filter includes one or more criteria corresponding to visual features of one or more materials for constructing sidewalks, wherein the filtering is based on the results of the image filter.