Method and apparatus for determining an intersection condition
A method, apparatus, and non-transitory computer-readable storage medium for determining an intersection condition are provided. An example embodiment may obtain entity count data of an area comprising an intersection, obtain map feature data of the area comprising the intersection, obtain temporal context data of the area comprising the intersection, and generate an intersection deadlock score based on the entity count data, the map feature data, the temporal context data, or a combination thereof. The generated intersection deadlock score may further be used to provide a prediction of a deadlock condition based on the intersection deadlock score to a traffic application and/or a navigation application.
1 . An apparatus comprising at least one processor and at least one memory including computer program code, the at least one memory and computer program code configured to, with the processor, cause the apparatus to at least:
receive, from at least one sensor data source, entity observation data of an area comprising an intersection of a road network;
obtain, from a map database, map feature data of the area comprising the intersection;
generate, as a function of a machine learning model, an intersection deadlock score based on the entity observation data and the map feature data;
provide a prediction of a deadlock condition based on the intersection deadlock score to a navigation application; and
cause the navigation application based on the prediction to: (1) alter a current path or route of a vehicle, (2) cause a routing engine to penalize a set of road elements to direct a routing algorithm to provide an alternate route, (3) cause the vehicle to initiate a mode change between an autonomous driving mode and a manual driving mode, or (4) a combination thereof.
2 . The apparatus of claim 1 , wherein the entity observation data further comprises an entity count within the area for a time window.
3 . The apparatus of claim 2 , wherein the entity count is further classified based on at least one of vehicular probe data, pedestrian probe data, image-based vehicle count data, or image-based pedestrian count data.
4 . The apparatus of claim 1 , wherein the intersection deadlock score is further based on temporal context data of the area comprising the intersection.
5 . The apparatus of claim 4 , wherein the temporal context data further comprises intersection deadlock scores of neighboring intersections of the intersection.
6 . The apparatus of claim 1 , wherein the map feature data of the area further comprises road feature data of a plurality of road segments connected to the intersection; and wherein the map feature data comprises one of a lane count, functional class, traffic light information, speed limit information, or adjacent POI information for each respective road segment.
7 . The apparatus of claim 1 , wherein the machine learning model is based on at least one generalized intersection deadlock vector.
8 . A method comprising:
receiving, from at least one sensor data source, entity observation data of an area comprising an intersection of a road network;
obtaining temporal context data of the area comprising the intersection;
generating an intersection state vector based on the entity observation data and the temporal context data;
generating, as a function of a machine learning model, an intersection deadlock score based on the intersection state vector;
providing a prediction of a deadlock condition based on the intersection deadlock score to a navigation application; and
causing the navigation application based on the prediction to: (1) alter a current path or route of a vehicle, (2) cause a routing engine to penalize a set of road elements to direct a routing algorithm to provide an alternate route, (3) cause the vehicle to initiate a mode change between an autonomous driving mode and a manual driving mode, or (4) a combination thereof.
9 . The method of claim 8 , wherein the entity observation data further comprises an entity count within the area for a time window.
10 . The method of claim 9 , wherein the entity count is further classified based on at least one of vehicular probe data, pedestrian probe data, image-based vehicle count data, or image-based pedestrian count data.
11 . The method of claim 8 , wherein the temporal context data further comprises intersection deadlock scores of neighboring intersections of the intersection.
12 . The method of claim 8 , further comprising obtaining map feature data of the area comprising the intersection.
13 . The method of claim 12 , wherein the map feature data of the area further comprises road feature data of a plurality of road segments connected to the intersection; and wherein the map feature data comprises one of a lane count, functional class, traffic light information, speed limit information, adjacent POI information, or a combination thereof for each respective road segment.
14 . The method of claim 8 , wherein the machine learning model is based on at least one generalized intersection deadlock vector.
15 . A non-transitory computer-readable storage medium having computer-executable program code instructions stored therein, the computer-executable program code instructions comprising program code instructions to:
obtain, via a program interface, entity count data of an area comprising an intersection of a road network;
obtain, from a map database, map feature data of the area comprising the intersection;
generate, as a function of a machine learning model, an intersection deadlock score based on the entity count data and the map feature data;
provide a prediction of a deadlock condition based on the intersection deadlock score to a traffic application; and
cause the navigation application based on the prediction to: (1) alter a current path or route of a vehicle, (2) cause a routing engine to penalize a set of road elements to direct a routing algorithm to provide an alternate route, (3) cause the vehicle to initiate a mode change between an autonomous driving mode and a manual driving mode, or (4) a combination thereof.
16 . The non-transitory computer-readable storage medium of claim 15 , wherein the entity count is further classified based on vehicle type or pedestrian entity.
17 . The non-transitory computer-readable storage medium of claim 15 , further comprising instructions to:
obtain temporal context data of the area comprising the intersection.
18 . The non-transitory computer-readable storage medium of claim 17 , wherein the temporal context data further comprises intersection deadlock scores of neighboring intersections of the intersection.
19 . The non-transitory computer-readable storage medium of claim 15 , wherein the map feature data of the area further comprises road feature data of a plurality of road segments connected to the intersection; and wherein the map feature data comprises one of a lane count, functional class, traffic light information, speed limit information, or adjacent POI information for each respective road segment.
20 . The non-transitory computer-readable storage medium of claim 15 , wherein the machine learning model is based on at least one generalized intersection deadlock vector.