Method and system of navigating an autonomous vehicle at an intersection of roads
Disclosed subject matter relates to a field of vehicle navigation system that performs a method for navigating an autonomous vehicle at an intersection of roads. An intersection management system may receive sensor data including at least one of depth of an object, images and a video of environment surrounding the autonomous vehicle. Further, traffic police and auxiliary objects associated with each traffic police are detected from plurality of objects of interest present in the images, when the autonomous vehicle is within a predefined distance from an intersection of roads. Thereafter, a correlation matrix comprising inferred data related to each traffic police and the auxiliary objects may be generated. Based on the correlation matrix, the video and the images, a gesture of the traffic police may be determined accurately. Finally, navigation information may be determined based on the correlation matrix and determined gesture, for navigating the autonomous vehicle.
1. A method of navigating an autonomous vehicle at an intersection of roads, the method comprising:
receiving, by a intersection management system configured in an autonomous vehicle, sensor data from one or more sensors configured in the autonomous vehicle, wherein the sensor data comprises at least one of depth of an object with respect to the autonomous vehicle, and one or more images and a video of environment surrounding the autonomous vehicle;
detecting, by the intersection management system, one or more traffic police and one or more auxiliary objects associated with each of the one or more traffic police from a plurality of objects of interest present in the one or more images, when the autonomous vehicle is within a predefined distance from an intersection of roads;
generating, by the intersection management system, a correlation matrix comprising inferred data related to each of the one or more traffic police and the one or more auxiliary objects associated with the corresponding one or more traffic police;
determining, by the intersection management system, a gesture of the one or more traffic police based on the correlation matrix, and at least one of the video and the one or more images; and
determining, by the intersection management system, navigation information for the autonomous vehicle based on the correlation matrix and the determined gesture, for navigating the autonomous vehicle.
2. The method as claimed in claim 1 , wherein the inferred data comprises at least one of an identifier of each of the one or more traffic police, location of each of the one or more traffic police with respect to the autonomous vehicle, indication of presence or absence of the one or more auxiliary objects, type of each of the one or more auxiliary objects, and values of each of the one or more auxiliary objects associated with each of the one or more traffic police, or position of each of the one or more auxiliary objects with respect to the corresponding one or more traffic police.
3. The method as claimed in claim 1 , wherein determining the gesture comprises:
determining a state of each of the one or more traffic police, wherein the state of each of the one or more traffic police is one of “idle” or “active”;
detecting presence of the one or more auxiliary objects associated with each of the one or more traffic police whose state is determined to be “active”, based on the correlation matrix;
selecting an auxiliary object among the one or more auxiliary objects associated with each of the active traffic police, based on at least one of a position, visibility or a type of each of the one or more auxiliary objects; and
inferring a value of the selected auxiliary object as the gesture of the corresponding active traffic police.
4. The method as claimed in claim 3 , wherein, when, at least one of,
the one or more auxiliary objects are absent,
or
the presence of one or more auxiliary objects is detected, and value of the detected one or more auxiliary objects is indeterminable,
the method comprises determining, by the intersection management system,
a dynamic hand gesture of the active traffic police based on at least one of a video and one or more images of environment surrounding an autonomous vehicle, using one or more Convolutional Neural Network (CNN) techniques.
5. The method as claimed in claim 1 , wherein determining the navigation information comprises:
determining a candidate traffic police, proximal to the autonomous vehicle, from the one or more traffic police whose state is determined to be “active”, based on a location of each of the one or more traffic police with respect to the autonomous vehicle; and
determining the navigation information for the autonomous vehicle based on the gesture of the candidate traffic police.
6. An intersection management system for navigating an autonomous vehicle at an intersection of roads, the intersection management system comprises:
a processor; and
a memory communicatively coupled to the processor, wherein the memory stores the processor-executable instructions, which, on execution, causes the processor to:
receive sensor data from one or more sensors configured in the autonomous vehicle, wherein the sensor data comprises at least one of depth of an object with respect to the autonomous vehicle, and one or more images and a video of environment surrounding the autonomous vehicle;
detect one or more traffic police and one or more auxiliary objects associated with each of the one or more traffic police from a plurality of objects of interest present in the one or more images, when the autonomous vehicle is within a predefined distance from an intersection of roads;
generate a correlation matrix comprising inferred data related to each of the one or more traffic police and the one or more auxiliary objects associated with the corresponding one or more traffic police;
determine a gesture of the one or more traffic police based on the correlation matrix, and at least one of the video and the one or more images; and
determine navigation information for the autonomous vehicle based on the correlation matrix and the determined gesture, for navigating the autonomous vehicle.
7. The intersection management system as claimed in claim 6 , wherein the inferred data comprises at least one of an identifier of each of the one or more traffic police, location of each of the one or more traffic police with respect to the autonomous vehicle, indication of presence or absence of the one or more auxiliary objects, type of each of the one or more auxiliary objects, and values of each of the one or more auxiliary objects associated with each of the one or more traffic police, or position of each of the one or more auxiliary objects with respect to the corresponding one or more traffic police.
8. The intersection management system as claimed in claim 6 , wherein to determine the gesture, the processor is configured to:
determine a state of each of the one or more traffic police, wherein the state of each of the one or more traffic police is one of “idle” or “active”;
detect presence of the one or more auxiliary objects associated with each of the one or more traffic police whose state is determined to be “active”, based on the correlation matrix;
select an auxiliary object among the one or more auxiliary objects associated with each of the active traffic police, based on at least one of a position, visibility or a type of each of the one or more auxiliary objects; and
infer a value of the selected auxiliary object as the gesture of the corresponding active traffic police.
9. The intersection management system as claimed in claim 8 , wherein, when, at least one of,
the one or more auxiliary objects are absent,
or
the presence of one or more auxiliary objects is detected, and value of the detected one or more auxiliary objects is indeterminable,
the processor is configured to determine a dynamic hand gesture of the active traffic police based on at least one of a video and one or more images of environment surrounding an autonomous vehicle, using one or more Convolutional Neural Network (CNN) techniques.
10. The intersection management system as claimed in claim 6 , wherein to determine the navigation information, the processor is configured to:
determine a candidate traffic police, proximal to the autonomous vehicle, from the one or more traffic police whose state is determined to be “active”, based on a location of each of the one or more traffic police with respect to the autonomous vehicle; and
determine the navigation information for the autonomous vehicle based on the gesture of the candidate traffic police.
11. A non-transitory computer readable medium including instructions stored thereon that when processed by at least one processor causes an intersection management system to perform operations comprising:
receiving sensor data from one or more sensors configured in the autonomous vehicle, wherein the sensor data comprises at least one of depth of an object with respect to the autonomous vehicle, and one or more images and a video of environment surrounding the autonomous vehicle;
detecting one or more traffic police and one or more auxiliary objects associated with each of the one or more traffic police from a plurality of objects of interest present in the one or more images, when the autonomous vehicle is within a predefined distance from an intersection of roads;
generating a correlation matrix comprising inferred data related to each of the one or more traffic police and the one or more auxiliary objects associated with the corresponding one or more traffic police;
determining a gesture of the one or more traffic police based on the correlation matrix, and at least one of the video and the one or more images; and
determining navigation information for the autonomous vehicle based on the correlation matrix and the determined gesture, for navigating the autonomous vehicle.