An event scheduling system for collecting event-related image data.

Abstract

This disclosure relates to an event scheduling system for collecting image data related to one or more events using one or more autonomous vehicles. An event scheduling system for collecting image data related to one or more events using one or more autonomous vehicles includes a centralized scheduling system that wirelessly communicates with the one or more autonomous vehicles. Each autonomous vehicle collects image data related to the one or more events while following a unique driving plan. The centralized scheduling system executes instructions to determine a unique driving plan for a particular autonomous vehicle, wherein the unique driving plan guides the particular autonomous vehicle to a specific location of a filtered event to collect image data.

Description

Technical Field

[0001] This disclosure relates to an event scheduling system for collecting image data related to one or more events using one or more autonomous vehicles, wherein the event scheduling system determines a unique driving plan for a particular autonomous vehicle. The autonomous vehicles collect image data related to one or more events while following their unique driving plans. The unique driving plan is determined based on a dynamic programming scheduling method or a greedy algorithm method. Background Technology

[0002] Autonomous vehicles can employ various technologies to collect sensor information to detect their surroundings, including but not limited to radar, laser, GPS, and cameras. In particular, autonomous vehicles use multiple cameras to extract 3D data about objects in their surroundings. However, cameras can sometimes encounter problems that prevent them from seeing some aspects of the autonomous vehicle's environment. For example, one possible problem is visual occlusion, which occurs when the features of an object are obscured by other subjects. In another example, the autonomous vehicle's cameras may malfunction and therefore fail to collect image data about surrounding objects. This problem can be further complicated when an event such as a traffic accident occurs in the surrounding environment and the cameras are unable to see the accident.

[0003] Therefore, while current vehicles have achieved their intended purpose, there is a need in the art for an improved method for collecting image data through autonomous vehicles. Summary of the Invention

[0004] According to several aspects, an event scheduling system for collecting image data related to one or more events via one or more autonomous vehicles is disclosed. The event scheduling system includes a centralized scheduling system that wirelessly communicates with one or more autonomous vehicles, wherein each autonomous vehicle collects image data related to one or more events while following a unique driving plan. The centralized scheduling system executes instructions to receive one or more notifications indicating that an event has occurred and creates an event pool storing the one or more events. The centralized scheduling system executes instructions to compare a predetermined route corresponding to a particular autonomous vehicle with a specific location corresponding to each event stored in the event pool to identify one or more filtered events. The centralized scheduling system executes instructions to determine the presence of a particular autonomous vehicle when a filtered event occurs based on the predetermined route and the specific location of the filtered event. The centralized scheduling system executes instructions to identify matching pairs, which include the filtered event and the predetermined route of the particular autonomous vehicle. Finally, the centralized scheduling system executes instructions to determine a unique driving plan for the particular autonomous vehicle based on the matching pairs, wherein the unique driving plan guides the particular autonomous vehicle to the specific location of the filtered event to collect image data.

[0005] In one aspect, the one or more events include traffic accidents involving one or more vehicles.

[0006] On the other hand, the one or more events indicate the appearance of an object.

[0007] In another respect, the objects are one of the following: potholes on roads, traffic signs, street signs, road markings, buildings, landmarks, cyclists and pedestrians.

[0008] In one aspect, the one or more notifications are generated by another autonomous vehicle or individual.

[0009] On the other hand, a unique driving plan can be determined based on dynamic programming scheduling methods or greedy algorithm methods.

[0010] On the other hand, a particular autonomous vehicle includes event observation capacity, which indicates the number of events in which a particular autonomous vehicle simultaneously observes and collects image data.

[0011] In one aspect, a centralized scheduling system executes instructions to run a dynamically programmed scheduling algorithm for a predetermined number of rounds to determine a unique driving plan, where the predetermined number of rounds is equal to the event observation capacity of a particular autonomous vehicle.

[0012] On the other hand, the centralized scheduling system executes instructions to carry out a greedy algorithm that sequentially introduces events into a particular autonomous vehicle's unique driving plan until the event observation capacity of that particular autonomous vehicle is reached.

[0013] On the other hand, the centralized scheduling system executes instructions to carry out a greedy algorithm, which introduces events based on the total number of events occurring at each event location until the event observation capacity of a particular autonomous vehicle is reached.

[0014] In one aspect, the centralized scheduling system executes instructions to perform a greedy algorithm that introduces events based on the minimum number of observers required for each event in the event pool until the event observation capacity of a particular autonomous vehicle is reached.

[0015] On the other hand, the minimum number of observers represents the minimum number of vehicles required to collect image data for a specific event.

[0016] On the other hand, the centralized scheduling system executes instructions to determine the maximum capacity percentage of each of one or more autonomous vehicles based on machine learning algorithms, where the maximum capacity percentage indicates the availability for performing unexpected tasks not included as part of the sole driving plan.

[0017] In one aspect, the centralized scheduling system executes instructions to calculate the cost function of a machine learning algorithm and solves for the output value, which is part of the cost function, indicating the maximum capacity percentage of a particular autonomous vehicle.

[0018] In one aspect, a method for determining a unique driving plan for a specific autonomous vehicle. The method includes receiving one or more notifications from a centralized scheduling system indicating that an event has occurred, wherein the centralized scheduling system wirelessly communicates with one or more autonomous vehicles, and each autonomous vehicle collects image data associated with the one or more events. The method includes creating an event pool by the centralized scheduling system to store the one or more events. The method further includes comparing a predetermined route corresponding to the specific autonomous vehicle with a specific location corresponding to each event stored in the event pool to identify one or more filtered events, wherein the specific autonomous vehicle travels to the specific location corresponding to the filtered event when following the predetermined route. The method includes determining the presence of the specific autonomous vehicle based on the predetermined route and the specific location of the filtered event when the filtered event occurs. The method also includes identifying matching pairs comprising the filtered events and the predetermined route of the specific autonomous vehicle. The method further includes determining a unique driving plan for the specific autonomous vehicle based on the matching pairs, wherein the unique driving plan guides the specific autonomous vehicle to the specific location of the filtered event to collect image data associated with the filtered event. The method also includes executing a dynamically programmed scheduling algorithm for a predetermined number of rounds to determine the unique driving plan, wherein the predetermined number of rounds is equal to the event observation capacity of the specific autonomous vehicle.

[0019] In one aspect, the method includes executing a greedy algorithm that sequentially introduces events into a unique driving plan for a particular autonomous vehicle until the event observation capacity of the particular autonomous vehicle is reached.

[0020] On the other hand, the method includes executing a greedy algorithm that introduces events based on the total number of events occurring at each event location until the event observation capacity of a particular autonomous vehicle is reached.

[0021] On another aspect, the method includes executing a greedy algorithm that introduces events based on the minimum number of observers required for each event in the event pool until the event observation capacity of a particular autonomous vehicle is reached.

[0022] This invention provides the following technical solutions:

[0023] 1. An event scheduling system for collecting image data related to one or more events via one or more autonomous vehicles, wherein the event scheduling system comprises:

[0024] A centralized dispatch system that wirelessly communicates with the one or more autonomous vehicles, wherein each autonomous vehicle collects image data related to the one or more events while following a unique driving plan, and wherein the centralized dispatch system executes instructions to:

[0025] Receive one or more notifications indicating that an event has occurred;

[0026] Create an event pool to store the one or more events;

[0027] A predetermined route corresponding to a specific autonomous vehicle is compared with a specific location corresponding to each event stored in the event pool to identify one or more filtered events;

[0028] Based on a predetermined route and the specific location of the filtered events, it is determined when a specific autonomous vehicle appears when the filtered events occur.

[0029] Identify matching pairs including the filtered events and the predetermined routes of the specific autonomous vehicle; and

[0030] Based on the matching pair, a unique driving plan is determined for the specific autonomous vehicle, wherein the unique driving plan guides the specific autonomous vehicle to a specific location of the filtered event to collect image data.

[0031] According to the event scheduling system of technical solution 1, the one or more events include traffic accidents involving one or more vehicles.

[0032] According to the event scheduling system of technical solution 1, the occurrence of one or more event indication objects.

[0033] According to the event scheduling system of technical solution 3, the object is one of the following: potholes on the road, traffic signs, street signs, road markings, buildings, landmarks, cyclists and pedestrians.

[0034] According to the event scheduling system of technical solution 1, the one or more notifications are generated by another autonomous vehicle or individual.

[0035] According to the event scheduling system of technical solution 1, the unique driving plan is determined based on a dynamic programming scheduling method or a greedy algorithm method.

[0036] According to the event scheduling system of technical solution 1, the specific autonomous vehicle includes an event observation capacity, which indicates the number of events in which the specific autonomous vehicle simultaneously observes and collects image data.

[0037] According to the event scheduling system described in technical solution 7, the centralized scheduling system executes instructions to:

[0038] The dynamic programming scheduling algorithm is executed for a predetermined number of rounds to determine a unique driving plan, wherein the predetermined number of rounds is equal to the event observation capacity of the particular autonomous vehicle.

[0039] According to the event scheduling system described in technical solution 7, the centralized scheduling system executes instructions to:

[0040] A greedy algorithm is executed, which sequentially introduces events into a specific autonomous vehicle's unique driving plan until the event observation capacity of the specific autonomous vehicle is reached.

[0041] According to the event scheduling system described in technical solution 7, the centralized scheduling system executes instructions to:

[0042] A greedy algorithm is executed, which introduces events based on the total number of events occurring at each event location until the event observation capacity of the particular autonomous vehicle is reached.

[0043] According to the event scheduling system described in technical solution 7, the centralized scheduling system executes instructions to:

[0044] A greedy algorithm is executed, which introduces events based on the minimum number of observers required for each event in the event pool, until the event observation capacity of the particular autonomous vehicle is reached.

[0045] According to the event scheduling system of technical solution 11, the minimum number of observers represents the minimum number of vehicles required to collect image data of a specific event.

[0046] According to the event scheduling system described in technical solution 1, the centralized scheduling system executes instructions to:

[0047] The maximum capacity percentage of each of the one or more autonomous vehicles is determined based on a machine learning algorithm, wherein the maximum capacity percentage indicates the availability for performing unexpected tasks not included as part of the sole driving plan.

[0048] According to the event scheduling system of technical solution 14, the centralized scheduling system executes instructions to:

[0049] Calculate the cost function of the machine learning algorithm; and

[0050] Solve for the output value as part of the cost function, where the output value indicates the maximum capacity percentage of a particular autonomous vehicle.

[0051] A method for determining a unique driving plan for a specific autonomous vehicle, the method comprising:

[0052] A centralized dispatch system receives one or more notifications indicating that an event has occurred, wherein the centralized dispatch system communicates wirelessly with one or more autonomous vehicles, and each autonomous vehicle collects image data related to the one or more events;

[0053] An event pool is created by the centralized scheduling system to store the one or more events;

[0054] A predetermined route corresponding to a specific autonomous vehicle is compared with a specific location corresponding to each event stored in the event pool to identify one or more filtered events, wherein the specific autonomous vehicle travels to a specific location corresponding to a filtered event while traveling along the predetermined route;

[0055] Based on the predetermined route and the specific location of the filtered events, it is determined when a specific autonomous vehicle appears when the filtered events occur;

[0056] Identify matching pairs, including filtered events and predetermined routes for the specific autonomous vehicle; and

[0057] Based on the matching pair, a unique driving plan is determined for the specific autonomous vehicle, wherein the unique driving plan guides the specific autonomous vehicle to a specific location of the filtered event in order to collect image data related to the filtered event.

[0058] The method according to technical solution 16 further includes:

[0059] The dynamic programming scheduling algorithm is executed for a predetermined number of rounds to determine a unique driving plan, wherein the predetermined number of rounds is equal to the event observation capacity of a particular autonomous vehicle.

[0060] The method according to technical solution 16 further includes:

[0061] A greedy algorithm is executed, which sequentially introduces events into the unique driving plan of the particular autonomous vehicle until the event observation capacity of the particular autonomous vehicle is reached.

[0062] The method according to technical solution 16 further includes:

[0063] A greedy algorithm is executed, which introduces events based on the total number of events occurring at each event location until the event observation capacity of a specific autonomous vehicle is reached.

[0064] The method according to technical solution 16 further includes:

[0065] A greedy algorithm is executed, which introduces events based on the minimum number of observers required for each event in the event pool, until the event observation capacity of the particular autonomous vehicle is reached.

[0066] Further areas of application will become apparent from the description provided herein. It should be understood that the descriptions and specific examples are for illustrative purposes only and are not intended to limit the scope of this disclosure. Attached Figure Description

[0067] The accompanying drawings described herein are for illustrative purposes only and are not intended to limit the scope of this disclosure in any way.

[0068] Figure 1 This is a schematic diagram of a disclosed event scheduling system according to an exemplary embodiment, the event scheduling system including a centralized scheduling system that communicates wirelessly with one or more autonomous vehicles;

[0069] Figure 2 This is according to an exemplary embodiment. Figure 1 The diagram shows the operation flowchart of a centralized scheduling system, which includes an event pool and individual driving plan blocks for determining the unique driving plan of a specific autonomous vehicle.

[0070] Figure 3 This is a timing diagram illustrating a dynamic programming scheduling method for determining a unique travel plan according to an exemplary embodiment;

[0071] Figure 4 This is a timing diagram illustrating a greedy algorithm method for determining a unique driving plan according to an exemplary embodiment; and

[0072] Figure 5 This is a flowchart illustrating a method for determining a unique driving plan for a specific autonomous vehicle using a disclosed event scheduling system, according to an exemplary embodiment. Detailed Implementation

[0073] The following description is exemplary in nature and is not intended to limit this disclosure, application, or use.

[0074] refer to Figure 1The illustration depicts an exemplary event dispatch system 10 for collecting image data related to one or more events. The event dispatch system 10 includes one or more autonomous vehicles 12 and a back-end office 16, wherein the one or more autonomous vehicles 12 wirelessly communicate with one or more centralized dispatch systems 18, which are part of the back-end office 16, via a network 26. Each autonomous vehicle 12 includes one or more autopilot controllers 20 that execute routing algorithms to determine a predetermined route 30 for a corresponding one of the autonomous vehicles 12, which the autonomous vehicle 12 follows. Each autonomous vehicle 12 shares its corresponding predetermined route 30 with the one or more centralized dispatch systems 18 of the back-end office 16. It should be understood that the autonomous vehicle 12 can be any type of vehicle, such as, but not limited to, a car, truck, SUV, van, or motorhome.

[0075] As explained below, the event scheduling system 10 determines a unique driving plan 28 for each autonomous vehicle 12 (see below). Figure 2 Each autonomous vehicle 12 includes at least one camera 22 in electronic communication with an autopilot controller 20, wherein the camera 22 collects image data related to one or more events while following a unique driving plan 28. The image data collected by the autonomous vehicle 12 is shared with one or more centralized dispatch systems 18 in a back-end office 16. The event dispatch system 10 can collect image data related to a specific event from multiple perspectives (i.e., image data collected by different vehicle cameras with different perspectives). In an embodiment, if a camera 22 of an autonomous vehicle 12 fails to collect data (e.g., if the camera 22 is malfunctioning or if visual occlusion occurs), the event dispatch system 10 can compensate for the problem by using image data collected by another autonomous vehicle 12.

[0076] An event is any accident or object captured by image data. For example, an event could be a traffic accident involving one or more vehicles. Alternatively, in another example, an event indicates the presence of an object. Some examples of objects include, but are not limited to, potholes on a road, traffic signs, street signs, road markings, buildings, landmarks such as monuments or statues, or individuals such as cyclists or pedestrians. It should be understood that the event remains valid for a period of time and does not occur instantaneously. Rather, the duration of the event is long enough for the event dispatch system 10 to dispatch one or more autonomous vehicles 12 to the corresponding location of the event to collect image data associated with the event. For example, if the event is a traffic accident between two vehicles, the duration of the traffic accident is long enough for one or more autonomous vehicles 12 to be dispatched to the corresponding location to collect image data.

[0077] It should be understood that each individual event includes a required or minimum number of observers, where the minimum number of observers represents the minimum number of vehicles required to collect image data for a particular event. The minimum number of observers is based on the nature and type of the event, where some types of events may require more observers than others. For example, an event such as a traffic accident involving more than two vehicles requires more observers than viewing potholes or street signs. Therefore, in embodiments, the disclosed event scheduling system 10 collects image data related to a particular event from multiple perspectives (i.e., image data collected by different vehicles). It should also be understood that each individual event occurs at a specific location. For example, if the individual event is a traffic accident, the specific location indicates where the traffic accident occurred on the road.

[0078] Figure 2 yes Figure 1 The flowchart shown is for a centralized dispatch system 18 used to determine a unique driving plan 28 for a specific autonomous vehicle 12. The centralized dispatch system 18 receives one or more notifications 36 indicating that an event has occurred. It should be understood that the notifications 36 are generated by another autonomous vehicle 12 connected to the network 26, or alternatively, by an individual. For example, an autonomous vehicle 12 can send a notification indicating a traffic accident via the network 26. Alternatively, an individual, such as a cloud operator or a vehicle driver, can send a notification 36 indicating that an event is occurring.

[0079] The centralized scheduling system 18 creates an event pool 40 and individual driving plan blocks 42. The event pool 40 stores events, and the individual driving plan block 42 determines a unique driving plan 28 for a specific autonomous vehicle 12. The individual driving plan block 42 includes a filter 50, a matching block 52, and a route determination block 54 that determines the unique driving plan 28. (Reference) Figure 1 and Figure 2 Each autonomous vehicle 12 has one or more automated driving controllers 20 that determine a predetermined route 30 based on a starting point 46, a destination point 48, and any stops between the starting and destination points. For example... Figure 2 As seen in the diagram, the predetermined route 30 determined by a specific autonomous vehicle 12 is transmitted via network 26 to the filter 50 of the individual driving plan block 42 of the one or more centralized dispatch systems 18.

[0080] The filter 50 of the one or more centralized scheduling systems 18 compares the predetermined route 30 of a particular autonomous vehicle 12 with a specific location corresponding to each event stored in the event pool 40, and determines whether the particular autonomous vehicle 12 has traveled to any of the specific locations of the events stored in the event pool 40. The filter 50 then identifies one or more filtered events 60 in which the particular autonomous vehicle 12 has traveled to a specific location corresponding to the filtered event 60.

[0081] Matching block 52 receives one or more filtered events 60 and determines that a specific autonomous vehicle appears when the filtered event occurs, based on a predetermined route 30 corresponding to a specific autonomous vehicle and a specific location of the filtered event. In an embodiment, centralized scheduling system 18 compares a first time interval in which a specific autonomous vehicle 12 appears at a specific location of a filtered event 60 while following the predetermined route 30 with a second time interval in which the filtered event 60 occurs, wherein if the first time interval overlaps with the second time interval, centralized scheduling system 18 determines that the specific autonomous vehicle 12 appears when the filtered event occurs.

[0082] In response to determining that a specific autonomous vehicle 12 is present at a specific location when a specific event occurs, matching block 52 identifies matching pairs 62. Matching pairs 62 include filtered events 60 and a predetermined route 30 for the specific autonomous vehicle 12. Individual driving plan block 42 identifies matching pairs for each filtered event 60 identified by filter 50. Route determination block 54 then determines a unique driving plan 28 based on the matching pairs 62. In other words, unique driving plan 28 includes each filtered event 60 as part of an event pool 40 that occurs simultaneously with the presence of the specific autonomous vehicle 12 to collect image data. Specifically, the specific autonomous vehicle 12 appears and collects image data at a specific location for each filtered event 60 while following unique driving plan 28. It should be understood that unique driving plan 28 can be determined based on a dynamic programming scheduling method or a greedy algorithm method, which are described in more detail below.

[0083] Figure 3 This is a timing diagram illustrating a dynamic programming scheduling method, and it shows three event sequences 70, each of which includes four events 72. The timing diagram includes an x-axis, where time T is represented along the x-axis. Figure 3 As shown, events 72 are disjointed, therefore they do not overlap within their respective durations. (See reference) Figure 2 and Figure 3 Both of these should be understood to mean that the duration includes not only the time spent recording event 72 in real time, but also the processing and uploading time. Specifically, the duration includes capturing video or images in real time via the camera 22 of the autonomous vehicle 12, storing the image data in the memory or hard drive of the autonomous driving controller 20, and uploading the image data to one or more centralized scheduling systems 18 via the network 26.

[0084] Continue to refer to Figure 2 and Figure 3 It should be understood that each autonomous vehicle 12 includes an event observation capacity, which indicates the number of events that a particular autonomous vehicle 12 simultaneously observes and collects data on. Figure 3 In the example shown, the event observation capacity is three, and therefore three event sequences 70 are shown. It should be understood that... Figure 3 This is merely an example, and the event observation capacity varies from vehicle to vehicle and depends on the available storage in the memory of one or more autonomous driving controllers 20 and the network speed of network 26.

[0085] When using a dynamic programming scheduling method, the individual driving plan block 42 of the centralized scheduling system 18 determines a unique driving plan 28 for a specific autonomous vehicle 12 by selecting as many events 72 as possible from each event sequence 70 for simultaneous observation without exceeding the event observation capacity of the specific autonomous vehicle 12. The individual driving plan block 42 of the centralized scheduling system 18 determines the unique driving plan 28 by executing the dynamic programming scheduling algorithm a predetermined number of rounds, where the predetermined number of rounds is equal to the event observation capacity. For example, if the specific autonomous vehicle 12 can observe three events simultaneously, the dynamic programming scheduling algorithm is executed three times. Each time the dynamic programming scheduling algorithm is executed, it determines one of the three event sequences 70.

[0086] For example, during the first execution of the dynamic programming scheduling algorithm, a first event sequence 70A is generated by selecting disjointed events 72 that do not overlap within their respective durations. It should be understood that the dynamic programming scheduling algorithm selects as many events 72 as possible for the event sequence without creating any temporal overlap between each event 72. During each event 72, the corresponding camera 22 of the specific autonomous vehicle 12 appears and collects image data at a specific location before the corresponding event 72 expires. Event 72 expires once the occurrence of the event is no longer defined. For example, if event 72 is a traffic accident, then event 72 expires once the traffic accident has been cleared and the vehicles involved have driven away. Once the specific autonomous vehicle 12 has completed its observation of the event, it can drive to the specific location associated with the next event 72, which is part of the first event sequence 70A. Similarly, during the second execution of the dynamic programming scheduling algorithm, a second event sequence 70B is generated, and during the third execution, a third event sequence 70C is generated. Once the dynamic programming scheduling algorithm has been executed for a predetermined number of rounds, the individual driving plan block 42 of the centralized scheduling system 18 determines a unique driving plan 28 for a specific autonomous vehicle 12 by merging events 72 together. Figure 3 In the example shown, merging event 72 includes merging all events 72 in the first event sequence 70A, the second event sequence 70B, and the third event sequence 70C together.

[0087] Figure 4This is a timing diagram illustrating a greedy algorithm method for determining a unique driving plan 28 according to an exemplary embodiment. In the example shown, the x-axis represents time T. Figure 4 In the non-limiting embodiment shown, the timing diagram illustrates seven events 170, numbered E1-E7. (See reference...) Figure 2 and Figure 4 In one embodiment, the individual driving plan block 42 of the centralized scheduling system 18 determines the unique driving plan 28 for a specific autonomous vehicle 12 by executing a greedy algorithm and sequentially introducing events into the unique driving plan 28 until the event observation capacity of the specific autonomous vehicle 12 is met. Once the event observation capacity of the specific autonomous vehicle 12 is met, the greedy algorithm ignores additional events and will not introduce additional events into the unique driving plan 28 until one of the selected events has been completed. Figure 4 In the example shown, the event observation capacity is two. Therefore, in the example shown, if events E3 and E4 are both selected as part of the unique driving plan 28, the fifth event E5 may not be introduced into the unique driving plan 28 until the third event E3 ends.

[0088] In one embodiment, the individual driving plan block 42 of the centralized scheduling system 18 determines the unique driving plan 28 for a specific autonomous vehicle 12 by executing a greedy algorithm and adding events to the unique driving plan 28 based on the total number of events occurring at a specific location for each event in the event pool 40 until the event observation capacity of the specific autonomous vehicle 12 is reached. Specifically, the greedy algorithm compares the total number of events occurring at a specific location for each event stored in the event pool 40 and selects the event corresponding to the specific location with the maximum number of events to add to the unique driving plan 28 for the specific autonomous vehicle 12. For example, if the first location A includes four events (e.g., events E1, E3, E5, E6) and the second location B includes three events (e.g., events E2, E4, E7), the greedy algorithm selects the fifth event E5 instead of the fourth event E4 because the fourth event E4 is located in the first location A where more events are located. Now that the first location A and the second location B have an equal number of events, the greedy algorithm can arbitrarily select the next event.

[0089] In one embodiment, the individual driving plan block 42 of the centralized scheduling system 18 determines the unique driving plan 28 for a specific autonomous vehicle 12 by executing a greedy algorithm and introducing events from the event pool 40 into the unique driving plan 28 based on the minimum number of observers required for each event in the event pool 40 until the event observation capacity of the specific autonomous vehicle is reached. That is, the greedy algorithm selects the event that requires the maximum number of observers. For example, if the minimum number of observers required for the fourth event E4 is 9, and the minimum number of observers required for the fifth event E5 is 10, then the greedy algorithm selects the fifth event E5 because the fifth event E5 includes the maximum number of observers. Now that the fourth event E4 and the fifth event E5 have an equal number of events, the greedy algorithm can arbitrarily select the next event.

[0090] refer to Figure 2 In one embodiment, the centralized scheduling system 18 determines the maximum capacity percentage of each autonomous vehicle 12 based on a machine learning algorithm, where the maximum capacity percentage indicates the availability for performing unexpected tasks not included as part of the sole driving plan 28. Specifically, the centralized scheduling system 18 calculates the cost function of the machine learning algorithm. The maximum capacity percentage of a particular autonomous vehicle 12 is determined by solving for the output value y, which indicates the maximum capacity percentage of that particular autonomous vehicle 12. In one embodiment, the cost function is determined by combining a standard linear regression formula with a regularization function and is expressed by the following equation:

[0091]

[0092] in It is assumed that m represents the number of data records used in the machine learning algorithm. The regularization parameter indicates what a machine learning algorithm should use, where x is the feature vector and y is the output value. The feature vector is represented as... x1 indicates the road type, such as highways, local roads, and residential roads; x2 indicates the road structure, such as ramps, intersections, and roundabouts; x3 indicates the road speed limit; x4 indicates the number of events occurring at a specific location; x5 indicates the geohash code; x6 indicates the percentage of vehicle capacity used; and x7 indicates how many unplanned events a particular autonomous vehicle 12 needs to perform at a specific location.

[0093] Figure 5 This is a flowchart illustrating the method 200 for determining a unique driving plan 28 for a specific autonomous vehicle 12. (Overall reference) Figure 1 , Figure 2 and Figure 5Method 200 may begin at box 202. In box 202, the centralized dispatch system 18 receives one or more notifications 36 indicating that an event has occurred. As mentioned above, notification 36 is generated by another autonomous vehicle 12 connected to network 26, or alternatively, by an individual. Method 200 may then proceed to box 204.

[0094] In box 204, the centralized scheduling system 18 creates an event pool 40 that stores one or more events. Method 200 can then proceed to box 206.

[0095] In block 206, the centralized scheduling system 18 compares a predetermined route corresponding to a particular autonomous vehicle 12 with a specific location corresponding to each event stored in the event pool 40 to identify one or more filtered events, wherein the particular autonomous vehicle 12 travels to the specific location corresponding to the filtered event while following the predetermined route 30. Method 200 can then proceed to block 208.

[0096] In block 208, the centralized scheduling system 18 determines whether a specific autonomous vehicle 12 is present when the filtered event occurs, based on a predetermined route 30 and the specific location of the filtered event. Specifically, in one embodiment, one or more centralized scheduling systems 18 compare a first time interval in which the specific autonomous vehicle appears at the specific location of the filtered event while following the predetermined route with a second time interval in which the filtered event occurs to determine whether the first time interval overlaps with the second time interval. If the first time interval overlaps with the second time interval, it indicates that the specific autonomous vehicle 12 is present when the filtered event occurs. If the first time interval does not overlap with the second time interval, method 200 can terminate. Otherwise, method 200 can then proceed to block 210.

[0097] In box 210, the centralized scheduling system 18 identifies matching pairs, which include filtered events and predetermined routes for specific autonomous vehicles. Method 200 can then proceed to box 212.

[0098] In box 212, the centralized scheduling system 18 determines a unique driving plan 28 for a particular autonomous vehicle 12 based on the matching pairs identified in box 210. Method 200 can then terminate.

[0099] Referring generally to the accompanying drawings, the disclosed event scheduling system provides various technical effects and benefits by collecting image data about one or more events. Multiple perspectives can be used to collect image data (because different vehicle cameras have different perspectives). Therefore, if one or more cameras of an autonomous vehicle cannot collect data at a specific location, image data collected by the event scheduling system can be used instead. Thus, the disclosed system provides a cost-effective and relatively simple method for collecting image data based on task scheduling among autonomous vehicles.

[0100] A controller can refer to electronic circuitry, combinational logic circuitry, a field-programmable gate array (FPGA), a processor (shared, dedicated, or grouped) that executes code, or a combination of some or all of the above (such as in a system-on-a-chip), or a portion thereof. Furthermore, the controller can be microprocessor-based, such as a computer having at least one processor, memory (RAM and / or ROM), and associated input and output buses. The processor can operate under the control of an operating system residing in memory. The operating system can manage computer resources such that computer program code, embodied as one or more computer software applications—such as applications residing in memory—can have instructions that are executed by the processor. In alternative embodiments, the processor can directly execute the application, in which case the operating system can be omitted.

[0101] The description in this disclosure is merely exemplary in nature, and variations thereof without departing from the spirit and scope of this disclosure are intended to be within its scope. Such variations should not be considered as departing from the spirit and scope of this disclosure.

Claims

1. An event scheduling system for collecting image data related to one or more events via one or more autonomous vehicles, wherein the event scheduling system comprises: A centralized dispatch system that wirelessly communicates with the one or more autonomous vehicles, wherein each autonomous vehicle, while following a unique driving plan, has image data related to the one or more events collected by a corresponding camera, and wherein the centralized dispatch system executes instructions to: Receive one or more notifications indicating that an event has occurred; Create an event pool to store the one or more events; A predetermined route corresponding to a specific autonomous vehicle is compared with a specific location corresponding to each event stored in the event pool to identify one or more filtered events; Based on a predetermined route and the specific location of the filtered events, it is determined when a specific autonomous vehicle appears when the filtered events occur. Identify matching pairs including the filtered events and the predetermined routes of the specific autonomous vehicle; and Based on the matching pair, a unique driving plan is determined for the specific autonomous vehicle, wherein the unique driving plan guides the specific autonomous vehicle to a specific location of the filtered event to collect image data, and wherein the corresponding camera of the specific autonomous vehicle appears and collects image data at the specific location before the corresponding event expires.

2. The event scheduling system according to claim 1, wherein, The one or more events include traffic accidents involving one or more vehicles.

3. The event scheduling system according to claim 1, wherein, The one or more events indicate the occurrence of the object.

4. The event scheduling system according to claim 3, wherein, The objects referred to are one of the following: potholes on roads, traffic signs, street signs, road markings, buildings, landmarks, cyclists, and pedestrians.

5. The event scheduling system according to claim 1, wherein, The one or more notifications are generated by another autonomous vehicle or individual.

6. The event scheduling system according to claim 1, wherein, The unique driving plan is determined based on a dynamic programming scheduling method or a greedy algorithm method.

7. The event scheduling system according to claim 1, wherein, The specific autonomous vehicle includes an event observation capacity, which indicates the number of events in which the specific autonomous vehicle simultaneously observes and collects image data.

8. The event scheduling system according to claim 7, wherein, The centralized scheduling system executes instructions as follows: The dynamic programming scheduling algorithm is executed for a predetermined number of rounds to determine a unique driving plan, wherein the predetermined number of rounds is equal to the event observation capacity of the particular autonomous vehicle.

9. The event scheduling system according to claim 7, wherein, The centralized scheduling system executes instructions as follows: A greedy algorithm is executed, which sequentially introduces events into a specific autonomous vehicle's unique driving plan until the event observation capacity of the specific autonomous vehicle is reached.

10. The event scheduling system according to claim 7, wherein, The centralized scheduling system executes instructions as follows: A greedy algorithm is executed, which introduces events based on the total number of events occurring at each event location until the event observation capacity of the particular autonomous vehicle is reached.

11. The event scheduling system according to claim 7, wherein, The centralized scheduling system executes instructions as follows: A greedy algorithm is executed, which introduces events based on the minimum number of observers required for each event in the event pool, until the event observation capacity of the particular autonomous vehicle is reached.

12. The event scheduling system according to claim 11, wherein, The minimum number of observers refers to the minimum number of vehicles required to collect image data for a specific event.

13. The event scheduling system according to claim 1, wherein, The centralized scheduling system executes instructions as follows: The maximum capacity percentage of each of the one or more autonomous vehicles is determined based on a machine learning algorithm, wherein the maximum capacity percentage indicates the availability for performing unexpected tasks not included as part of the sole driving plan.

14. The event scheduling system according to claim 13, wherein, The centralized scheduling system executes instructions as follows: Calculate the cost function of the machine learning algorithm; and Solve for the output value as part of the cost function, where the output value indicates the maximum capacity percentage of a particular autonomous vehicle.

15. A method for determining a unique driving plan for a specific autonomous vehicle, the method comprising: A centralized dispatch system receives one or more notifications indicating that an event has occurred, wherein the centralized dispatch system communicates wirelessly with one or more autonomous vehicles, and each autonomous vehicle collects image data related to the one or more events; An event pool is created by the centralized scheduling system to store the one or more events; A predetermined route corresponding to a specific autonomous vehicle is compared with a specific location corresponding to each event stored in the event pool to identify one or more filtered events, wherein the specific autonomous vehicle travels to a specific location corresponding to a filtered event when following the predetermined route; Based on the predetermined route and the specific location of the filtered events, it is determined when a specific autonomous vehicle appears when the filtered events occur; Identify matching pairs, including filtered events and predetermined routes for the specific autonomous vehicle; and Based on the matching pair, a unique driving plan is determined for the specific autonomous vehicle, wherein the unique driving plan guides the specific autonomous vehicle to a specific location of the filtered event to collect image data related to the filtered event, and wherein a corresponding camera of the specific autonomous vehicle appears and collects image data at the specific location before the corresponding event expires.

16. The method of claim 15, further comprising: The dynamic programming scheduling algorithm is executed for a predetermined number of rounds to determine a unique driving plan, wherein the predetermined number of rounds is equal to the event observation capacity of a particular autonomous vehicle.

17. The method of claim 15, further comprising: A greedy algorithm is executed, which sequentially introduces events into the unique driving plan of the particular autonomous vehicle until the event observation capacity of the particular autonomous vehicle is reached.

18. The method of claim 15, further comprising: A greedy algorithm is executed, which introduces events based on the total number of events occurring at each event location until the event observation capacity of a specific autonomous vehicle is reached.

19. The method of claim 15, further comprising: A greedy algorithm is executed, which introduces events based on the minimum number of observers required for each event in the event pool, until the event observation capacity of the particular autonomous vehicle is reached.

20. The method of claim 15, further comprising: The maximum capacity percentage of each of the one or more autonomous vehicles is determined based on a machine learning algorithm, wherein the maximum capacity percentage indicates the availability for performing unexpected tasks not included as part of the sole driving plan.

Images