Method and system for implementing traffic signal management for the operation of autonomous vehicles

Abstract

Method (200) for implementing traffic signal management for the operation of autonomous vehicles, the method comprising: (205, 220) Received, at an administrative system (110), of vehicle data (115) from two or more vehicles (100) at an intersection with one or more traffic lights (130), wherein the vehicle data (115) includes a vehicle location, a vehicle speed and image information or images; (200, 300, 600) Determine, on the management system (110), at least one of three types of information about the one or more traffic lights (130) based on the vehicle data (115), wherein the three types of information include a location of the one or more traffic lights (130), a signal phase and timing (SPaT) of the one or more traffic lights (130), and a lane correspondence of the one or more traffic lights (130); and (340, 440) Providing, through the management system (110), at least one of the three types of information about the one or more traffic lights (130) for the operation of autonomous vehicles (100); wherein determining the location (240) of the one or more traffic lights (130) includes the use of the image information or images and the use of the vehicle location from the vehicle data (115) obtained from the two or more vehicles (100) at the intersection, and wherein the method (200) also includes: Received, at the management system, of additional information from one or more other sources, wherein the additional information includes a verified location of the one or more traffic lights (130) and the one or more other sources include the two or more vehicles (100) relaying a message from the intersection, an urban traffic center information system or a commercial traffic center information system (230), (250) Determining a consensus between the location of the one or more traffic lights (130), determined using the vehicle data (115) of the two or more vehicles (100) at the intersection, and the verified location, obtained from the additional information using a standard deviation or a mean absolute deviation, and (260) Proceeding with the provision of the location of one or more traffic lights (130) for the operation of autonomous vehicles (100) on the basis of the consensus which is greater than a consensus threshold, and (270) Increasing the number of the two or more vehicles (100) from which the vehicle data (115) are obtained, based on the consensus being less than the consensus threshold.

Description

[0001] The subject matter of the invention relates to traffic signal management for autonomous vehicle operation.

[0002] The autonomous operation of a vehicle requires the gathering of information and decision-making at a speed sufficient to act and react in real-time situations. Examples of vehicles that can operate autonomously include cars, trucks, construction equipment, agricultural machinery, and automated factory equipment. Autonomous operation requires a vehicle to observe and consider many of the same inputs as a human driver. For example, when an autonomous vehicle approaches an intersection, it must observe and consider the color of the illuminated traffic light. To do this, the autonomous vehicle must first accurately determine the location of the traffic light within its field of view. The autonomous vehicle must also determine which traffic light corresponds to the lane it is traveling in. Therefore, it is desirable to implement traffic signal management for autonomous operation.

[0003] US 2019 / 0251838A1 concerns a method for controlling the signal phase and timing (SPaT) of traffic lights and for transmitting notification messages. The method involves collecting information about traffic flow, incidents, and emergency vehicles on a predetermined segment of a road network. The method dynamically adjusts the SPaT information for a traffic light in the predetermined segment of the road network based on the collected information and provides a change in the SPaT status (normal or abnormal) along with the adjusted SPaT information.

[0004] DE 10 2018 007 962 A1 relates to a method for detecting traffic light positions, in which a traffic scene at a traffic light intersection is recorded by a majority of fleet vehicles using a camera and traffic light candidates are determined in the recorded traffic scene and uploaded to a central backend.

[0005] The objective can be considered to be to provide an improved method and system for implementing traffic signal management for the operation of autonomous vehicles, thereby ensuring the timely measures necessary for safe and efficient operation. This objective is achieved by the subject matter of claim 1 and claim 5.

[0006] The inventive method for implementing traffic signal management for the operation of autonomous vehicles comprises obtaining vehicle data from two or more vehicles at an intersection with one or more traffic lights. The vehicle data includes vehicle location, vehicle speed, and image information or images. The method also comprises determining at least one of three types of information about the one or more traffic lights based on the vehicle data. The three types of information include the location of the one or more traffic lights, a signal phase and timing (SPaT) of the one or more traffic lights, and a lane correspondence of the one or more traffic lights. At least one of the three types of information about the one or more traffic lights is provided for the operation of autonomous vehicles.

[0007] Determining the location of one or more traffic lights involves using image information or images and using the vehicle location from the vehicle data obtained from the two or more vehicles at the intersection.

[0008] The procedure also includes obtaining additional information from one or more other sources. This additional information includes a verified location of the one or more traffic lights and the one or more other sources, including the two or more vehicles relaying a message from the intersection, an urban traffic center information system, or a commercial traffic center information system.

[0009] The procedure also includes determining a consensus between the location of the one or more traffic lights, which is determined based on the vehicle data of the two or more vehicles at the intersection, and the verified location, which is determined from the additional information using a standard deviation or a mean absolute deviation.

[0010] The procedure also includes proceeding with providing the location of one or more traffic lights for the operation of autonomous vehicles based on a consensus greater than a consensus threshold, and increasing the number of the two or more vehicles from which vehicle data is obtained based on a consensus less than the consensus threshold.

[0011] According to one embodiment, the method also includes obtaining SPaT observations for the one or more traffic lights from the two or more vehicles based on image information or images, and determining, based on a consensus among the SPaT observations from the two or more vehicles that is greater than a first threshold, whether the SPaT of the one or more traffic lights is static, so that the SPaT is always the same, or dynamic, so that the SPaT changes based on traffic flow or time of day. The method also includes obtaining motion-based SPaT for the one or more traffic lights based on the vehicle location and the vehicle speed specified in the vehicle data of the two or more vehicles, and determining a consensus between the SPaT observations and the motion-based SPaT.

[0012] The procedure includes, based on the consensus between the SPaT observations and the motion-based SPaT, which is greater than a second threshold, also continuing to provide the SPaT of one or more traffic lights, based on the fact that the SPaT of one or more traffic lights is static, and continuing to provide the SPaT of one or more traffic lights together with a confidence value that decreases with reduced recency of the reporting of vehicle data to the management system, based on the fact that the SPaT of one or more traffic lights is dynamic.

[0013] According to one embodiment, the method also includes determining the lane correspondence based on the vehicle location and image information or images of the two or more vehicles.

[0014] According to one embodiment, the provision of at least one of the three types of information about the one or more traffic lights for the operation of autonomous vehicles is done as an application programming interface (API) or as an extended map.

[0015] The system according to the invention for implementing traffic signal management for the operation of autonomous vehicles comprises a communication module for receiving vehicle data from two or more vehicles at an intersection with one or more traffic lights. The vehicle data includes vehicle location, vehicle speed, and image information or images. The system also includes a processor for determining at least one of three types of information about the one or more traffic lights based on the vehicle data, wherein the three types of information include the location of the one or more traffic lights, the signal phase and timing (SPaT) of the one or more traffic lights, and the lane correspondence of the one or more traffic lights, and for providing the at least one of the three types of information about the one or more traffic lights for the operation of autonomous vehicles.

[0016] The processor determines the location of one or more traffic lights based on the image information or images and using the vehicle location from the vehicle data obtained from the two or more vehicles at the intersection.

[0017] The communication module receives additional information from one or more other sources, the additional information including a verified location of the one or more traffic lights, and the one or more other sources including the two or more vehicles relaying a message from the intersection, a municipal traffic center information system, or a commercial traffic center information system.

[0018] Based on the vehicle data of the two or more vehicles at the intersection, the processor determines a consensus between the location of one or more traffic lights and the verified location, which is obtained from the additional information using a standard deviation or a mean absolute deviation.

[0019] The processor provides the location of one or more traffic lights for the operation of autonomous vehicles, based on the consensus that is greater than a consensus threshold, and increases the number of two or more vehicles from which vehicle data will be obtained based on the consensus that is less than the consensus threshold.

[0020] According to one embodiment, the processor receives SPaT observations for the one or more traffic lights from the two or more vehicles based on the image information or images, in order to determine, based on a consensus among the SPaT observations of the two or more vehicles that is greater than a first threshold, whether the SPaT of the one or more traffic lights is static, so that the SPaT is always the same, or dynamic, so that the SPaT changes based on the traffic flow or the time of day.

[0021] The processor receives motion-based SPaT for the one or more traffic lights based on the vehicle location and vehicle speed specified in the vehicle data of the two or more vehicles, and determines a consensus between the SPaT observations and the motion-based SPaT.

[0022] Based on the consensus between the SPaT observations and the motion-based SPaT, which is greater than a second threshold, the processor provides the SPaT of one or more traffic lights, assuming that the SPaT of one or more traffic lights is static, and provides the SPaT of one or more traffic lights together with a confidence value that decreases with reduced recency of vehicle data reporting to the management system, assuming that the SPaT of one or more traffic lights is dynamic.

[0023] According to one embodiment, the processor determines the lane correspondence based on the vehicle location and image information or images of the two or more vehicles.

[0024] According to one embodiment, the processor provides at least one of the three types of information about the one or more traffic lights for the operation of autonomous vehicles as an application programming interface (API) or as an extended map.

[0025] The features and advantages mentioned above, as well as other features and advantages of the invention, are easily apparent from the following detailed description when taken in conjunction with the accompanying drawings.

[0026] Further features, advantages and details appear only as examples in the following detailed description, which refers to the drawings in which: Fig. 1 a block diagram of a system that facilitates traffic signal management for autonomous vehicle operation, according to one or more embodiments; Fig. 2 a procedure flow of a method for carrying out traffic signal management for autonomous vehicle operation by determining the location of a traffic light, according to one or more embodiments; Fig. 3 a process flow of a procedure for carrying out a traffic signal management for autonomous vehicle operation by determining a static time pattern of a traffic light, according to one or more embodiments; Fig. 4 a process flow of a procedure for carrying out a traffic signal management for autonomous vehicle operation by determining a dynamic time pattern of a traffic light, according to one or more embodiments; Fig. 5 processes are shown that relate to Fig. 3 and Fig. 4. Further detail the discussed consensus determination; and Fig. Figure 6 shows an exemplary map that is extended based on a traffic signal management system for autonomous vehicle operation, according to one or more embodiments.

[0027] The following description is merely exemplary and is not intended to limit the present invention, its application, or uses. It should be understood that throughout the drawings, corresponding reference numbers point to identical or corresponding parts and features.

[0028] As previously mentioned, autonomous vehicles require information conveying the real-time scenario and must act or react to this information in a timely manner. Any prior knowledge that reduces the real-time processing required by the autonomous vehicle can ensure the timely actions necessary for safe and efficient operation. Implementations of the systems and methods described here relate to traffic signal management for autonomous operation. The traffic signal management can be cloud-based, as described in the exemplary implementation presented here. In the specific context of an intersection with traffic lights, three types of information can be provided to an approaching autonomous vehicle. This information can be in the form of an application programming interface (API) or used to augment the map that autonomous vehicles use for navigation.

[0029] The first of the three types of information is the traffic light's location in global coordinates (i.e., latitude, longitude, altitude). This information allows for faster detection of the traffic light's location within the autonomous vehicle's field of view. Such information reduces the time and computational effort required to locate the traffic light. The second type of information is the traffic light's timing model (e.g., signal phase and timing (SPaT)). This information specifies the duration of each traffic light and can aid in decision-making (e.g., speed control before approaching an intersection) or enable a faster response to a traffic light turning green (e.g., activating the start / stop function or resuming automatic driving from a standstill). The third type of information specifies the particular traffic light (i.e.,A set of two or three traffic lights corresponds to each lane at an intersection if there is more than one set of traffic lights at a given intersection. The three types of information are independent of each other, so only one or two of the information types may be available at a given intersection, instead of all three.

[0030] According to an exemplary embodiment, Fig. Figure 1 shows a block diagram of a system that enables traffic signal management for autonomous vehicle operation. Three example vehicles 100a, 100b, and 100c (generally referred to as 100) are shown. Any or all of the vehicles 100 can be operated autonomously. One of the example vehicles 100a is a truck, and the other two example vehicles 100b and 100c are cars. As shown for vehicle 100c, each of the vehicles 100 can contain one or more cameras 105 and a vehicle controller 107. The in Fig. The exemplary cameras 105 shown in Figure 1 are not intended to limit the number or locations of cameras 105 in alternative configurations. As further explained, one or both cameras 105 can provide raw images for processing or provide information from the acquired images. Fig. Figure 1 shows two sets of traffic lights 130a, 130b (generally referred to as 130) and indicates that both are in the exemplary field of view (FOV) shown in dashed lines for cameras 105.

[0031] The vehicle controller 107 can be a collection of vehicle controllers 107 that communicate with each other to perform the functionality discussed here. The vehicle controller 107 can perform or facilitate both communication functions and autonomous control functions. The communication can include vehicle-to-infrastructure (V2I) communication, which involves the exchange of DSRC (dedicated short range communication) messages 135 with a smart intersection 140, such as the one that controls the traffic light 130 in the example of Fig. 1 includes. The DSRC messages 135 can transmit specifications of the traffic light 130, such as the time pattern for the operation of the traffic light or SPaT, or local intersection data indicating which lane 601 ( Fig. 6) which traffic light corresponds to 130.

[0032] The communication carried out or facilitated by the vehicle control unit 107 of the vehicle 100 can also include the provision of vehicle data 115 to a management system 110, which can be implemented, for example, using a cloud-based server. For instance, the vehicle 100 can provide images or information based on processed images using one or both cameras 105. The vehicle control unit 107 can receive other information via a Controller Area Network (CAN) to communicate with the management system 110. This other information can include the vehicle 100's location based on a Global Navigation Satellite System (GNSS) (e.g., Global Positioning System (GPS)), speed, yaw rate, and other location and motion data. The vehicle data 115 can also include the DSRC messages 135 from the smart intersection 140, which are forwarded to the management system 110.The vehicle control unit 107 and the management system 110 may contain a processing circuit that may include an application-specific integrated circuit (ASIC), an electronic circuit, a processor (shared, dedicated, or grouped), memory that executes one or more software or firmware programs, a combined logic circuit, and / or other suitable components that provide the functionality described. The vehicle control unit 107 and the management system 110 also contain a communication module for performing wireless communication according to known communication protocols and schemes.

[0033] The management system 110 can obtain vehicle data 115 from many vehicles 100, as in Fig. 1 specified. In this way, the management system 110 can collect data (e.g., location, speed, dwell period, dwell time) from the vehicles 100, which allows the management system 110 to derive the SPaT or lane correspondence of the traffic light 130 based on the movement pattern of the vehicles 100. In addition to the vehicle 100 information, the management system 110 can exchange information 125 with a commercial or municipal traffic center information system 120. The traffic center information system 120 can provide similar information to the DSRC messages 135 from a smart intersection 140. For example, a commercial entity or municipality can conduct a survey and specify the location of the traffic light 130. As mentioned earlier, the management system 110 can use the vehicle data 115 and information 125, and the processing of the vehicle data 115 and information 125, to create an API or an extended map 600 ( Fig. 6) to provide for autonomous vehicles 100. As also mentioned previously, one or more of three types of information relating to traffic light 130 may be of particular interest. These three types of information include the global location of traffic light 130, the SPaT (Speed ​​and Time of Change), which indicates the temporal pattern of the light changes, and the lane correspondence of each traffic light 130.

[0034] Fig. 2 is a process flow of a method 200 for implementing traffic signal management for autonomous vehicle operation according to one or more embodiments. Specifically, this refers to the following: Fig. The two described procedures 200 relate to the first type of information, which specifies a global location (latitude, longitude, altitude) of the traffic light 130. The processes can be carried out by the management system 110, and the application of the processes to a given intersection with one or more traffic lights 130 is discussed for explanatory purposes.

[0035] In Block 205, receiving an input can involve multiple sources. In Block 210, receiving smart intersection information involves receiving the location of traffic signal 130 (e.g., based on a survey) at smart intersection 140 from vehicles 100. If the given intersection is a smart intersection 140, the vehicles 100 receive V2I communication (e.g., DSRC messages) from the smart intersection 140 and then forward the information to the management system 110. The number of vehicles 100 uploading this information to the management system 110 can be limited (e.g., every second vehicle 100 or every tenth vehicle 100, depending on the traffic volume at smart intersection 140). This selective notification can be controlled by the management system 110 via communication with the vehicles 100 to manage communication and computing resources.

[0036] Receiving inputs in Block 205 can also include receiving vehicle data 115 in Block 220. Vehicle data 115 includes CAN data (e.g., GPS location, speed) and image information or images from vehicles 100. The vehicle data 115 provided to the management system 110 by each vehicle 100 can be restricted to vehicle data 115 collected within a detection region (e.g., 5 to 15 feet before an intersection of interest). This improves system accuracy by collecting samples only in areas where the camera-based range estimation is more accurate. Additionally, the number of vehicles 100 providing vehicle data 115 can be controlled by the management system 110 in a selective reporting scheme similar to the one discussed with respect to Block 210.The image information or images that are part of the vehicle data 115 can be obtained from different cameras 105 of each vehicle 100. For example, image information can be information obtained from one camera 105 based on images, while images can be raw images obtained from another camera 105 that allow for a three-dimensional (3D) reconstruction of the scene. One or both types of cameras 105 can be available in a given vehicle 100.

[0037] Receiving inputs in Block 205 can also include receiving traffic center information from a commercial and / or municipal traffic center information system 120 in Block 230. Like the smart intersection 140, the commercial and / or municipal traffic center information system 120 can also provide a verified location of the traffic light 130 or at least the intersection. The traffic center information system 120 can be cloud-based and can perform cloud-to-cloud communication with the management system 110. In Block 240, determining the latitude, longitude, and altitude of one or more traffic lights 130 at the given intersection refers to determining the location based on each of the various inputs available in Block 205.In the case of intelligent intersection information (at block 210) and traffic center information (at block 230), the location of traffic light 130 can be specified, or at least its location at the time of the check. In the case of vehicle data 115 (at block 220), the location of traffic light 130 can be determined, for example, from images or image information. This means that the management system 110 can additionally process the vehicle data 115 to determine the location. The CAN data that is part of the vehicle data 115 (e.g., GPS location of vehicle 100) facilitates the determination of which traffic light 130 the location information refers to.

[0038] In block 250, the quantification of confidence in the location of traffic light 130 refers to the cross-validation of the determined location (in block 240) from the various input sources (in block 205). A standard deviation, a mean absolute deviation, or another indicator can be derived from the location provided by each input source. The vehicles 100 can be processed collectively (e.g., the mean of the location provided by each source) or individually for the purpose of cross-validation of the location.

[0039] Block 260 checks whether the confidence exceeds a threshold. If so, the location specified in Block 265 refers to the management system 110, which provides the location of the traffic light 130 at the given intersection via an API or on a map 600 ( Fig. 6) specifies, e.g., for use by autonomous vehicles, 100. If, based on the check in block 260, the confidence does not exceed the threshold, then in block 270 the increase in the number of reporting vehicles to 100 refers to the control of selective reporting.

[0040] Specifically, more vehicles can be instructed to report intelligent intersection information (at block 210) or vehicle data (at block 220).

[0041] Fig. 3 is a process flow of a method 300 for implementing traffic signal management for autonomous vehicle operation according to one or more embodiments. Specifically, this refers to Fig. The three depicted procedures 300 relate to the second type of information, which indicates a time pattern or SPaT of the traffic light 130. The processes can be carried out by the management system 110, and the application of the processes to a given intersection with one or more traffic lights 130 is discussed for explanatory purposes. In contrast to the determination of the first type of information, the location, which is determined with reference to Fig. As discussed in section 2, determining the SPaT (time pattern) for a traffic light (traffic light 130) is complicated by the fact that a traffic light 130 can have a static or a dynamic SPaT. A static SPaT means that the time pattern according to which the traffic light 130 operates is always the same. A dynamic SPaT means that the time pattern according to which the traffic light 130 operates changes. The change can, for example, be based on the time of day, so that there are two or more specific patterns. Alternatively, a dynamic SPaT can refer to a traffic light 130 that is operated dynamically according to the traffic flow without any specific pattern.

[0042] Once a SPaT has been determined for a specific traffic light 130, the SPaT can be verified and maintained as long as updated SPaT information is provided to the management system 110 within a defined time (i.e., frequently enough). If the updated SPaT information is not provided within the defined time, or if an SPaT has not yet been determined for a specific traffic light 130, the SPaT will be revoked. Fig. The 3 processes shown are carried out. In block 310, the receipt of vehicle data 115 from vehicles 100 refers to the management system 110, which receives CAN data (e.g. GPS location, speed) and image information or images from vehicles 100.

[0043] Block 320 checks whether the vehicle data 115 (received in Block 310) is sufficiently up-to-date. If the vehicle data 115 is not sufficiently up-to-date, then any SPaT that was available for the given traffic light 130 is made unavailable in Block 325. This is because the SPaT has not been verified and no current data is available to verify the SPaT that the management system 110 previously provided for the given traffic light 130. If the vehicle data 115 is sufficiently up-to-date (based on the check in Block 320), Block 330 determines whether there is a consensus among the SPaT observations for a specific duration. SPaT observations are the SPaT determinations at each of the vehicles 100 for the given traffic light 130, as determined by images or image processing. The specified duration can be, for example, two days. The duration makes it easier to determine whether the given traffic light 130 has a static or dynamic SPaT (speed-to-time). Consensus can, for example,based on determining whether the standard deviation or another metric is below a threshold.

[0044] If the check in block 330 finds that there is no consensus between the SPaT observations of different vehicles 100 over a certain duration, the given traffic light 130 can be treated as a dynamic SPaT in block 335. This is done with reference to Fig. 4. This is discussed further. If it is determined (in Block 330) that there is a consensus among the SPaT observations of vehicles 100 over the specified duration, then Block 340 determines whether the SPaT observations agree with motion-based SPaT for the same traffic light 130. Motion-based SPaT refers to the management system 110, which derives the time pattern of traffic light 130 based on motion information (e.g., how long vehicle 100 was stopped at the location of traffic light 130) supplied by vehicles 100 as part of the vehicle data 115. The consensus in Block 340 can be determined based on the standard deviation or a similar metric. If the check in Block 340 indicates that there is no consensus between SPaT observations and motion-based SPaT, then the process of making a previous SPaT unavailable is carried out in Block 325.If the check at block 340 indicates that there is a consensus between SPaT observations and motion-based SPaT, then the management system 110 provides the SPaT of the traffic light 130 via an API or on a map 600 (. Fig. 6) available, e.g. for use by autonomous vehicles 100 at Block 345.

[0045] Fig. 4 is a process flow of a method 400 for implementing traffic signal management for autonomous vehicle operation according to one or more embodiments. Specifically, this refers to the following: Fig. The fourth depicted procedure 400 relates to the second type of information, which indicates a time pattern or SPaT of the traffic light 130, and more precisely, procedure 400 refers to a dynamic SPaT. To determine a dynamic SPaT, frequency (e.g., a minimum rate of the management system 110 for recording observations of vehicles 100), recency (e.g., the age of the observations reported to the management system 110), and regularity (e.g., a consistent distribution of the observations reported to the management system 110) are important factors. This is because, with frequent, current, and regular inputs, the management system 110 can treat a dynamic SPaT like a static SPaT with a limited duration. As with block 340 ( Fig. 3) In this discussion, motion-based SPaT and SPaT observations are examined. A decreasing confidence estimate can be provided if the two do not agree. Furthermore, 110 parameters of the traffic control cycle can be made available to the management system to support the validation of SPaT. The traffic control cycle parameters can, for example, include a minimum time for each phase for a given speed limit.

[0046] In block 405, the vehicle data 115 is received from the vehicles 100 as input into the management system 110. This is similar to the procedure in block 310 ( Fig. 3) Block 410 checks whether the inputs received by management system 110 from vehicles 100 are sufficiently frequent. Sufficient frequency can be determined empirically, for example. If the inputs are not frequent enough, any previous SPaT provided by management system 110 for the given traffic light 130 is deactivated in block 415. If the check in block 410 shows that the inputs are received by management system 110 frequently enough, then the application of the static SPaT procedure in block 420 refers to the application of procedure 300, which is based on... Fig. 3 is discussed. As already mentioned, sufficiently frequent entries into the management system 110 facilitate the treatment of a dynamic SPaT like a static SPaT. However, there is an additional feature related to the decreasing confidence value of the SPaT, as explained.

[0047] Block 430 checks whether the SPaT observations of vehicles 100 match the movement-based SPaT determination by the management system 110. This is similar to the check in Block 340 ( Fig. 3) The determination of conformity can be based, for example, on a standard deviation threshold or another metric. The in Fig. Procedure 300, as shown in block 345, indicates that SPaT is made available for use by autonomous vehicles 100 based on a consensus between SPaT observations and motion-based SPaT (at block 340). However, with dynamic SPaT, the outcome is less straightforward. If the check in block 430 reveals that the SPaT observations do not match the motion-based SPaT, then the process of making the previously available SPaT unavailable is carried out in block 415. However, if the check in block 440 shows a match between SPaT observations and motion-based SPaT, then the processes in block 440 are carried out. In block 440, the SPaT is made available for use by autonomous vehicles 100, but with a confidence level. This confidence level is based on the timeliness and regularity of the inputs to the management system 110.Therefore, the process in block 440 also includes checking the recency of the inputs and reducing the confidence level as the recency of the inputs decreases. If the check in block 450 shows that the confidence is not greater than a threshold, the process of making previously available SPaTs unavailable is carried out in block 415. As long as the confidence level is greater than the threshold (according to the check in block 450), the processes in block 440 continue.

[0048] Fig. Figure 5 shows the processes 500, which determine consensus in block 340 ( Fig. 3) and Block 430 ( Fig. 4) Further details. Thus, further reference is made to the following: Fig. 3 and Fig. The processes shown in block 4 are referenced. In block 510, the collection of inputs refers to obtaining vehicle data 115 from vehicles 100 in block 310 or block 405. The recency or regularity of the inputs can be checked if necessary. In block 520, the generation of a training dataset and a validation dataset involves using SPaT observations as the training dataset and motion-based SPaT as the validation dataset. In block 530, matching SPaT models with the training dataset facilitates the search for the most suitable model. In block 540, validating the most suitable model with the validation dataset leads to a confidence level (e.g., based on a standard deviation or another metric).In block 550, the removal of outlier data based on the validation result (in block 540) refers to the removal of outliers from the input to the management system 110 when the confidence level is, for example, below a threshold.

[0049] Fig. Figure 6 shows an exemplary map 600, which is extended by the management system 110 to display one or more of the three types of information according to one or more embodiments. Three lanes 601a, 601b, and 601c (generally referred to as 601) are shown. The direction of travel or turning for each lane 601 is indicated. Each lane 601 is associated with a virtual traffic control device (VTCD) VTCD1, VTCD2, and VTCD3. The assignment of the VTCDs to the lanes 601, the SPaT of the VTCDs, and the global location of each VTCD (i.e., latitude, longitude, altitude) can be displayed on the map 600. Other information (e.g., speed limits) can also be indicated.

[0050] The third type of information, lane localization or the specific traffic light 130 that corresponds to each lane 601, can be considered an extension of the information relating to Fig. The process flow discussed in sections 2-5 can be determined. Specifically, the CAN data acquired in block 220 provides, for example, the location and movement of each reporting vehicle 100. In an example scenario, each lane 601 can have a corresponding traffic light 130. In this case, the location of each traffic light 130 (i.e., the first type of information related to Fig.2 (discussed) can be used to assign the traffic light to a specific lane 601. Even if two or more lanes 601 are controlled by the same traffic light 130, the image information or raw images supplied by the vehicles 100 can be used by the management system 110 to assign the lanes 601 to the traffic light 130. If SPaT information is available, the movement of the vehicles 100 on the different lanes 601, obtained through CAN data that is part of the input to the management system 110, can be used to assign a specific traffic light 130 to a specific lane 601. Alternatively, lane localization can also be achieved without determining the first type (global location) or the second type (SPaT) of information. The management system 110 can use the vehicle data 115 (e.g.,Use images (image information) to link lanes 601 with traffic light 130 without additionally determining the location of traffic lights 130.

Claims

[1] Method (200) for implementing traffic signal management for the operation of autonomous vehicles, the method comprising: (205, 220) Received, at an administrative system (110), of vehicle data (115) from two or more vehicles (100) at an intersection with one or more traffic lights (130), wherein the vehicle data (115) includes a vehicle location, a vehicle speed and image information or images; (200, 300, 600) Determine, on the management system (110), at least one of three types of information about the one or more traffic lights (130) based on the vehicle data (115), wherein the three types of information include a location of the one or more traffic lights (130), a signal phase and timing (SPaT) of the one or more traffic lights (130), and a lane correspondence of the one or more traffic lights (130); and (340, 440) Providing, through the management system (110), at least one of the three types of information about the one or more traffic lights (130) for the operation of autonomous vehicles (100); wherein determining the location (240) of the one or more traffic lights (130) includes the use of the image information or images and the use of the vehicle location from the vehicle data (115) obtained from the two or more vehicles (100) at the intersection, and wherein the method (200) also includes: Received, at the management system, of additional information from one or more other sources, wherein the additional information includes a verified location of the one or more traffic lights (130) and the one or more other sources include the two or more vehicles (100) relaying a message from the intersection, an urban traffic center information system or a commercial traffic center information system (230), (250) Determining a consensus between the location of the one or more traffic lights (130), determined using the vehicle data (115) of the two or more vehicles (100) at the intersection, and the verified location, obtained from the additional information using a standard deviation or a mean absolute deviation, and (260) Proceeding with the provision of the location of one or more traffic lights (130) for the operation of autonomous vehicles (100) on the basis of the consensus which is greater than a consensus threshold, and (270) Increasing the number of the two or more vehicles (100) from which the vehicle data (115) are obtained, based on the consensus being less than the consensus threshold. [2] Method (200) according to claim 1, further comprising (310) Receiving SPAt observations for the one or more traffic lights (130) from the two or more vehicles (100) based on the image information or images, and (320, 330) Determine, based on a consensus among the SPaT observations of the two or more vehicles (100) that is greater than a first threshold, whether the SPaT of the one or more traffic lights (130) is static, so that the SPaT is always the same, or dynamic, so that the SPaT changes depending on the traffic flow or time of day, Obtain motion-based SPaT for the one or more traffic lights (130) based on the vehicle location and the vehicle speed specified in the vehicle data (115) of the two or more vehicles (100), and Determining a consensus between the SPaT observations and the motion-based SPaT, and based on the consensus between the SPaT observations and the motion-based SPaT, which is greater than a second threshold, proceed with providing the SPaT of the one or more traffic lights (130), based on the SPaT of the one or more traffic lights (130) that is static, and Continue by providing the SPaT of the one or more traffic lights (130) together with a confidence value that falls with reduced timeliness of the reporting of vehicle data (115) to the management system, based on the SPaT of the one or more traffic lights (130) which is dynamic. [3] Method (200) according to claim 1, further comprising (600) determining the lane correspondence based on the vehicle location and the image information or images of the two or more vehicles (100). [4] Method (200) according to claim 1, wherein the provision of at least one of the three types of information about the one or more traffic lights (130) for the operation of autonomous vehicles (100) is carried out as an application programming interface (API) or as an extended map. [5] System (110) for performing traffic signal management for the operation of autonomous vehicles (100), wherein the system (110) comprises: a communication module configured to receive vehicle data (115) from two or more vehicles (100) at an intersection with one or more traffic lights (130), wherein the vehicle data (115) includes a vehicle location, a vehicle speed and image information or images; and a processor configured to determine at least one of three types of information about the one or more traffic lights (130) based on the vehicle data (115), wherein the three types of information include a location of the one or more traffic lights (130), a signal phase and timing (SPaT) of the one or more traffic lights (130) and a lane correspondence of the one or more traffic lights (130), and that it provides the at least one of the three types of information about the one or more traffic lights (130) for the operation of autonomous vehicles (100); wherein the processor is configured to determine the location of one or more traffic lights (130) using the image information or images and using the vehicle location from the vehicle data obtained from the two or more vehicles (100) at the intersection, wherein the communication module is further configured to receive additional information from one or more other sources, wherein the additional information includes a verified location of the one or more traffic lights (130) and the one or more other sources include the two or more vehicles (100) relaying a message from the intersection, a municipal traffic center information system or a commercial traffic center information system (120), wherein the processor is further configured to determine a consensus between the location of the one or more traffic lights (130) using the vehicle data (115) from the two or more vehicles (100) at the intersection and the verified location obtained from the additional information using a standard deviation or a mean absolute deviation, and wherein the processor is configured to provide the location of the one or more traffic lights (130) for the operation of autonomous vehicles (100) based on the consensus greater than a consensus threshold, and is configured to increase the number of the two or more vehicles (100) from which the vehicle data is obtained based on the consensus less than the consensus threshold. [6] System (110) according to claim 5, wherein the processor is configured to receive SPaT observations for the one or more traffic lights (130) from the two or more vehicles (100) based on the image information or images, and that, based on a consensus among the SPaT observations from the two or more vehicles (100) that is greater than a first threshold, it determines whether the SPaT of the one or more traffic lights (130) is static, so that the SPaT is always the same, or dynamic, so that the SPaT changes based on the traffic flow or the time of day, wherein the processor is configured to obtain motion-based SPaT for the one or more traffic lights (130), based on the vehicle location and vehicle speed specified in the vehicle data from the two or more vehicles (100), and to determine a consensus between the SPaT observations and the motion-based SPaT, and wherein, based on the consensus between the SPaT observations and the motion-based SPaT, which is greater than a second threshold, the processor is configured to provide the SPaT of one or more traffic lights (130) based on the SPaT of one or more traffic lights (130) that is static, and to provide the SPaT of one or more traffic lights (130) together with a confidence value that decreases with reduced recency of the reporting of vehicle data to the management system based on the SPaT of one or more traffic lights (130) that is dynamic. [7] System (110) according to claim 5, wherein the processor is configured to determine the lane correspondence based on the vehicle location and the image information or images of the two or more vehicles (100). [8] System (110) according to claim 5, wherein the processor is configured to provide at least one of the three types of information about the one or more traffic lights (130) for the operation of autonomous vehicles (100) as an application programming interface (API) or as an extended map.

Images