A control method and device for a highway network connected autonomous vehicle

Abstract

The embodiment of the specification discloses a control method for expressway network-connected automatic driving vehicles, comprising the following steps: a cloud control platform acquires vehicle driving data of the network-connected automatic driving vehicles and road environment data collected by a roadside sensing device; whether any two adjacent network-connected automatic driving vehicles satisfy a preset multi-vehicle queue driving condition is judged according to the road environment data and the driving data of the network-connected automatic driving vehicles; if yes, a first control instruction of multi-vehicle queue driving is generated; if not, a second control instruction of single-vehicle driving is generated; the first control instruction or the second control instruction is sent to the any two adjacent network-connected automatic driving vehicles, so that the network-connected automatic driving vehicles automatically drive according to the corresponding control instruction; based on this, the driving efficiency and safety of the network-connected automatic driving vehicles on the expressway are improved, and the use cost of the vehicles is also reduced.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, specifically to a control method and device for a highway connected autonomous driving vehicle. Background Technology

[0002] Vehicle-road cooperative autonomous driving technology enables comprehensive coordination between vehicles, roads, the cloud, and pedestrians through efficient communication between vehicles, roads, pedestrians, and the cloud. It meets the application needs of different levels of connected autonomous vehicles, such as assisted driving and medium-to-high-level autonomous driving, thereby promoting the development goals of optimizing individual autonomous vehicles and optimizing the overall traffic system.

[0003] Vehicle-road cooperative technology is most easily implemented in closed, structured roads, and highways are a typical example of closed, structured roads. Therefore, the application of connected autonomous driving technology on highways is conducive to the rapid commercialization of vehicle-road cooperative technology. While existing technologies have linked vehicle-road collaboration, effective collaboration in application remains lacking. For example, on highways, when a connected autonomous vehicle activates its autonomous driving mode, the presence of non-connected autonomous vehicles along the route causes the connected autonomous vehicle to be unable to follow its planned trajectory during interactions. It needs to continuously adjust its speed and heading angle, placing operational pressure on the onboard server and thus reducing safety in autonomous driving mode. Furthermore, under the same autonomous driving capabilities, there are inherent contradictions between the safety, applicability, and economic viability of connected autonomous vehicles. To improve the safety of connected autonomous vehicles, measures such as adding higher-precision sensors are often employed. However, the relatively high price of these high-precision sensors increases the production and usage costs of connected autonomous vehicles, reducing their economic viability and hindering the large-scale commercialization of autonomous driving technology.

[0004] In order to promote the large-scale and commercial application of autonomous driving technology, there is an urgent need for an autonomous driving technology that is safer and more economical, so as to improve the level of autonomous driving capabilities.

[0005] In view of this, the present invention proposes a control method and device for connected autonomous vehicles on highways. By using a cloud control platform, it provides vehicle guidance services for connected autonomous vehicles, which can improve the safety of connected autonomous vehicles during driving and reduce the cost of vehicle use. Summary of the Invention

[0006] This specification provides a control method and apparatus for a connected autonomous vehicle on highways, addressing the problem that existing methods of adding expensive sensing devices to improve the safety of autonomous vehicles increase the operating costs of connected autonomous vehicles, resulting in reduced economic efficiency despite ensuring safety. In other words, if the existing technology does not add more expensive and better-performing sensor devices, the maintenance and operating costs of the vehicle will not increase, but this will not improve the safety of the autonomous vehicle, and consequently, it will not improve the driving efficiency of the autonomous vehicle on highways.

[0007] To solve the above-mentioned technical problems, the embodiments in this specification are implemented as follows:

[0008] Firstly, embodiments of this specification provide a control method for a highway connected autonomous driving vehicle, characterized by comprising:

[0009] The cloud control platform acquires vehicle driving data from connected autonomous vehicles and road environment data collected by roadside sensing devices;

[0010] Based on the road environment data and the driving data of the connected autonomous vehicles, determine whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions.

[0011] If so, then based on the road environment data and the vehicle driving data of the connected autonomous vehicles, a first control command for multi-vehicle platoon driving is generated; the first control command is sent to any two adjacent connected autonomous vehicles, so that the two adjacent connected autonomous vehicles drive as the same platoon.

[0012] If not, then based on the road environment data and the driving data of the connected autonomous vehicles, a second control command for single-vehicle driving is generated; the second control command is sent to any two adjacent connected autonomous vehicles, so that the two adjacent connected autonomous vehicles drive in single-vehicle driving mode.

[0013] Secondly, embodiments of this specification provide a control device for a highway connected autonomous driving vehicle, including:

[0014] The acquisition module is used by the cloud control platform to acquire vehicle driving data of the connected autonomous vehicle and road environment data collected by roadside perception devices;

[0015] The judgment module is used to determine, based on the road environment data and the driving data of the connected autonomous vehicles, whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions.

[0016] The first generation module is configured to, if so, generate a first control command for multi-vehicle platooning based on the road environment data and the vehicle driving data of the connected autonomous vehicles; and send the first control command to any two adjacent connected autonomous vehicles, so that the two adjacent connected autonomous vehicles can drive in the same platoon.

[0017] The second generation module is used to generate a second control command for single-vehicle driving based on the road environment data and the driving data of the connected autonomous vehicles if no; and to send the second control command to any two adjacent connected autonomous vehicles so that the two adjacent connected autonomous vehicles drive in single-vehicle driving mode.

[0018] The embodiments of this specification, employing at least one of the above-described technical solutions, can achieve the following beneficial effects:

[0019] The system acquires vehicle driving data from connected autonomous vehicles and road environment data collected by roadside sensing devices through a cloud control platform. Based on the road environment data and the driving data of the connected autonomous vehicles, it determines whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions. If so, a first control command for multi-vehicle platoon driving is generated based on the road environment data and the vehicle driving data of the connected autonomous vehicles. The first control command is sent to any two adjacent connected autonomous vehicles, causing them to drive in the same platoon. If not, a second control command for single-vehicle driving is generated based on the road environment data and the driving data of the connected autonomous vehicles. The second control command is sent to any two adjacent connected autonomous vehicles, causing them to drive in a single-vehicle driving mode. Based on this, the driving efficiency and safety of connected autonomous vehicles on highways are improved, while also reducing vehicle operating costs. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the accompanying drawings used in the description of the embodiments or prior art will be briefly introduced below. The illustrative embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation of this application. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0021] Figure 1 This is a flowchart illustrating a control method for a highway connected autonomous driving vehicle provided in the embodiments of this specification.

[0022] Figure 2 This is a schematic diagram of the process architecture of a multi-vehicle platooning method for highway connected autonomous vehicles provided in the embodiments of this specification.

[0023] Figure 3 This is a schematic flowchart illustrating a lane-changing method for a connected autonomous vehicle on a highway, as provided in the embodiments of this specification.

[0024] Figure 4 This is a schematic diagram of the process architecture of a lane-changing method for a highway connected autonomous vehicle provided in the embodiments of this specification.

[0025] Figure 5 This is a schematic diagram of the cumulative reward in a cooperative lane-changing method for a highway connected autonomous vehicle provided in the embodiments of this specification.

[0026] Figure 6 This is a schematic diagram of the action space of an intelligent agent in multi-vehicle cooperative lane changing for a connected autonomous vehicle on a highway, provided by an embodiment of this specification.

[0027] Figure 7 This is a schematic diagram illustrating the steps of Monte Carlo search tree in multi-vehicle cooperative lane changing for connected autonomous vehicles on highways, as provided in the embodiments of this specification.

[0028] Figure 8 This is a schematic diagram illustrating the multi-agent action simulation of a multi-vehicle cooperative lane-changing process for a connected autonomous vehicle on a highway, as provided in the embodiments of this specification.

[0029] Figure 9 This is a schematic diagram illustrating the random simulation of multi-vehicle actions in a control method for a highway connected autonomous driving vehicle provided in the embodiments of this specification.

[0030] Figure 10 This is a schematic diagram of a control method for a highway connected autonomous driving vehicle provided in the embodiments of this specification, applied to a multi-vehicle collaborative service system in a highway unmanned logistics scenario.

[0031] Figure 11 This is a schematic diagram of a control device for a highway connected autonomous driving vehicle provided in the embodiments of this specification. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0033] It should be noted that all expressions such as "first", "second", "third", and "fourth" used in the embodiments of the present invention are for the purpose of distinguishing two entities with the same name but different names or different parameters. It can be seen that "first", "second", "third", and "fourth" are only for the convenience and clarity of expression. Furthermore, the functional modules corresponding to the devices used in the embodiments of the present invention can be one set, two sets, or more sets, depending on the requirements. They should not be construed as limiting the embodiments of the present invention, and subsequent embodiments will not describe them one by one.

[0034] The technical solutions provided in the various embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0035] In existing technologies, although vehicle-road collaboration has been established, effective collaboration is still lacking in practical applications. For example, on highways, when a connected autonomous vehicle activates its autonomous driving mode, it cannot follow its original trajectory when interacting with non-connected autonomous vehicles along the path. It needs to continuously adjust its speed and heading angle, placing operational pressure on the onboard server and thus reducing safety in autonomous driving mode. To address this issue, measures such as adding sensors with higher measurement accuracy are employed to improve the safety of single-vehicle intelligent autonomous driving. However, the relatively high price and maintenance difficulty of these high-precision sensors increase the production and usage costs of connected autonomous vehicles, hindering the commercialization and large-scale development of autonomous driving technology.

[0036] To address the shortcomings of existing technologies, this solution provides the following embodiments:

[0037] Figure 1This is a flowchart illustrating a control method for a highway connected autonomous vehicle provided in an embodiment of this specification. It should be noted that, from a procedural perspective, the executing entity of this application can be a cloud control platform or other service platforms equipped with the corresponding algorithm of this method; the cloud control platform is a platform used to remotely provide services to autonomous vehicles. This platform can be a cluster of devices composed of multiple computers or servers, and can also be called an intelligent connected cloud control platform, an autonomous driving control platform, or a cloud control algorithm platform. The cloud control platform includes at least one or more of the following modules: simulation testing module, information interconnection module, data fusion module, standardization module, and cloud collaboration module, etc. This specification does not specifically limit the name of the platform; it only uses the cloud control platform as an embodiment of this solution for description. The connected autonomous vehicle described in this specification can also be a connected vehicle capable of receiving remote control commands. All other embodiments obtained by those skilled in the art based on the embodiments of this specification without creative effort should fall within the scope of protection of this application.

[0038] like Figure 1 As shown, the method may include the following steps:

[0039] Step 110: The cloud control platform acquires the vehicle driving data of the connected autonomous vehicle and the road environment data collected by the roadside perception device; the vehicle driving data of the connected autonomous vehicle includes vehicle status data, origin and destination.

[0040] In step 110, the cloud control platform acquires the vehicle driving data of the connected autonomous vehicle, and can also acquire road environment data collected by roadside perception devices on the road where the connected autonomous vehicle is traveling. The connected autonomous vehicle can be a connected autonomous vehicle with autonomous driving capabilities or a driverless vehicle. In this embodiment, the connected autonomous vehicle is one with autonomous driving capabilities. If the vehicle has its autonomous driving function activated, the cloud control platform can acquire autonomous driving strategies to better control the vehicle and complete autonomous driving services. The vehicle driving data of the connected autonomous vehicle includes the vehicle's status data during its journey on the road, data collected by the vehicle's own sensors, and the vehicle's planned route, including the origin and destination. The vehicle status data of the connected autonomous vehicle includes at least the vehicle's speed, heading angle, and whether the vehicle is in a state where it can drive normally.

[0041] In practical applications, after obtaining the above data, the cloud control platform can analyze the data based on the preset algorithms of the cloud control platform, formulate control strategies for connected autonomous vehicles, and issue control strategies to the corresponding connected autonomous vehicles. This reduces the running time of the connected autonomous vehicle's own algorithms, improves decision-making efficiency, and thus shortens the vehicle's response time and ensures safety.

[0042] Step 120: Based on the road environment data and the driving data of the connected autonomous vehicles, determine whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions.

[0043] In step 120, the cloud control platform uses the acquired road environment data and driving data of the connected autonomous vehicles to perform technical and planning functions based on the multi-vehicle collaborative algorithm pre-deployed by the cloud control algorithm platform, and uses the multi-vehicle platooning algorithm to determine whether any two adjacent connected autonomous vehicles can be platooned.

[0044] Specifically, preset conditions for whether to platoon vehicles can be set in advance. These preset conditions can be numerous, and this solution does not limit them to any specific conditions. In practical applications, platooning preset conditions can be set specifically for different road sections and driving areas.

[0045] As an example, the preset conditions in this solution may include preset conditions in the following scenarios.

[0046] Scenario 1: When the distance between two adjacent connected autonomous vehicles is short and there are no other non-connected autonomous vehicles between them, the preset condition can be distance, such as 100m.

[0047] Scenario 2: When two adjacent vehicles have the same speed and size, to avoid differences between vehicles, vehicles with significantly different models can be excluded from the same driving queue. For example, a preset condition can be set to uniformly queue connected autonomous freight vehicles, or a preset condition can be set to uniformly queue connected autonomous cars. In other words, to improve safety, a preset condition can be set to exclude cars from the freight vehicle queue.

[0048] Step 130: If so, then based on the road environment data and the vehicle driving data of the connected autonomous vehicles, generate a first control command for multi-vehicle platoon driving; send the first control command to any two adjacent connected autonomous vehicles, so that the two adjacent connected autonomous vehicles drive as the same platoon.

[0049] Step 140: If not, then generate a second control command for single-vehicle driving based on the road environment data and the driving data of the connected autonomous vehicles; send the second control command to any two adjacent connected autonomous vehicles so that the two adjacent connected autonomous vehicles drive in single-vehicle driving mode.

[0050] In steps 130 to 140, when it is determined that two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions, step 130 is executed. The cloud control platform generates a first control command for multi-vehicle platoon driving and sends it to the corresponding connected autonomous vehicle. After receiving the first control command, the connected autonomous vehicle controls the vehicle to drive in the same platoon according to the first control command.

[0051] As an example, the cloud control platform generates a first control command based on the road environment data and the vehicle driving data of the connected autonomous vehicles. This first control command includes remote control commands issued to multiple connected autonomous vehicles that can be platooned. For example, if three connected autonomous vehicles, A, B, and C, are traveling on a highway, the cloud control platform determines, based on the road environment data of the road they are traveling on and the vehicle driving data of the three vehicles, that the three vehicles meet the preset multi-vehicle platooning driving conditions and can be platooned. Then, based on the algorithm on the cloud control platform, a first control command is generated. The first control command includes control commands for connected autonomous vehicles A, B, and C respectively. The cloud control platform issues the control commands for connected autonomous vehicles A, B, and C respectively to connected autonomous vehicles A, B, and C. Based on this, the cloud control platform provides connected autonomous vehicles A, B, and C with control commands including vehicle speed, heading angle, and throttle opening, so that connected autonomous vehicles A, B, and C can travel in the same platoon.

[0052] Furthermore, when it is determined that two adjacent connected autonomous vehicles do not meet the preset multi-vehicle platoon driving conditions, step 140 is executed. The cloud control platform generates a second control command for single-vehicle driving of the connected autonomous vehicle and sends it to the corresponding connected autonomous vehicle. After receiving the second control command, the autonomous vehicle controls the vehicle to drive in single-vehicle driving mode according to the second control command.

[0053] As an example, the cloud control platform generates a second control command based on the road environment data and the vehicle driving data of the connected autonomous vehicles. This second control command includes single-vehicle driving control commands issued to multiple connected autonomous vehicles that cannot form a platoon. For example, if three connected autonomous vehicles, A', B', and C', are traveling on a highway, the cloud control platform determines, based on the road environment data of the road and the vehicle driving data of these three vehicles, that the three vehicles do not meet the preset multi-vehicle platooning conditions and cannot form a platoon. Therefore, based on... The algorithm on the cloud control platform generates a second control command; the second control command includes control commands for connected autonomous vehicles A', B', and C' respectively; the cloud control platform sends the control commands for connected autonomous vehicles A', B', and C' respectively to connected autonomous vehicles A', B', and C'; based on this, the cloud control platform provides connected autonomous vehicles A', B', and C' with control commands including vehicle speed, heading angle, and throttle opening, enabling connected autonomous vehicles A', B', and C' to drive in single-vehicle driving mode respectively.

[0054] Based on steps 110 to 140, the cloud control platform analyzes the driving data and road environment data of the connected autonomous vehicle to obtain driving control commands for the vehicle. This reduces the need for extensive data analysis by the connected autonomous vehicle, or even eliminates the need for data analysis altogether, thus lowering the operational burden on the vehicle's onboard server. Furthermore, by combining data collected from roadside perception devices, more road environment data is gathered, making the decisions made by the cloud control platform more consistent with reality and improving the driving safety of the connected autonomous vehicle. Based on the above method, by utilizing the cloud control platform, connected autonomous vehicles with a certain level of autonomous driving capability can achieve at least one level of improved autonomous driving capability without the need to install additional sensor data collection equipment. Compared to the traditional approach of adding higher-performance sensor equipment, this reduces the manufacturing and operating costs of the vehicle.

[0055] Specifically, in step 120, the cloud control platform includes a multi-vehicle platooning algorithm, which includes a platooning partitioning algorithm and a speed planning algorithm. The platooning partitioning algorithm is used to determine a multi-vehicle platooning partitioning strategy for the connected autonomous vehicles based on the road environment data and the driving data of the connected autonomous vehicles. The speed planning algorithm is used to determine a speed control strategy for the connected autonomous vehicles based on the road environment data and the driving data of the connected autonomous vehicles. Based on the vehicle platooning partitioning strategy and the speed control strategy, the first control command and the second control command are generated.

[0056] Specifically, the formation partitioning algorithm is used to determine whether each vehicle belongs to a formation. Of course, it can also further determine which formation a vehicle belongs to. The output of the formation partitioning algorithm is a sequence of queue numbers.

[0057] For example, in practical applications, a dedicated lane for connected autonomous vehicles can be set up to facilitate the autonomous driving control of these vehicles, enabling them to provide better services to people.

[0058] In the dedicated lane, the sequence number [1,0,2,2,2,……,11,11,11,11,0,12,0,0] has the following characteristics: the first element is 1, indicating that the first vehicle (the first vehicle in the front) is in the first platoon; the second element is 0, indicating that the second vehicle in the dedicated lane is a non-connected autonomous vehicle (non-connected autonomous vehicles may also appear in the dedicated lane, but these vehicles will not be included in the platoon or have their speed planned); and the third element is 2, indicating that the third vehicle in the dedicated lane is in the second platoon. The cloud control platform can remotely control vehicles in the dedicated lane whose elements are not 0, but cannot remotely control vehicles whose elements are 0.

[0059] The output of the formation partitioning algorithm can be the input of the speed planning algorithm, which outputs the target speed of all connected autonomous vehicles on the dedicated lane.

[0060] For example, in the sequence [17.9,-1,11.9,11.9,11.9,……,10.11,9.98,10.25,11.37,-1,7.39], the first element in the sequence is 1, indicating that the target speed of the first vehicle (the first car) in the dedicated lane is 17.9. The second element in the sequence is -1, indicating that the second vehicle in the dedicated lane is a non-connected autonomous vehicle, and no speed is issued to that vehicle.

[0061] Specifically, before determining whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions, the process includes: determining, based on the road environment data, whether any connected autonomous vehicles have entered or left the platoon, or whether platooning has not been performed at the current time, or whether a preset time has elapsed since the last platooning.

[0062] It should be noted that the term "connected autonomous vehicles" in this step is a general term for all connected autonomous vehicles on the road. These can be vehicles for which the cloud control platform has planned control commands or vehicles for which control commands have not been planned. This solution does not specifically limit the state information of connected autonomous vehicles. The term "any two adjacent connected autonomous vehicles" in this step refers to vehicles for which the cloud control platform plans to plan control commands, such as determining whether to form a platoon or issue speed control commands based on the driving data of these two adjacent connected autonomous vehicles.

[0063] The determination of whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions specifically includes: if yes, determining whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions; if no, directly calling the speed planning algorithm to determine the speed control strategy of the connected autonomous vehicles. It should be noted that directly calling the speed planning algorithm to determine the speed control strategy of the connected autonomous vehicles indicates that the original formation has not been disrupted, or the current formation does not need to be updated within a preset time. In this case, only the speed planning algorithm needs to be called to control the speed of the connected autonomous vehicles within the formation.

[0064] Furthermore, determining whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platooning driving conditions may include: determining the first driving distance between any two adjacent connected autonomous vehicles based on the road environment data; and determining whether the first driving distance between the two adjacent vehicles is less than or equal to a preset driving distance for vehicle platooning based on the first driving distance and the vehicle driving data of the two adjacent connected autonomous vehicles.

[0065] In the embodiments of this specification, vehicle information of a target vehicle whose first driving distance is less than or equal to a first preset distance can be obtained. The first driving distance can refer to the distance from the first vehicle to a preset time interval, such as 3 seconds. The time interval refers to the distance between the vehicle and the vehicle in front / the speed of the vehicle.

[0066] In practical applications, while ensuring driving safety, the preset distance can be adjusted as needed to guide vehicles into platooning more efficiently.

[0067] The step of generating a first control command for multi-vehicle platooning based on the road environment data and the vehicle driving data of the connected autonomous vehicle may include:

[0068] If the first driving distance between two adjacent vehicles is less than or equal to the preset driving distance for vehicle platooning, then the two adjacent connected autonomous vehicles are grouped into a platoon, and the queue number corresponding to each connected autonomous vehicle is updated.

[0069] As an example, in a dedicated lane, within the sequence [1,1,0,2,2,2,...], the first element is 1, indicating that the first vehicle (the first car) in the dedicated lane is in the first platoon; the second element is 1, indicating that the second vehicle in the dedicated lane is in the same platoon as the first vehicle; the third element is 0, indicating that the second vehicle in the dedicated lane is a non-connected autonomous vehicle; and the fourth to sixth elements are 2, indicating that the fourth to sixth vehicles in the dedicated lane are in the second platoon. At a certain moment, if the third element leaves the dedicated lane, the first travel distance between the original second and fourth elements is determined; if this distance is less than or equal to the preset travel distance for vehicle platooning, the two adjacent connected autonomous vehicles are grouped into one platoon, and the queue number corresponding to each connected autonomous vehicle is updated to [1,1,1,1,1,...].

[0070] Furthermore, the first control command for generating a multi-vehicle platoon may include: determining a first multi-vehicle platoon partitioning strategy using a platoon partitioning algorithm based on the road environment data and the vehicle driving data of the connected autonomous vehicles; and determining a first speed planning strategy using a speed planning algorithm based on the road environment data and the vehicle driving data of the connected autonomous vehicles, combined with the first multi-vehicle platoon partitioning strategy. The platoon partitioning algorithm is used to determine whether each vehicle belongs to a platoon, and if so, which platoon it belongs to; the output of the platoon partitioning algorithm is a sequence of platoon numbers.

[0071] In practical applications, the target speed of the vehicle can be determined and transmitted to the vehicle based on the vehicle's position and speed information relative to the vehicle in front. The calculation and transmission frequency of the target speed can both be 0.2 seconds per transmission. The vehicle can be equipped with a high-precision positioning unit and an onboard communication unit, which can upload the vehicle's accurate position and speed information to the cloud control platform. At the same time, the onboard communication unit can receive the target speed transmitted from the cloud control platform, thereby achieving precise and efficient communication between the vehicle and the cloud control platform.

[0072] The second control command for generating single-vehicle driving may include: determining the first single-vehicle queue division strategy using a platooning algorithm based on the road environment data and the vehicle driving data of the connected autonomous vehicle; and determining the second speed planning strategy using a speed planning algorithm based on the road environment data and the vehicle driving data of the connected autonomous vehicle, combined with the first single-vehicle queue division strategy. The platooning algorithm is also used to determine whether each vehicle belongs to a platoon. If it does not belong to any platoon, it is determined that the connected autonomous vehicle is driving in a single-vehicle queue, i.e., one vehicle in one platoon driving on a dedicated lane. The platooning method of this connected autonomous vehicle corresponds to the first single-vehicle queue division strategy. The cloud control platform provides speed planning for the vehicle based on its driving status and road environment data using a speed planning algorithm; that is, the second speed planning strategy for the connected autonomous vehicle corresponding to the first single-vehicle queue division strategy. The second speed planning strategy may include speed planning strategies for acceleration, deceleration, or constant speed driving.

[0073] As an example, in a dedicated lane, within the sequence [1,1, 0,2,2,2,...], the first element is 1, indicating that the first vehicle (the first car) in the dedicated lane is in the first platoon; the second element is 1, indicating that the second vehicle in the dedicated lane is in the same platoon as the first vehicle; the third element is 0, indicating that the second vehicle in the dedicated lane is a non-connected autonomous vehicle; and the fourth to sixth elements are 2, indicating that the fourth to sixth vehicles in the dedicated lane are in the second platoon. At a certain moment, if a non-connected autonomous vehicle enters between the first and second elements and does not leave within a preset time, to ensure driving safety, the connected vehicles in the dedicated lane need to be re-platooned, and a speed control command needs to be issued. In the above situation, after re-platooning, the sequence number [1,0,2, 0,3,3,3,...] is updated; at this time, the first and third elements are both autonomous vehicles traveling in single-vehicle platoons.

[0074] Furthermore, in the formation process, the formation partitioning algorithm can be triggered by timed triggering, combined with / or by vehicles entering / exiting the dedicated lane. Triggering conditions can include a connected autonomous vehicle entering or leaving the dedicated formation lane, and / or no formation partitioning has been performed at the current time, or t_cluster seconds (t_cluster is set to 5) have elapsed since the last formation partitioning. Speed ​​planning is called periodically, and because it is a control command, the call frequency is very high.

[0075] like Figure 2 As shown, Figure 2This is a schematic diagram of the process architecture for a multi-vehicle platooning method for connected autonomous vehicles on highways, provided in the embodiments of this specification. It should be noted that the execution logic of this process framework is performed once every t_speedPlan seconds, where t_speedPlan is set to 0.2.

[0076] Step 210: Search for vehicle information, location information, and speed information of all vehicles in the dedicated platooning lane to determine whether they are connected autonomous vehicles. Specifically, the cloud control platform determines the vehicle information of vehicles in the dedicated platooning lane based on the acquired road environment data and vehicle driving data, such as vehicle model and license plate information; further determines the vehicle's location and speed information; and can also determine whether the vehicles in the dedicated lane are connected autonomous vehicles.

[0077] Step 220: Determine if there are no connected autonomous vehicles in the dedicated lane. If yes, proceed to the end step; otherwise, proceed to step 230.

[0078] Step 230: Determine if any connected autonomous vehicles have entered or left the platooning lane, and / or if platooning has not been performed at the current time, and / or if t_cluster seconds have elapsed since the last platooning, where t_cluster is set to 5. If yes, proceed to step 240; otherwise, proceed to step 260.

[0079] Step 240: Invoke the platooning algorithm; if a platooning error occurs, the process ends, and fault code 1 is issued to all connected autonomous vehicles in the dedicated platooning lane. Fault code 1 indicates that a vehicle has experienced a platooning error, and this fault code serves as a notification to other connected autonomous vehicles. A platooning error indicates that platooning cannot be performed. For example, at a certain moment, a non-connected autonomous vehicle enters between vehicles A and B that are about to be platooned and fails to exit within a preset time; or vehicle B needs to change lanes among the three vehicles A, B, and C in the platoon; or the distance between vehicles A and B increases due to speed, causing the distance to no longer meet the preset conditions for platooning; or the driver of vehicle A or B voluntarily exits and no longer receives autonomous driving planning instructions. In these cases, it is impossible to platoon the connected vehicles in the dedicated lane at that moment, and a platooning error occurs.

[0080] Step 250: Update the queue number for each connected autonomous vehicle.

[0081] Step 260: Invoke the speed planning algorithm; if a speed planning error occurs, the process ends, and fault code 2 is issued to all connected autonomous vehicles in the dedicated platoon lane. Fault code 2 indicates that a vehicle has a speed planning error, and the fault code serves as a notification to other connected autonomous vehicles. Specifically, a speed planning error indicates that speed planning cannot be performed. For example, after vehicles A and B have been platooned and their platoon numbers updated, and the speed planning algorithm is invoked, a non-connected autonomous vehicle suddenly enters the planned space between vehicles A and B, causing the connected autonomous vehicles behind it to brake and decelerate, resulting in speed planning failure; or, if vehicle B in the platoon of A, B, and C needs to change lanes to exit the highway, it needs to coordinate with the traffic conditions in its target lane, which will also lead to a speed planning error in the platoon; or, the drivers of vehicles A and B may voluntarily disengage and no longer receive autonomous driving planning instructions.

[0082] Step 270: Send the target speed to each connected autonomous vehicle that is performing speed planning;

[0083] Completing steps 210 to 270 above completes one calculation of the multi-vehicle platooning algorithm on the cloud control platform.

[0084] In steps 210 to 270, specifically, the cloud control platform obtains the driving data of the connected autonomous vehicles on the highway, as well as the road environment data collected by the roadside perception devices. In step 210, the cloud control platform searches for information on all vehicles in the dedicated lane of the highway based on the obtained data, including vehicle location information, speed information, vehicle model, etc., to determine whether the vehicle currently traveling in the dedicated lane is a connected autonomous vehicle, and makes corresponding planning strategies based on the search results.

[0085] Further, in step 220: if it is determined that there is no connected autonomous vehicle in the dedicated lane for high-speed connected autonomous vehicles, the planning algorithm is not executed, and the task ends after one determination. If it is determined that there is a connected autonomous vehicle in the dedicated lane, the planning algorithm is pre-activated.

[0086] Further, step 230 is executed: determining whether the connected autonomous vehicles currently traveling in the dedicated lane meet the preset multi-vehicle platooning driving conditions. These conditions include: a connected autonomous vehicle entering or leaving the dedicated platooning lane, and / or no platooning division has been performed at the current time, and / or t_cluster seconds have elapsed since the last platooning division, where t_cluster is set to 5.

[0087] Furthermore, if it is determined that the preset multi-vehicle platooning conditions are not met, then step 260 is executed to call the speed planning algorithm to formulate a single-vehicle driving speed plan for the connected autonomous vehicles; step 270 is executed to send the target vehicle speed to each corresponding connected autonomous vehicle and control its speed.

[0088] Furthermore, if the preset multi-vehicle queuing conditions are met, then step 240 is executed. If no errors occur in the queuing and speed planning algorithms, steps 250 to 270 are executed sequentially.

[0089] Based on this, by using platooning algorithms, connected autonomous vehicles are actively grouped into platoons. By controlling the vehicle speed, the distance between vehicles in the platoon is kept small, which avoids frequent acceleration and deceleration, reduces platoon wind resistance, thereby improving fuel efficiency and road throughput, enhancing the response speed of following vehicles, and ensuring the safety and stability of the platoon.

[0090] Optionally, the method in the embodiments of this specification may further include setting a second preset distance.

[0091] Based on the second vehicle information of the second network of the dedicated lane, identify the abnormal vehicle located ahead of the convoy in the dedicated lane in the convoy's direction of travel;

[0092] If the distance between the abnormal vehicle and the lead vehicle in the platoon is less than or equal to a second preset distance, a takeover command is sent to each vehicle in the platoon.

[0093] In practical applications, the cloud control platform can acquire road information collected by roadside sensing devices on dedicated roads, including abnormal vehicles and obstacles. When the distance between an abnormal vehicle in the convoy's driving direction and the lead vehicle in the convoy is less than or equal to a second preset distance, a takeover command is sent to each vehicle in the convoy. If there is a driver, the driver takes over; if it is an unmanned vehicle, it operates using the autonomous cruise function, and the vehicles in the convoy sequentially leave the dedicated lane to avoid the abnormal vehicle. Under the condition of ensuring safe driving, a second preset distance, such as 300m, can be set as needed.

[0094] In practical applications, the cloud control platform can proactively break up platoons, and vehicles within a platoon can also choose whether to leave the platoon. Autonomous vehicles can calculate the distance between themselves and the vehicle in front using a distance feedback control algorithm. When this distance does not match the distance within the platoon, the cloud control platform can send a target speed to the vehicles, enabling them to maintain a distance appropriate for the platoon, thereby reducing air resistance and energy consumption.

[0095] Optionally, the methods in the embodiments of this specification may further include:

[0096] When the dedicated lane includes a road section with a gradient greater than a first gradient threshold, obtain the vehicle information of the lead vehicle in the same convoy;

[0097] Obtain the road slope information of the aforementioned road section;

[0098] The target speed is determined based on the information of the leading vehicle and the road slope information of the road segment;

[0099] The target speed is sent to the lead vehicle.

[0100] Specifically, a 2% gradient can be used as the starting point for the road segment. When the lead vehicle travels in the direction of travel to a road segment with a gradient greater than 2%, the target speed of the lead vehicle can be determined using an energy-saving cruise algorithm based on vehicle information and road gradient information.

[0101] In practical applications, the target speed of autonomous vehicles within the service area can be centrally planned. Speed ​​planning can also be performed on following vehicles in a platoon so that they can maintain the same speed as the lead vehicle. Following vehicles refer to all vehicles in the platoon except the lead vehicle. Speed ​​planning can be performed on the lead vehicle. When the road geometry is unknown, an adaptive cruise algorithm can be used to determine the target speed of the lead vehicle. When the road geometry is known, an energy-saving cruise algorithm can be used to determine the target speed of the lead vehicle. The lead vehicle travels at the aforementioned target speed, and the following vehicles in the platoon maintain the same driving status as the lead vehicle and follow it.

[0102] Optionally, the road segment is divided into four consecutive stages according to the vehicle's direction of travel: an uphill segment, a hill crest segment, a first downhill segment, and a second downhill segment. The first boundary between the uphill segment and the hill crest segment is the position where the leading vehicle's speed reaches a first preset speed. The second boundary between the hill crest segment and the first downhill segment is the position where the segment's gradient equals the first gradient threshold. The third boundary between the first downhill segment and the second downhill segment is the position where the leading vehicle's speed reaches the lane speed limit or the position where the segment's gradient equals the second gradient threshold.

[0103] Determining the target speed based on the information of the leading vehicle and the road slope information of the road segment may specifically include:

[0104] Based on the vehicle's location information, when the lead vehicle is located on the uphill section, a first target speed is generated based on a preset deceleration, so that the lead vehicle travels on the uphill section at the first target speed;

[0105] When the speed of the lead vehicle reaches the first preset speed, the first preset speed is used as the second target speed so that the lead vehicle travels on the hill crest section at the second target speed;

[0106] When the lead vehicle reaches the second dividing point, it accelerates using the gravitational potential energy of the lead vehicle in order to control the lead vehicle to travel on the first downhill section.

[0107] When the lead vehicle reaches the third dividing point, the speed at the third dividing point is taken as the third target speed, so that the vehicle can drive the second downhill section at the third target speed.

[0108] It's important to note that the platooning algorithm controls the longitudinal behavior of the vehicles, while the expected lateral behavior of each connected autonomous vehicle is the same. For each connected autonomous vehicle, after entering the highway, it changes lanes to the left until it reaches the dedicated logistics lane on the left. Then, approaching the merging point, such as when it's 1.5 kilometers from the expected merging point, it begins to change lanes to the right and then merges. The benefit of platooning is high road throughput, but the smaller distance between vehicles can hinder lane changes. Therefore, when performing multi-vehicle platooning algorithm calculations on the cloud control platform, it also needs to be used in conjunction with a lane-changing algorithm.

[0109] As an example, such as Figure 3 As shown, Figure 3 This is a schematic flowchart illustrating a lane-changing method for a connected autonomous vehicle on a highway, as provided in the embodiments of this specification.

[0110] Step 310: The cloud control platform determines whether the connected autonomous vehicle needs to perform vehicle-coordinated lane changing based on the road environment data and the driving data of the connected autonomous vehicle.

[0111] Step 320: If so, then based on the road environment data and the vehicle driving data of the connected autonomous vehicle, a third control command for coordinated lane changing is generated, and the third control command is sent to the connected autonomous vehicle, so that the connected autonomous vehicle controls the vehicle to change lanes according to the third control command.

[0112] Specifically, the third control command for generating coordinated lane changing includes: generating a control command for coordinated lane changing to the left, or generating a control command for coordinated lane changing to the right.

[0113] Furthermore, the third control command for generating coordinated lane changing specifically includes: determining, based on the road environment data, whether there are other vehicles in the target lane of the first connected autonomous vehicle that would affect its lane changing; the target lane is the lane the first connected autonomous vehicle intends to travel in after executing the lane changing strategy. Here, "first connected autonomous vehicle" refers to a connected autonomous vehicle in a dedicated lane that needs to change lanes. The "first" in this step is only used to distinguish between connected autonomous vehicles that need to change lanes and those that do not, and is not used for other limitations.

[0114] If not, then determine whether a second connected autonomous vehicle exists within a preset distance range. Here, a second connected autonomous vehicle refers to a vehicle that coordinates with the first connected autonomous vehicle to change lanes.

[0115] If so, the first connected autonomous vehicle is paired with the second connected autonomous vehicle closest to it; based on the paired first and second connected autonomous vehicles, a cooperative lane-changing algorithm is invoked to determine a lane-changing strategy for the first connected autonomous vehicle; and a third control command for the cooperative lane-changing is generated according to the lane-changing strategy for the first connected autonomous vehicle.

[0116] The cooperative lane-changing algorithm is activated when a connected autonomous vehicle wants to change lanes but is unable to do so due to obstruction from other vehicles. The algorithm then searches for the closest connected autonomous vehicle in its target lane, pairs the two vehicles together, and repeatedly executes the cooperative lane-changing strategy.

[0117] As an example, such as Figure 4 As shown, Figure 4 This is a schematic diagram of the process architecture of a lane-changing method for a highway connected autonomous vehicle provided in the embodiments of this specification.

[0118] Step 410: Determine whether the target lane of the connected autonomous vehicle is the same as its current lane; if the target lane is the same as its current lane, it means that the connected autonomous vehicle has changed lanes to the target lane it wants to drive in, and at this time the target lane is the same as its current lane.

[0119] As an example, it can be included that if the target lanes of two connected autonomous vehicles that are cooperating in lane changing are the same as their current lanes, then both connected autonomous vehicles are considered to have completed the lane change; if the target lane of any one of them is different from its current lane, then that vehicle is considered not to have completed the lane change and the cooperative lane changing strategy needs to continue.

[0120] Step 420: Determine whether no collaborative decision has been made yet and / or whether t_mcts seconds have elapsed since the last collaborative decision, where t_mcts is set to 1. If yes, proceed to step 430; otherwise, proceed to step 440.

[0121] Step 430: Invoke the collaborative decision-making algorithm.

[0122] Step 440: Determine whether the latest collaborative decision instruction is not "unsolvable"; if yes, it means there is no solution; no solution means that the collaborative lane-changing instruction cannot be executed. For example, the connected autonomous vehicle may be unable to change lanes according to the collaborative decision instruction, and changing lanes will cause traffic safety problems; if not, proceed to step 450.

[0123] Step 450: For each connected autonomous vehicle making collaborative decisions; if the latest collaborative decision instruction is acceleration, its target speed is the minimum of the road speed limit and "current speed + acceleration meters per second squared * t_speedPlan"; if the latest collaborative decision instruction is deceleration, its target speed is the minimum of the road speed limit and "current speed + decel meters per second squared * t_speedPlan"; if the latest collaborative decision instruction is to maintain a constant speed or change lanes to the left / right, its target speed is "current speed"; if the latest collaborative decision instruction is to change lanes to the right, the lane change instruction is to change lanes to the right; if the latest collaborative decision instruction is to change lanes to the left, the lane change instruction is to change lanes to the left.

[0124] Step 460: Send lane change / speed command to the connected autonomous vehicle.

[0125] In steps 410 to 460, step 410 is executed to determine whether the two connected autonomous vehicles cooperating in lane changing have already reached their target lanes. If the current driving lanes of the two connected autonomous vehicles cooperating in lane changing are their target lanes, the coordinated lane changing process ends. If the current driving lanes of the two connected autonomous vehicles cooperating in lane changing are not their target lanes, a coordinated decision is made; steps 420 and 430 are executed; and the latest coordinated lane changing decision instruction is obtained through the coordinated decision.

[0126] Furthermore, based on the latest decision instructions, the target speed and lane-changing instructions for each vehicle are determined, and steps 440 and 450 are executed to obtain lane-changing / speed instructions; step 460 is executed to send the lane-changing / speed instructions to the connected autonomous vehicle, and the connected autonomous vehicle executes the lane change.

[0127] Specifically, based on the lane-changing strategy for the first connected autonomous vehicle, a third control command for the coordinated lane-changing is generated, which specifically includes: determining whether the paired first connected autonomous vehicle and the second connected autonomous vehicle have changed lanes to their respective planned target lanes.

[0128] If not, determine whether the two paired connected autonomous vehicles meet the preset lane-changing conditions. The preset lane-changing conditions include that the paired connected autonomous vehicles have not yet implemented a cooperative strategy, or that the current time has exceeded a preset time since the last cooperative lane-changing strategy was implemented. If yes, call the cooperative lane-changing algorithm and output at least one of the following cooperative lane-changing decision instructions for any connected autonomous vehicle cooperating in lane-changing. Determine the third control instruction for cooperative lane-changing for each connected autonomous vehicle cooperating in lane-changing based on the cooperative lane-changing decision instructions.

[0129] Optionally, if the cooperative lane-changing decision instruction is acceleration, then its target speed is the minimum value between the road speed limit and the expected target speed; the expected target speed represents the current speed of the connected autonomous vehicle plus the expected acceleration multiplied by 0.2 seconds.

[0130] Optionally, if the cooperative lane-changing decision instruction is deceleration, then its target vehicle speed is the current speed of the connected autonomous vehicle and the expected deceleration rate multiplied by 0.2 seconds.

[0131] Optionally, if the cooperative lane-changing decision instruction is to maintain a constant speed or change lanes to the left / right, then the target vehicle speed is the current speed of the connected autonomous vehicle.

[0132] Optionally, if the coordinated lane-changing decision instruction is to change lanes to the right, the lane-changing instruction is to change lanes to the right.

[0133] Optionally, if the coordinated lane-changing decision instruction is to change lanes to the left, the lane-changing instruction is to change lanes to the left.

[0134] The collaborative decision-making process is initiated once per second, continuing until the connected autonomous vehicles complete the collaboration and reach their designated lanes. The collaborative decision output consists of one of five commands for each of the two connected autonomous vehicles. Based on these commands, the target speed and lane-changing behavior for each frame (0.2 seconds per frame) are calculated. The target speed (expected speed in the next frame, 0.2 seconds later) for the acceleration command is the current speed plus the expected acceleration multiplied by the two-frame interval. The target speed (expected speed in the next frame, 0.2 seconds later) for the deceleration command is the current speed plus the expected deceleration multiplied by the two-frame interval. For constant speed, left lane change, and right lane change, the target speed (expected speed in the next frame, 0.2 seconds later) is the current speed.

[0135] The cloud control platform completes a collaborative lane-changing calculation based on steps 410 to 460. If it is determined that the vehicle that needs to change lanes has not completed the lane change, steps 410 to 460 need to be executed repeatedly to realize the lane-changing behavior of the connected autonomous vehicle that needs to change lanes. This makes lane changing more convenient, resolves conflicts in lane changing, and improves traffic efficiency.

[0136] Optionally, during the calculation and decision-making process of the cloud control platform based on the road environment data and the vehicle driving data of the connected autonomous vehicles, a preset simulation method can be used to simulate the execution process of any control command or strategy to obtain feedback data. Based on the feedback data, the control commands can be further optimized to obtain the optimal control commands, thereby improving the safety of the connected autonomous vehicles when driving according to the control commands issued by the cloud control platform. The preset simulation method can be any existing method capable of simulating the execution of control commands issued from the cloud. In this embodiment, Markov and Monte Carlo search trees are used to simulate the decision-making process to achieve the resolution and optimal coordination of behavioral conflicts among multiple objects, while ensuring the safe and efficient driving of each connected autonomous vehicle on its expected path.

[0137] Optional, such as Figure 5 As shown, Figure 5 This diagram illustrates the cumulative reward in a cooperative lane-changing method for a connected autonomous vehicle on a highway, as provided in the embodiments of this specification. Here, r represents the reward, s represents the vehicle state, and a represents the policy or control command.

[0138] Markov decision processes simulate the stochastic policies and rewards of an agent in an environment. In the current state, following a certain policy, a specific action is performed and a reward is received due to the influence of the environment. The reward is the feedback from the environment after the action is performed, which is used to represent the environmental influence on the specific action. Furthermore, in a Markov decision process, a set of states and a set of rewards can be obtained. The expected reward r that can be obtained by following the policy from state s can be obtained.

[0139] Furthermore, the cumulative reward G in the simulation process is the accumulation of the reward over time steps, as shown below, where the discount factor γ represents the difference in importance between future rewards and present rewards.

[0140]

[0141] The goal of the simulation is to find an optimal strategy that maximizes the cumulative reward. Here, T represents time, r represents the reward, s represents the vehicle state, and a represents the strategy.

[0142] Furthermore, such as Figure 6 As shown, Figure 6This is a schematic diagram of the action space of an agent in multi-vehicle cooperative lane changing for a connected autonomous vehicle on a highway, provided by an embodiment of this specification. It should be noted that adding Markov properties to a stochastic process yields a Markov process; adding a reward further results in a Markov Reward Process (MRP); and adding an external stimulus, such as an agent's action, yields a Markov Decision Process.

[0143] In this diagram, the states of the multiple agents in multi-vehicle cooperative lane changing can be centrally represented, namely the speed and lane of the connected autonomous vehicle. The intelligent actions of the multiple agents (connected autonomous vehicles) can include the following: changing lanes to the left, changing lanes to the right, maintaining the current state (doing nothing), accelerating, and decelerating.

[0144] Specifically, such as Figure 7 As shown, Figure 7 This is a schematic diagram illustrating the steps of Monte Carlo search tree in multi-vehicle cooperative lane changing for connected autonomous vehicles on highways, as provided in the embodiments of this specification.

[0145] After multi-vehicle cooperative lane changing is expressed as a Markov decision process, Monte Carlo search tree is used to solve this optimization problem. As shown in the figure below, Monte Carlo search tree can be roughly divided into the following four steps: selection, expansion, simulation, and backpropagation.

[0146] Specifically, in the selection phase, we need to start from the root node, which is the state R where the decision needs to be made, and select the node N that most urgently needs to be expanded. State R is the first node checked in each iteration. The goal is to find a node at the bottom of the search tree after repeated iterations, so that we can continue with the subsequent steps.

[0147] Furthermore, after the selection phase ends, we find the node N that most urgently needs to be expanded, and the next action A that has not yet been expanded. We create a new node Nn in the search tree as a new child node of N. The state of Nn is the state of node N after it has performed action A.

[0148] Furthermore, during the simulation phase, in order to obtain an initial score for Nn, control instructions can be randomly executed starting from Nn until an execution ends. This outcome will be used as the initial score.

[0149] Furthermore, during the backpropagation phase, after the simulation of Nn ends, its parent node N and all nodes on the path from the root node to N will add their own cumulative scores based on the results of this simulation. If a simulation outcome is found directly during the selection phase, the scores will be updated based on that outcome.

[0150] Each iteration expands the search tree, and the size of the search tree increases with the number of iterations. The process terminates after a certain number of iterations or after a set time, selecting the best child node under the root node as the result of this decision.

[0151] Optionally, stochastic simulation of multi-agent actions can be used. For each action combination a that can be executed in the current state s, a stochastic policy can be simulated within a fixed time to obtain the cumulative rewards G1 to Gn, such as... Figure 8 As shown, Figure 8 This is a schematic diagram illustrating the multi-agent action simulation of a multi-vehicle cooperative lane-changing process for a connected autonomous vehicle on a highway, as provided in the embodiments of this specification.

[0152] Furthermore, the average gain of each action combination can be calculated as the expected cumulative return q, and the action combination with the largest average gain can be selected as the current optimal action combination a*.

[0153]

[0154]

[0155] Where s represents the current state, a represents the action combination, and G represents the reward after simulating the policy.

[0156] Furthermore, in the simulation of Monte Carlo tree decision-making, the actions of each agent are randomly selected to obtain the next virtual state, such as... Figure 9 As shown, Figure 9 This is a schematic diagram illustrating the random simulation of multi-vehicle actions in a control method for a highway connected autonomous driving vehicle provided in the embodiments of this specification.

[0157] Based on the objective function of the multi-agent global optimum problem obtained by the collaborative objective determination module, the reward function in the Monte Carlo search tree is expressed. The reward function includes individual vehicle state rewards and multi-vehicle conflict penalties. Individual vehicle state rewards are positive if the vehicle approaches the target state, which includes the target lane and target speed. The reward function is negative if multiple vehicles experience route conflicts.

[0158] exist Figure 8 In this process, the average gain of each action combination is calculated as the expected cumulative return q. The intelligent actions of the multi-agent in different lanes can include the following: lane change left, lane change right, do nothing, accelerate, and decelerate. Furthermore, the optimal action combination a for the connected autonomous vehicle at different times in different lanes is obtained, thus obtaining a schematic diagram of the random simulation of multi-vehicle actions, providing data support for the cloud control platform to make optimal decisions.

[0159] The above collaborative decision-making algorithm enables the collaborative goal determination module to first define the objective function based on the optimal criteria, which is the basis for policy updates. Then, a multi-agent Markov decision process model is established, and this model is used to uniformly express the multi-vehicle driving process, achieving the optimal coordination method, which is the Monte Carlo search tree. This enables the resolution and optimal coordination of behavioral conflicts among multiple objects on highways, while ensuring the safe and efficient driving of each connected autonomous vehicle on its expected path.

[0160] As another example, such as Figure 10 As shown, Figure 10 This is a schematic diagram of a control method for a highway connected autonomous driving vehicle provided in the embodiments of this specification, applied to a multi-vehicle collaborative service system in a highway unmanned logistics scenario.

[0161] Currently, most unmanned logistics vehicles operate within industrial parks. On highway logistics routes, many logistics vehicles are still manually driven. Current highway unmanned logistics vehicle demos mainly only achieve multi-vehicle convoy operation, with the lead vehicle controlled by a driver and the following vehicles being unmanned.

[0162] Vehicle-road cooperative technology is most easily implemented in closed, structured roads, and highways are a typical example of closed, structured roads, thus facilitating the rapid commercialization of vehicle-road cooperative technology. However, although vehicle and road technologies are already linked, effective coordination in application remains lacking. Trucks are a crucial transportation vehicle, and their driving characteristics are more suitable for implementing vehicle-road cooperative technology; therefore, it is necessary to conduct research on vehicle-road cooperative technology and its applications in typical highway truck scenarios.

[0163] Based on this, this embodiment provides a method for multi-vehicle collaborative services in unmanned logistics scenarios on highways, achieving vehicle-road-cloud integration around trunk logistics to help unmanned logistics vehicles achieve efficient operation. The connected autonomous vehicles in the architecture possess basic autonomous driving functions and are connected via dedicated communication terminals, serving as the implementation vehicle for cloud-controlled platooning / cooperative lane changing. The cloud control platform, leveraging multi-vehicle collaborative algorithms, helps unmanned logistics achieve commercial operational capabilities. Highways have roadside sensors and other infrastructure that assist vehicles in achieving positioning and communication capabilities on highways.

[0164] This multi-vehicle collaborative service system includes vehicle and roadside components, highway roadside sensing devices, and a cloud control platform decision-making end. The vehicle and roadside components include connected and non-connected autonomous vehicles; connected autonomous vehicles further include connected and non-connected autonomous freight vehicles and connected and non-freight vehicles. Connected vehicles can collect surrounding road information through their installed sensors and transmit the collected road environment information and their own status information to the cloud control platform via a data gateway. Roadside sensing devices installed on highways can transmit road environment data collected back to the server wirelessly or via wired connections, or transmit it to the cloud control platform via a data gateway. Roadside sensing devices include cameras, LiDAR, and fusion sensing computing units. The cloud control platform can be configured with high-precision map information or can directly obtain map information and high-precision positioning information from third-party map platforms via a data gateway. After obtaining the above data, the cloud control platform uses a pre-set multi-vehicle collaborative algorithm to obtain speed or lane-changing control commands for the connected autonomous logistics vehicles, thus providing guidance services for them.

[0165] For logistics vehicles traveling on highways, this embodiment proposes to open a dedicated logistics lane on the highway logistics trunk line, and use a cloud control platform to guide connected and autonomous logistics vehicles into the dedicated lane; use the cloud control platform to provide platooning guidance services for connected and autonomous logistics vehicles in the dedicated lane; and use the cloud control platform to guide connected and autonomous logistics vehicles out of the dedicated lane.

[0166] The specific method for providing vehicle guidance services to connected logistics vehicles traveling on highways using a cloud control platform is the same as the method for providing vehicle guidance services to connected autonomous vehicles described above, and will not be repeated here.

[0167] By employing the above methods, drivers can be eliminated from connected autonomous logistics vehicles, reducing personnel requirements. Furthermore, control commands issued using platooning and cooperative lane-changing algorithms based on a cloud control platform can reduce fuel consumption and driver labor costs (the ratio of vehicle fuel consumption, driver labor costs, and highway tolls in the total vehicle lifecycle cost is approximately 1:1:1). Placing reduces energy consumption, and autonomous driving functions reduce the number of drivers on duty, lowering personnel costs. Improved response speed of following vehicles reduces daily wear and tear and accidents, significantly reducing vehicle operating costs. This improves the efficiency and safety of connected autonomous vehicles on highways while simultaneously reducing overall vehicle operating costs.

[0168] Based on the same idea, embodiments of this specification also provide apparatus corresponding to the above methods. Figure 11 This is a schematic diagram of a control device for a highway connected autonomous driving vehicle provided in the embodiments of this specification.

[0169] like Figure 11 As shown, the device may include:

[0170] The acquisition module 1110 is used by the cloud control platform to acquire vehicle driving data of connected autonomous vehicles and road environment data collected by roadside perception devices;

[0171] The judgment module 1120 is used to determine, based on the road environment data and the driving data of the connected autonomous vehicles, whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions.

[0172] The first generation module 1130 is used to generate a first control command for multi-vehicle platoon driving based on the road environment data and the vehicle driving data of the connected autonomous vehicles if the conditions are met; and to send the first control command to any two adjacent connected autonomous vehicles so that the two adjacent connected autonomous vehicles drive as the same platoon.

[0173] The second generation module 1140 is used to generate a second control command for single-vehicle driving based on the road environment data and the driving data of the connected autonomous vehicles if no; and to send the second control command to any two adjacent connected autonomous vehicles so that the two adjacent connected autonomous vehicles drive in single-vehicle driving mode.

[0174] like Figure 11The device shown acquires vehicle driving data from connected autonomous vehicles and road environment data collected by roadside sensing devices through an acquisition module. Based on the road environment data and the driving data of the connected autonomous vehicles, a judgment module determines whether any two adjacent connected autonomous vehicles meet preset multi-vehicle platoon driving conditions. If so, a first control command for multi-vehicle platoon driving is generated through a first generation module; if not, a second control command for single-vehicle driving is generated through a second generation module. The first or second control command is sent to the two adjacent connected autonomous vehicles, causing them to drive according to the corresponding driving mode. This improves the driving efficiency and safety of connected autonomous vehicles on highways while also reducing vehicle operating costs.

[0175] It should also be noted that the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0176] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, for... Figure 11 As the apparatus embodiment shown is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be referred to in the description of the method embodiment.

[0177] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A control method for a highway connected autonomous driving vehicle, characterized in that, include: The cloud control platform acquires vehicle driving data from connected autonomous vehicles and road environment data collected by roadside sensing devices; Based on the road environment data and the vehicle driving data of the connected autonomous vehicles, determine whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions. If so, then based on the road environment data and the vehicle driving data of the connected autonomous vehicles, a first control command for multi-vehicle platooning is generated, including: if, and a non-connected autonomous vehicle leaves the platooning lane, then determining whether the first driving distance between the two connected autonomous vehicles adjacent to the non-connected autonomous vehicle leaving the platooning lane is less than or equal to a preset driving distance for vehicle platooning; wherein, the first connected autonomous vehicle among the two connected autonomous vehicles adjacent to the non-connected autonomous vehicle leaving the platooning lane is in the first platoon, and the third connected autonomous vehicle is in the second platoon; if the distance is less than or equal to the preset driving distance for vehicle platooning, then The first and second formations are grouped into the same queue, and the queue number corresponding to each connected autonomous vehicle is updated to the same queue number. Based on the road environment data and the vehicle driving data of the connected autonomous vehicles, a first multi-vehicle queue division strategy is determined using a formation division algorithm. Based on the road environment data and the vehicle driving data of the connected autonomous vehicles, and in conjunction with the first multi-vehicle queue division strategy, a first speed planning strategy is determined using a speed planning algorithm. The first control command is sent to any connected autonomous vehicle with the same queue number, so that any connected autonomous vehicle with the same queue number travels as the same queue. If not, then based on the road environment data and the vehicle driving data of the connected autonomous vehicles, a second control command for single-vehicle driving is generated, including: if not, and a non-connected autonomous vehicle enters the dedicated platooning lane and does not leave within a preset time, then the connected autonomous vehicles in the dedicated lane are re-platooned, and a first single-vehicle queue division strategy is determined using a platooning division algorithm based on the road environment data and the vehicle driving data of the connected autonomous vehicles; based on the road environment data and the vehicle driving data of the connected autonomous vehicles, combined with the first single-vehicle queue division strategy, a second speed planning strategy is determined using a speed planning algorithm; after re-platooning, the queue number corresponding to each connected autonomous vehicle is updated, and the second control command is sent to the two connected autonomous vehicles adjacent to the non-connected autonomous vehicles that entered the dedicated platooning lane, so that the two connected autonomous vehicles adjacent to the non-connected autonomous vehicles that entered the dedicated platooning lane drive in single-vehicle driving mode.

2. The method as described in claim 1, characterized in that, The cloud control platform includes a multi-vehicle platooning algorithm, which includes a platooning partitioning algorithm and a speed planning algorithm. The formation partitioning algorithm is used to determine the multi-vehicle queue partitioning strategy for the connected autonomous vehicle based on the road environment data and the vehicle driving data of the connected autonomous vehicle. The speed planning algorithm is used to determine the speed control strategy for the connected autonomous vehicle based on the road environment data and the vehicle driving data of the connected autonomous vehicle. Based on the multi-vehicle queue division strategy and speed control strategy, the first control command and the second control command are generated.

3. The method as described in claim 2, characterized in that, Before determining whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions, the following steps are included: Based on the road environment data, determine whether there are connected autonomous vehicles entering or leaving the platoon, or whether the platoon has not been divided at the current moment, or whether a preset time has passed since the last platoon division. The determination of whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions specifically includes: if so, determining whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions.

4. The method as described in claim 3, characterized in that, The method further includes: If not, the speed planning algorithm is directly invoked to determine the speed control strategy of the connected autonomous vehicle.

5. The method as described in claim 1, characterized in that, Also includes: The cloud control platform determines whether the connected autonomous vehicle needs to perform coordinated lane changing based on the road environment data and the vehicle driving data of the connected autonomous vehicle. If so, a third control command for coordinated lane changing is generated based on the road environment data and the vehicle driving data of the connected autonomous vehicle, and the third control command is sent to the connected autonomous vehicle, so that the connected autonomous vehicle controls the vehicle to change lanes according to the third control command.

6. The method as described in claim 5, characterized in that, The third control command for generating coordinated lane changing includes: generating a control command for coordinated lane changing to the left, or generating a control command for coordinated lane changing to the right; The third control command for generating coordinated lane changing specifically includes: Based on the road environment data, it is determined whether there are other vehicles in the target lane of the first connected autonomous vehicle that would affect the first connected autonomous vehicle's lane change; the target lane is the lane that the first connected autonomous vehicle will travel in after executing the lane change strategy; If not, determine whether a second connected autonomous vehicle exists within the preset distance range; If a second connected autonomous vehicle exists, the first connected autonomous vehicle is paired with the second connected autonomous vehicle closest to it. Based on the paired first and second connected autonomous vehicles, a cooperative lane-changing algorithm is invoked to determine a lane-changing strategy for the first connected autonomous vehicle. According to the lane-changing strategy for the first connected autonomous vehicle, a third control command for the cooperative lane-changing is generated.

7. The method according to any one of claims 1 to 6, characterized in that, The connected autonomous driving vehicle includes a connected autonomous driving logistics vehicle; The acquisition of vehicle driving data of the connected autonomous vehicle includes acquiring vehicle driving data when the connected autonomous logistics vehicle is driving on a dedicated logistics lane set up on a high-speed logistics trunk line. The acquisition of road environment data collected by roadside sensing devices includes acquiring road environment data on dedicated logistics lanes set up on high-speed logistics trunk lines.

8. A control device for a highway connected autonomous driving vehicle, characterized in that, include: The acquisition module is used by the cloud control platform to acquire vehicle driving data of the connected autonomous vehicle and road environment data collected by roadside perception devices. The judgment module is used to determine whether any two adjacent connected autonomous vehicles meet the preset multi-vehicle platoon driving conditions based on the road environment data and the vehicle driving data of the connected autonomous vehicles. The first generation module is configured to, if so, generate a first control command for multi-vehicle platooning based on the road environment data and the vehicle driving data of the connected autonomous vehicles, and further configured to: if, and a non-connected autonomous vehicle leaves the platooning lane, determine whether a first driving distance between two connected autonomous vehicles adjacent to the non-connected autonomous vehicle leaving the platooning lane is less than or equal to a preset driving distance for vehicle platooning; wherein, the first connected autonomous vehicle among the two connected autonomous vehicles adjacent to the non-connected autonomous vehicle leaving the platooning lane is in the first platoon, and the third connected autonomous vehicle is in the second platoon; if the distance is less than or equal to the preset driving distance for vehicle platooning... If the driving distance of the convoy is determined, the first and second convoys are grouped into the same queue, and the queue number corresponding to each connected autonomous vehicle is updated to the same queue number. Based on the road environment data and the vehicle driving data of the connected autonomous vehicles, a convoy partitioning algorithm is used to determine a first multi-vehicle queue partitioning strategy. Based on the road environment data and the vehicle driving data of the connected autonomous vehicles, and in combination with the first multi-vehicle queue partitioning strategy, a speed planning algorithm is used to determine a first speed planning strategy. The first control command is sent to any two adjacent connected autonomous vehicles with the same queue number, so that any connected autonomous vehicles with the same queue number drive as the same queue. The second generation module is used to generate a second control command for single-vehicle driving based on the road environment data and the vehicle driving data of the connected autonomous vehicles if no, and is further used to: if no, and a non-connected autonomous vehicle enters the dedicated platooning lane and does not leave within a preset time, then re-platoon the connected autonomous vehicles in the dedicated lane, and determine a first single-vehicle queue division strategy using a platooning division algorithm based on the road environment data and the vehicle driving data of the connected autonomous vehicles; determine a second speed planning strategy using a speed planning algorithm based on the road environment data and the vehicle driving data of the connected autonomous vehicles, combined with the first single-vehicle queue division strategy; update the queue number corresponding to each connected autonomous vehicle after re-platooning, and send the second control command to the two connected autonomous vehicles adjacent to the non-connected autonomous vehicles that entered the dedicated platooning lane, so that the two connected autonomous vehicles adjacent to the non-connected autonomous vehicles that entered the dedicated platooning lane drive in single-vehicle driving mode.

Images