Vehicle control method and vehicle

Abstract

The application provides a vehicle control method and a vehicle, and belongs to the technical field of intelligent driving. The method comprises the following steps: acquiring navigation attribute information and trajectory attribute information corresponding to at least one candidate driving trajectory; determining a target driving trajectory from the at least one candidate driving trajectory according to the navigation attribute information and the trajectory attribute information; in the case where a lane corresponding to the target driving trajectory and a lane currently occupied by the vehicle do not have a topological relationship, determining a temporary transition trajectory from the at least one candidate driving trajectory; controlling the vehicle to drive according to the temporary transition trajectory, and repeatedly performing the operations of determining the target driving trajectory and determining the temporary transition trajectory until the vehicle reaches the lane corresponding to the target driving trajectory. The scheme can not only take into account the driving quality in a local environment and the direction correctness of global navigation, but also avoid path deviation and safety risks, and ensure that the vehicle can safely reach the target lane in a scene where the topological relationship is incomplete.

Description

Technical Field

[0001] This application relates to the field of intelligent driving technology, and more specifically, to a vehicle control method and a vehicle in the field of intelligent driving technology. Background Technology

[0002] End-to-end autonomous driving is an important development direction in the field of intelligent driving. Its core idea is to directly map sensor inputs (such as camera images and LiDAR point clouds) into driving decisions through neural network models, and output multiple candidate driving trajectories and their confidence scores. These scores are used to reflect indicators such as the safety and smoothness of the trajectory in the local environment.

[0003] Currently, for multiple candidate driving trajectories output by neural network models, existing solutions typically adopt an optimization strategy based on confidence scores, that is, directly selecting the candidate driving trajectory with the highest confidence score as the final driving path and controlling the vehicle to drive according to the final driving path.

[0004] The existing solutions have at least the following problems: First, the confidence score primarily reflects the driving quality in a local environment. When the navigation requirement (such as turning left at the intersection ahead) is inconsistent with the driving direction of the locally optimal trajectory (such as going straight), the system is prone to selecting a locally safe but directional trajectory, causing the vehicle to deviate from the navigation plan. Second, this optimization strategy also relies on an implicit premise: that there is a passable topological connection between the lane corresponding to the selected trajectory and the current lane. Once the lane topological connection is broken or incomplete, the vehicle will either abandon the lane change and miss the navigation steering opportunity, or try to change lanes again. Figure 1 Making aggressive lane changes by crossing multiple lanes at once poses a serious safety hazard. Summary of the Invention

[0005] This application provides a vehicle control method and a vehicle. This method can not only take into account the driving quality in the local environment and the directional accuracy of global navigation, avoiding path deviation problems, but also avoid safety risks, ensuring that the vehicle can safely reach the target lane even in scenarios with incomplete topological relationships.

[0006] In a first aspect, this application provides a vehicle control method, including: Obtain navigation attribute information and trajectory attribute information corresponding to at least one candidate driving trajectory, wherein the navigation attribute information is used to indicate the ability of the lane to support the vehicle to continue driving along the global navigation path, and the trajectory attribute information includes behavioral attributes corresponding to each candidate driving trajectory, wherein the behavioral attributes represent the lateral movement trend of the vehicle when driving along the corresponding candidate driving trajectory; Based on the navigation attribute information and the trajectory attribute information, a target driving trajectory is determined from the at least one candidate driving trajectory; If it is determined that the lane corresponding to the target driving trajectory has no topological relationship with the lane where the vehicle is currently located, a temporary transition trajectory is determined from the at least one candidate driving trajectory based on the behavioral attributes of the target driving trajectory. The vehicle is controlled to travel along the temporary transition trajectory, and the operations of determining the target travel trajectory and determining the temporary transition trajectory are repeated until the vehicle reaches the lane corresponding to the target travel trajectory.

[0007] In the above technical solution, by using both navigation attribute information and trajectory attribute information as the basis for determining the target driving trajectory, the selection of the target driving trajectory takes into account both the driving quality in the local environment and the directional accuracy of global navigation, avoiding the path deviation problem caused by the confidence score only reflecting local safety and ignoring navigation needs. When there is no topological relationship between the lane corresponding to the target driving trajectory and the current lane, a temporary transition trajectory is determined based on the behavioral attributes of the target driving trajectory. This allows the vehicle to obtain an executable lane change path without relying on a complete topological path in the scenario of lane topology breaks. By repeatedly executing the determination operations of the target driving trajectory and the temporary transition trajectory, the vehicle gradually approaches the target lane by changing only one lane at a time. This avoids the safety risks caused by abandoning lane changes due to the inability to construct a complete path due to topological breaks, or by aggressively changing lanes across multiple lanes at once. This ensures that the vehicle can safely and compliantly reach the target lane even in scenarios with incomplete topological relationships.

[0008] In conjunction with the first aspect, in some possible implementations, the navigation attribute information includes the passable distance corresponding to at least one lane, where the passable distance represents the distance the vehicle can continue to travel in the current lane along the direction of travel. The trajectory attribute information also includes the confidence score corresponding to each candidate driving trajectory. The step of determining the target driving trajectory from the at least one candidate driving trajectory based on the navigation attribute information and the trajectory attribute information includes: Based on the passable distance, a navigation adaptation score is determined for each candidate driving trajectory. The navigation adaptation score is used to characterize the degree of matching between the candidate driving trajectory and the global navigation path. The target driving trajectory is determined from the at least one candidate driving trajectory based on the navigation adaptation score and the confidence score corresponding to each candidate driving trajectory.

[0009] In combination with the first aspect and the above implementation methods, in some possible implementation methods, determining the target driving trajectory from the at least one candidate driving trajectory based on the navigation adaptation score and the confidence score corresponding to each candidate driving trajectory includes: Based on the navigation adaptation score and the confidence score corresponding to each candidate driving trajectory, the comprehensive score corresponding to each candidate driving trajectory is determined; The candidate driving trajectory with the highest comprehensive score is determined as the target driving trajectory.

[0010] In the above technical solution, by using the passable distance as the basis for calculating the navigation adaptation score, and combining the navigation adaptation score with the confidence score output by the trajectory model in the candidate driving trajectory selection, the selection of the target driving trajectory takes into account both the driving quality in the local environment and the directional correctness of the global navigation path. After the two are combined, the candidate driving trajectory with the highest comprehensive score is selected as the target, thereby avoiding the path deviation problem caused by the confidence score only reflecting local safety at the source of decision-making. At the same time, the calculation of the navigation adaptation score only depends on the navigation attribute information of the passable distance, without the need to introduce complex path planning algorithms, and the implementation method is simple and efficient.

[0011] In combination with the first aspect and the above implementation methods, in some possible implementation methods, determining a temporary transition trajectory from the at least one candidate driving trajectory based on the behavioral attributes of the target driving trajectory includes: The lateral direction of the target is determined based on the lateral movement trend represented by the behavioral attributes of the target's driving trajectory. The candidate driving trajectory with the highest confidence score among the at least one candidate driving trajectory is determined as the temporary transition trajectory, provided that the behavioral attributes are consistent with those of the target driving trajectory, the corresponding lane is located on one side of the target lateral direction of the lane where the vehicle is currently located, and the candidate driving trajectory has the highest confidence score.

[0012] In the aforementioned technical solution, when the topological relationship is absent, the lateral direction of the target is determined based on the behavioral attributes of the target driving trajectory. Then, by combining the dual conditions of "consistent behavioral attributes" and "adjacent lanes," a temporary transitional trajectory is selected from the candidate trajectories. This allows the vehicle to obtain an executable driving path even in scenarios where the lane topology is broken or missing, completely eliminating the dependence on complete high-precision map topological relationships. This significantly expands the applicability of end-to-end autonomous driving systems, especially suitable for complex scenarios with incomplete topologies, such as areas without maps, road construction detours, and temporary failures of the perception system. By strictly limiting the spatial range of each lane change to adjacent lanes, the vehicle gradually approaches the target direction by moving only one lane laterally at a time. This effectively avoids aggressive lane-changing behavior that involves crossing multiple lanes at once due to directly selecting the target lane trajectory. The complex multi-lane lane-changing problem is decomposed into multiple independent single-lane lane-changing decisions. Each single-lane lane change only needs to judge the traffic conditions of adjacent lanes, significantly reducing the decision-making dimensionality and significantly improving safety. By further selecting the trajectory with the highest confidence score from the candidate set that meets the spatial and behavioral conditions, each temporary transition trajectory is the optimal choice in the local environment, ensuring the safety and smoothness of each lane change.

[0013] In combination with the first aspect and the above implementation methods, in some possible implementation methods, after determining the target driving trajectory from the at least one candidate driving trajectory, the method further includes: If the lateral movement trend characterized by the behavioral attributes of the target driving trajectory does not match the expected lateral movement trend indicated by the navigation requirements, and the passable distance is less than a set distance threshold, a lane change mode is entered. The lane change mode is used to determine a temporary transition trajectory from the at least one candidate driving trajectory.

[0014] In the above technical solution, the lane-changing mode is triggered only when the target driving trajectory behavior attributes do not match the expected lateral movement trend indicated by the navigation requirements and the current lane's passable distance is less than a set threshold. This ensures that the initiation of lane-changing behavior is constrained by two conditions: "whether the direction is correct" and "whether the remaining distance is sufficient." The mismatch in direction ensures that lane changing will not be done blindly when the direction is correct; the insufficient remaining distance ensures that lane changing will not be initiated too early when there is sufficient distance, nor will it be initiated when the distance is exhausted, thus avoiding missing the navigation turning opportunity. This combination of dual conditions ensures that the vehicle enters lane-changing mode only when "a lane change is indeed needed" and "a lane change must be done now," thereby efficiently completing the task of approaching the target lane.

[0015] In combination with the first aspect and the above implementation methods, in some possible implementation methods, determining the navigation adaptation score corresponding to each candidate driving trajectory based on the passable distance includes: Obtain lane attribute information corresponding to each of the at least one lane, wherein the lane attribute information includes lane type and / or lane usage rules; The target lane corresponding to each candidate driving trajectory is determined, and the availability correction coefficient corresponding to each target lane is determined based on the degree of matching between the lane attribute information and the vehicle type attribute of the vehicle. Based on the passable distance and availability correction coefficient corresponding to each target lane, the navigation adaptation score corresponding to each candidate driving trajectory is determined.

[0016] In the above technical solution, lane attribute information is introduced into the process of determining the navigation adaptation score. A usability correction coefficient is determined based on the matching degree between the lane attribute information and the vehicle type attribute. The passable distance is then corrected using this correction coefficient to determine the final navigation adaptation score. This means that the calculation of the navigation adaptation score no longer relies solely on the physical indicator of passable distance, but adds compliance as an independent evaluation dimension. As a result, when evaluating the navigation adaptation degree of each candidate trajectory, the system can simultaneously consider both the "physical passability" and "actual passability" of the lane. This effectively avoids the situation where a long passable distance is mistakenly judged as a highly adapted trajectory, significantly improving the accuracy of the navigation adaptation score in representing actual road usage rules.

[0017] In combination with the first aspect and the above implementation methods, in some possible implementation methods, repeating the operations of determining the target driving trajectory and determining the temporary transition trajectory until the vehicle reaches the lane corresponding to the target driving trajectory includes: The system repeatedly performs the operations of obtaining navigation attribute information and trajectory attribute information, determining the target driving trajectory, and determining the temporary transition trajectory at set time intervals, and in each cycle, the target driving trajectory and the temporary transition trajectory are re-determined based on the updated current position of the vehicle.

[0018] In the above technical solution, by cyclically executing the aforementioned steps at set time intervals and re-determining the target driving trajectory and temporary transition trajectory based on the updated vehicle's current position in each cycle, the system eliminates the need to pre-plan the complete lane-changing path before the lane change begins. Instead, it decomposes the lane-changing process into multiple independent single-step decision cycles, with each cycle deciding the next course of action based solely on the latest current position. If an unexpected situation occurs during the lane change (such as an obstacle vehicle suddenly appearing in an adjacent lane), the system can reselect a temporary transition trajectory in the next cycle based on updated perception information, avoiding the risk of collision caused by following the preset path. Simultaneously, the cyclical mechanism with set time intervals ensures the continuity and fault tolerance of the decision-making process; even if a suitable temporary transition trajectory cannot be selected in a certain cycle due to perception anomalies, the vehicle can still retry in the next cycle.

[0019] In combination with the first aspect and the above implementation methods, in some possible implementations, after the controlled vehicle travels according to the temporary transition trajectory, the method further includes: Obtain the traffic conditions information of the lane where the vehicle is currently located; If the traffic condition information indicates that the vehicle cannot continue to change lanes to the lane corresponding to the target driving trajectory, the vehicle is controlled to drive away from the lane corresponding to the target driving trajectory in order to return to the lane where the vehicle was before executing this temporary transition trajectory.

[0020] In the aforementioned technical solution, after controlling the vehicle to travel along a temporary transition trajectory, the system obtains information on the current lane's traffic conditions. When this information indicates that the vehicle cannot continue changing lanes in the target direction, the system controls the vehicle to retreat to the previous lane. This allows the vehicle to proactively exit the current lane-changing process and return to a known safe lane if it encounters sudden obstacles or insufficient safety clearance during a lane change, effectively preventing the vehicle from getting stuck in a dilemma halfway through the change. Furthermore, after retreating, the system can replan the subsequent path based on the new location, allowing the vehicle to continue approaching the target lane while prioritizing safety, thus balancing safety and target accessibility during the lane-changing process.

[0021] In combination with the first aspect and the above implementation methods, in some possible implementations, the method further includes: If it is determined that there is a topological relationship between the lane corresponding to the target driving trajectory and the lane where the vehicle is currently located, the sequence of intermediate lanes from the lane where the vehicle is currently located to the lane corresponding to the target driving trajectory is determined according to the topological relationship. The vehicle is controlled to travel sequentially along the middle lane sequence towards the lane corresponding to the target driving trajectory.

[0022] In the above technical solution, by determining the intermediate lane sequence based on the topological relationship when it exists and controlling the vehicle to change lanes sequentially, the vehicle can gradually reach the target lane by changing only one lane at a time. This avoids the aggressive lane-changing behavior of crossing multiple lanes at once caused by directly selecting the trajectory corresponding to the target lane. The complex one-time multi-lane lane-changing task is decomposed into multiple independent single-lane lane-changing actions. The decision space and safety judgment dimension of each single-lane lane-changing action are simpler and clearer, which greatly reduces the decision complexity and collision risk in the lane-changing process and improves the smoothness, compliance and passenger comfort of the lane-changing operation.

[0023] Secondly, this application provides a vehicle control device, the device comprising: The acquisition module is used to acquire navigation attribute information and trajectory attribute information corresponding to at least one candidate driving trajectory. The navigation attribute information is used to indicate the ability of the lane to support the vehicle to continue driving along the global navigation path. The trajectory attribute information includes behavioral attributes corresponding to each candidate driving trajectory. The behavioral attributes represent the lateral movement trend of the vehicle when driving along the corresponding candidate driving trajectory. The determination module is configured to determine a target driving trajectory from the at least one candidate driving trajectory based on the navigation attribute information and the trajectory attribute information, and to determine a temporary transition trajectory from the at least one candidate driving trajectory based on the behavioral attributes of the target driving trajectory when it is determined that the lane corresponding to the target driving trajectory has no topological relationship with the lane where the vehicle is currently located. The control module is used to control the vehicle to travel according to the temporary transition trajectory, and to repeatedly execute the operations of determining the target driving trajectory and determining the temporary transition trajectory until the vehicle reaches the lane corresponding to the target driving trajectory.

[0024] Thirdly, this application provides a vehicle, including: Memory, used to store executable program code; A processor is configured to call and run the executable program code from the memory, causing the vehicle to perform the method described in the first aspect or any possible implementation thereof.

[0025] Fourthly, a computer program product is provided, comprising: computer program code, which, when run on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof.

[0026] Fifthly, a computer-readable storage medium is provided that stores computer program code, which, when executed on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof. Attached Figure Description

[0027] Figure 1 This is a flowchart of a vehicle control method provided in an embodiment of this application; Figure 2 This is a schematic diagram illustrating the application of a vehicle control method provided in an embodiment of this application; Figure 3 This is another application diagram of a vehicle control method provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of a vehicle control device provided in an embodiment of this application; Figure 5 This is a schematic diagram of the structure of a vehicle provided in an embodiment of this application. Detailed Implementation

[0028] The technical solutions in this application will be clearly and thoroughly described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. "And / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0029] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0030] Currently, for multiple candidate driving trajectories output by neural network models used in end-to-end autonomous driving, existing solutions typically employ a confidence score-based optimization strategy. This involves directly selecting the candidate trajectory with the highest confidence score as the final driving path and controlling the vehicle to follow it. This optimization strategy has at least the following problems: First, the confidence score primarily reflects the driving quality in a local environment. When navigation requirements (such as turning left at an upcoming intersection) are inconsistent with the driving direction of the locally optimal trajectory (such as going straight), the system is prone to selecting a locally safe but directional trajectory, causing the vehicle to deviate from the navigation plan. Second, this optimization strategy also relies on an implicit premise that there is a passable topological connection between the lane corresponding to the selected trajectory and the current lane. Once the lane topological connection is broken or incomplete, the vehicle either abandons the lane change and misses the navigation steering opportunity, or attempts to change lanes again. Figure 1 Aggressive lane changes involving crossing multiple lanes at once pose serious safety hazards. Therefore, this application provides a vehicle control method.

[0031] Figure 1 This is a flowchart illustrating a vehicle control method provided in an embodiment of this application. The execution entity of this method can be a domain controller, a microcontroller unit (MCU), a system-on-chip (SoC), etc., and is not limited thereto. Figure 1 As shown, the method includes the following steps: Step 101: Obtain navigation attribute information and trajectory attribute information corresponding to at least one candidate driving trajectory.

[0032] Among them, navigation attribute information is used to indicate the ability of the lane to support the vehicle to continue driving along the global navigation path, and trajectory attribute information includes the behavioral attributes corresponding to each candidate driving trajectory. The behavioral attributes represent the lateral movement trend of the vehicle when driving along the corresponding candidate driving trajectory.

[0033] Step 102: Determine the target driving trajectory from at least one candidate driving trajectory based on navigation attribute information and trajectory attribute information.

[0034] Step 103: If it is determined that there is no topological relationship between the lane corresponding to the target driving trajectory and the lane where the vehicle is currently located, determine a temporary transition trajectory from at least one candidate driving trajectory based on the behavioral attributes of the target driving trajectory.

[0035] Step 104: Control the vehicle to travel along the temporary transition trajectory, and repeat the operations of determining the target driving trajectory and determining the temporary transition trajectory until the vehicle reaches the lane corresponding to the target driving trajectory.

[0036] In practical applications, it is first necessary to obtain navigation attribute information and trajectory attribute information corresponding to at least one candidate driving trajectory. The navigation attribute information indicates the ability of each lane to support the vehicle's continued travel along the global navigation path. This navigation attribute information includes at least the lane markings and travel distance for each lane, i.e., the distance the vehicle can continue to travel in the current lane along the direction of travel. In specific implementation, driving commands provided by the vehicle navigation system can be used as input to a pre-trained navigation model to obtain the navigation attribute information for each lane output by the navigation model.

[0037] The trajectory attribute information can include: lane identification for at least one candidate driving trajectory, behavioral attributes (i.e., whether the vehicle needs to change lanes to the left, right, or continue straight when executing the trajectory), confidence scores for each candidate driving trajectory, and lane topology relationships (representing the left-right and front-back relationships between lanes). In practice, the vehicle's cameras, millimeter-wave radar, lidar, and other sensors continuously collect surrounding environmental data (such as camera images, lidar point clouds, etc.) and transmit them to a pre-trained trajectory generation model via in-vehicle Ethernet to obtain the trajectory attribute information output by the trajectory generation model. It should be noted that both the navigation model and the trajectory generation model mentioned above can be pre-trained end-to-end neural network models.

[0038] Then, based on navigation attribute information and trajectory attribute information, the target driving trajectory is determined from at least one candidate driving trajectory. Specifically: One implementation method is to divide each candidate driving trajectory into different levels based on navigation attribute information: First layer (optimal layer): The passable distance of the lane corresponding to the candidate driving trajectory is greater than the set threshold (such as 1000 meters), and the driving direction of the lane is completely consistent with the navigation target (such as the navigation requires a left turn, and the lane is a left turn lane).

[0039] Second layer (suboptimal layer): Passable distance is in the medium range (e.g., 500-1000 meters), or the driving direction is partially consistent with the navigation target (e.g., if the navigation requires a left turn, the lane is a combined straight and left turn lane).

[0040] The third layer (usable layer): the passable distance is in a lower range (e.g., 200-500 meters), but at least the direction does not conflict with the navigation target.

[0041] Unqualified layers: those with a passable distance below the minimum threshold (e.g., <200 meters) or completely opposite to the navigation target direction are excluded.

[0042] In practice, the candidate driving trajectory with the highest confidence score is selected as the target driving trajectory in the first layer. If the first layer is empty, the candidate driving trajectory with the highest confidence score is selected from the second layer as the target driving trajectory. If the second layer is still empty, the candidate driving trajectory with the highest confidence score is selected from the third layer as the target driving trajectory.

[0043] As another implementation method, the navigation adaptation score corresponding to each candidate driving trajectory can be determined based on the traversable distance. The navigation adaptation score is used to characterize the degree of matching between the candidate driving trajectory and the global navigation path. Based on the navigation adaptation score and confidence score corresponding to each candidate driving trajectory, the comprehensive score corresponding to each candidate driving trajectory is determined. The candidate driving trajectory with the highest comprehensive score is determined as the target driving trajectory.

[0044] The following explanation uses a system-on-a-chip (hereinafter referred to as the system) as an example to illustrate this solution: In practical applications, each candidate driving trajectory is marked with a lane identifier (i.e., the lane number of the trajectory). Based on the lane identifier of each candidate driving trajectory, the system retrieves the permissible distance of the corresponding lane from the navigation attribute information. The permissible distance indicates the distance a vehicle can continue traveling in that lane along the current direction; that is, if a vehicle selects this lane and continues traveling in the current direction, how far it can travel before encountering an intersection, lane termination, or being unable to continue along the navigation path. A longer permissible distance indicates that the lane better supports the vehicle's continuous travel along the global navigation path, and provides a more ample time window for the vehicle to make steering or lane-changing decisions. A shorter permissible distance indicates that the lane is about to end or deviate from the navigation direction, requiring the vehicle to make adjustments as quickly as possible.

[0045] Based on this, the system can calculate the navigation adaptation score for each candidate driving trajectory according to the passable distance. This navigation adaptation score is used to quantitatively characterize the degree of matching between the candidate driving trajectory and the global navigation path. Its calculation logic follows an intuitive rule: the longer the passable distance, the higher the navigation adaptation score. It can be understood that if a lane allows a vehicle to travel a long distance in the correct direction, it means that the lane is highly matched with the navigation path; conversely, if a lane only has a few hundred meters or even tens of meters of passable space, it means that the lane is about to end or turn, and if a vehicle chooses this lane, it will soon face being forced to change lanes or deviate from the navigation, indicating a significantly lower degree of matching with the global navigation path. The specific mapping relationship between passable distance and navigation adaptation score can be implemented in various ways, such as direct linear mapping (the longer the distance, the higher the score), piecewise mapping (different distance intervals correspond to different score levels), or nonlinear mapping (introducing exponential or logarithmic functions to make the score advantage of long-distance lanes more significant). These will not be listed here, but regardless of the mapping method used, the core relationship is monotonically increasing: the longer the passable distance, the higher the navigation adaptation score.

[0046] In one optional embodiment, determining the navigation adaptation score corresponding to each candidate driving trajectory based on the drivable distance includes the following steps: obtaining lane attribute information corresponding to at least one lane, the lane attribute information including lane type and / or lane usage rules; determining the target lane corresponding to each candidate driving trajectory (the target lane is a lane among at least one lane), and determining the availability correction coefficient corresponding to each target lane based on the matching degree between the lane attribute information and the vehicle type attribute; determining the navigation adaptation score corresponding to each candidate driving trajectory based on the drivable distance corresponding to each target lane and the availability correction coefficient. The availability correction coefficient is used to characterize the actual availability of the target lane for the current vehicle after considering the lane type and / or lane usage rules.

[0047] The lane type indicates the functional attributes of the lane, such as a regular lane, bus lane, emergency lane, reversible lane, or high-occupancy vehicle lane (HOV). Lane usage rules indicate the traffic restrictions for the lane, such as prohibited hours (e.g., bus lanes are closed to private vehicles during weekday morning rush hour from 7:00 to 9:00), passenger capacity requirements (e.g., HOV lanes require at least two or three passengers), and driving direction restrictions (e.g., reversible lanes guide in different directions at different times). The vehicle type attribute indicates the basic category information of the vehicle, such as private vehicle, bus, taxi, emergency vehicle, etc., as well as the number of passengers and whether it enjoys special traffic privileges.

[0048] In practical applications, after obtaining lane attribute information and vehicle type attributes, the system can first determine the target lane corresponding to each candidate driving trajectory. Then, it determines the availability correction coefficient for each target lane based on the degree of matching between the lane attribute information and the vehicle type attribute. The rules for determining the value of this correction coefficient are as follows: When the lane type and vehicle type match perfectly (e.g., a private vehicle is in a regular lane, or a bus is in a bus lane and passage is permitted at the current time), the correction coefficient is 1, indicating that the lane is fully available and the passable distance can be directly included in the navigation adaptation score calculation without any discount. When the lane type and vehicle type do not match perfectly (e.g., a private vehicle is in a bus lane and it is currently during a restricted period, or a regular vehicle is in an emergency lane), the correction coefficient is 0, indicating that the lane is completely unavailable, the corresponding navigation adaptation score is zero, and all candidate trajectories corresponding to this lane will not be selected as target driving trajectories. When the lane type and vehicle type partially match or the degree of matching is unclear, the correction coefficient can be an intermediate value between 0 and 1 (such as 0.3, 0.5, 0.7, etc.). The specific value is determined according to the degree of matching. The higher the degree of matching, the closer the correction coefficient is to 1; the lower the degree of matching, the closer the correction coefficient is to 0.

[0049] After determining the availability correction coefficient, the system obtains the target lane corresponding to each candidate driving trajectory and determines the navigation adaptation score of the candidate driving trajectory based on the passable distance of the target lane and the correction coefficient. As an optional implementation: for each candidate driving trajectory, the original navigation adaptation score can be calculated first based on the passable distance of the target lane (the longer the passable distance, the higher the original score). Then, the original navigation adaptation score is multiplied by the availability correction coefficient corresponding to the target lane to obtain the corrected navigation adaptation score for the candidate driving trajectory. For example, a candidate driving trajectory might have a target lane that is a bus lane with a passable distance of 3000 meters and an original navigation adaptation score of 95. However, since this lane is a bus lane and is currently closed to other vehicles, its availability correction coefficient is 0, resulting in a corrected navigation adaptation score of 0. Another candidate driving trajectory might have a target lane that is a regular lane with a passable distance of 1000 meters and an original navigation adaptation score of 60. As a regular lane, its availability correction coefficient is 1.0, resulting in a corrected navigation adaptation score of 60. Even though the bus lane has a longer passable distance after correction, its corrected navigation adaptation score is 0, far lower than the 60 points of the regular lane. Therefore, the candidate driving trajectory corresponding to the bus lane will not be selected as the target driving trajectory.

[0050] By incorporating lane attribute information (lane type and / or lane usage rules) into the determination of navigation adaptation scores, and determining an availability correction coefficient based on the matching degree between this lane attribute information and vehicle type attributes, and then using this correction coefficient to adjust the passable distance to determine the final navigation adaptation score, the calculation of navigation adaptation scores no longer relies solely on the physical indicator of passable distance. Instead, it adds compliance (lane type and lane usage rules) as an independent evaluation dimension. This allows the system to consider both the "physical passability" and "actual passability" aspects when evaluating the navigation adaptation degree of each candidate trajectory. This effectively avoids misjudging highly adapted trajectories due to long passable distances that are actually not allowed to pass (such as bus lanes during restricted hours or emergency lanes where private vehicles are prohibited), significantly improving the accuracy of the navigation adaptation score in representing actual road usage rules.

[0051] After obtaining the navigation adaptation score of each candidate driving trajectory, the system merges it with the original confidence score of the candidate driving trajectories to determine the target driving trajectory from multiple candidate trajectories. The confidence score reflects the driving quality of the trajectory in a local environment, such as whether the trajectory is smooth, whether it maintains a safe distance from surrounding obstacles, and whether it conforms to vehicle kinematic constraints. A high confidence score means that the trajectory is safe and comfortable in a local area, but it lacks the ability to determine the global navigation direction. The navigation adaptation score precisely compensates for this deficiency; it does not care about whether the candidate driving trajectory is smooth locally, but only whether "this road can be continued." After merging the two, for a candidate driving trajectory to be selected, it must simultaneously meet two conditions: good local driving quality (high confidence score) and correct global direction (high navigation adaptation score).

[0052] The method for integrating the confidence score and the navigation adaptation score can be flexibly chosen. For example, the two can be directly added together or weighted to obtain a comprehensive score, and the one with the highest comprehensive score can be used as the target driving trajectory. Alternatively, the confidence score can be used as the base score, and preference rewards or penalties can be applied according to the navigation adaptation score. Trajectories with high navigation adaptation scores receive bonus points, while trajectories with low navigation adaptation scores are penalized with deduction points or directly filtered out of the candidate set, and so on. No limitation is made here.

[0053] To illustrate, let's consider an example: Suppose a vehicle is traveling north on a main urban road, and the navigation system indicates "turn left at the intersection 300 meters ahead to enter the east-west main road." The trajectory generation model generates four candidate trajectories based on the current perception information, each corresponding to a different lane and behavioral attribute, as detailed below: Candidate driving trajectory 1: Corresponds to the middle straight lane, with a confidence score of 92 (the highest among all candidate driving trajectories). This candidate driving trajectory is smooth and safe, and the vehicle can comfortably travel straight along the current lane. However, this straight lane ends at the next intersection, with only 250 meters of passable distance remaining. Due to the very short passable distance, the navigation adaptation score is only 25.

[0054] Candidate driving trajectory 2: Corresponding to the leftmost left-turn lane, with a confidence score of 78. This lane allows vehicles to turn left at the upcoming intersection, and after turning left, they can enter a continuous east-west main road extending for 3000 meters, providing a travel distance of up to 3000 meters. Therefore, the navigation adaptation score for this candidate driving trajectory is as high as 95.

[0055] Candidate driving trajectory 3: The corresponding left-turn lane (also the leftmost lane as candidate driving trajectory 2, but with a different offset) has a confidence score of 74, slightly lower than trajectory 2. Its corresponding left-turn lane also has a passable distance of 3000 meters, and its navigation adaptation score is 95.

[0056] Candidate driving trajectory 4: The rightmost right turn lane has a confidence score of 85 points. However, the right turn ends and enters the auxiliary road 100 meters ahead, and the passable distance is only 100 meters. Moreover, the right turn direction is completely opposite to the left turn required by the navigation. The navigation adaptation score is only 10 points.

[0057] At this point, the confidence score and navigation adaptation score can be combined for decision-making, and the comprehensive score can be calculated using a linearly weighted fusion method (each weight accounts for 50%). Candidate driving trajectory 1 = 92 × 0.5 + 25 × 0.5 = 58.5 points; Candidate driving trajectory 2 = 78 × 0.5 + 95 × 0.5 = 86.5 points; Candidate driving trajectory 3 = 74 × 0.5 + 95 × 0.5 = 84.5 points; Candidate driving trajectory 4 = 85 × 0.5 + 10 × 0.5 = 47.5 points.

[0058] As can be seen, candidate driving trajectory 2 has the highest overall score (86.5 points) and is therefore selected as the target driving trajectory.

[0059] By using the drivable distance as the basis for calculating the navigation adaptation score, and combining this navigation adaptation score with the confidence score output by the trajectory model in the candidate driving trajectory selection, the selection of the target driving trajectory takes into account both the driving quality in the local environment and the directional correctness of the global navigation path. After the two are combined, the candidate driving trajectory with the highest comprehensive score is selected as the target, thus avoiding the path deviation problem caused by the confidence score only reflecting local safety at the source of decision-making. At the same time, the calculation of the navigation adaptation score only relies on the navigation attribute information of drivable distance, without the need to introduce complex path planning algorithms, making the implementation simple and efficient.

[0060] After determining the target driving trajectory, the system does not immediately perform a lane change operation. Instead, it first determines whether it is indeed necessary to enter the lane change mode. If the lateral movement trend represented by the behavioral attributes of the target driving trajectory does not match the expected lateral movement trend indicated by the navigation requirements, and the passable distance is less than the set distance threshold, the system enters the lane change mode. The lane change mode is used to determine a temporary transition trajectory from at least one candidate driving trajectory.

[0061] In practice, the decision to enter lane-changing mode can be based on the following two conditions: 1. Does the behavioral attributes of the target driving trajectory match the expected lateral movement trend indicated by the navigation requirements? If they match, it means that the currently selected target driving trajectory already meets the navigation requirements in terms of direction, and the vehicle can continue driving along the current path without actively changing lanes. If they do not match, it means that the vehicle's currently selected driving direction conflicts with the driving direction required by the navigation, and a lane change adjustment is required.

[0062] 2. Whether the passable distance in the lane the vehicle is currently in is less than the set distance threshold. The set distance threshold can be flexibly configured according to the actual scenario. For example, it can be set to 300 meters, or it can be dynamically adjusted according to the current vehicle speed. The higher the vehicle speed, the longer the distance required to complete the lane change, and the higher the threshold will be; the lower the vehicle speed, the lower the threshold will be.

[0063] If the lateral movement trend represented by the behavioral attributes of the target driving trajectory does not match the expected lateral movement trend indicated by the navigation requirements, and the passable distance is less than the set distance threshold, then the lane change mode is activated.

[0064] By triggering lane-change mode only when the target driving trajectory behavior attributes do not match the expected lateral movement trend indicated by the navigation requirements and the current lane's passable distance is less than a set threshold, the initiation of lane-change behavior is constrained by two conditions: "whether the direction is correct" and "whether the remaining distance is sufficient." The mismatch in direction ensures that lane-change will not be initiated blindly when the direction is correct; the insufficient remaining distance ensures that lane-change will not be initiated too early when there is sufficient distance, nor will it be initiated when the distance is exhausted, thus avoiding missing the navigation turning opportunity. This combination of dual conditions ensures that the vehicle only enters lane-change mode when "a lane change is indeed necessary" and "a lane change must be initiated now," thereby efficiently completing the task of approaching the target lane.

[0065] Next, it is determined whether there is a topological relationship between the lane corresponding to the target driving trajectory and the lane where the vehicle is currently located. If it is determined that there is a topological relationship between the lane corresponding to the target driving trajectory and the lane where the vehicle is currently located, the sequence of intermediate lanes that the vehicle passes through from the lane where the vehicle is currently located to the lane corresponding to the target driving trajectory is determined according to the topological relationship; the vehicle is then controlled to drive along the sequence of intermediate lanes to the lane corresponding to the target driving trajectory.

[0066] In practical applications, if a topological relationship exists, it indicates that there is a connected path from the current lane to the target lane, consisting of "preceding and succeeding" and "left and right adjacent" relationships. The system can determine a complete lane-changing path based on this topological relationship. Specifically, the system, starting from the current lane, searches step-by-step along the topological relationship based on the left-right adjacent and / or preceding and succeeding relationships between lanes in the topological relationship. It sequentially determines the sequence of intermediate lanes that the vehicle needs to enter to reach the target lane. After determining the intermediate lane sequence, the system selects the candidate driving trajectory corresponding to the next lane in the sequence from the candidate driving trajectories as the actual driving trajectory, controls the vehicle to complete one lane change, updates the vehicle position, and continues to perform the next lane change along the intermediate lane sequence until the vehicle reaches the lane corresponding to the target driving trajectory.

[0067] To facilitate understanding, the following will be combined with... Figure 2 For example: exist Figure 2There are 5 lanes in the system, numbered 1-1, 1-2, 1-3, 1-4, and 1-5. The vehicle is currently in lane 1-1. The confidence scores of the candidate trajectories for each lane are 800, 600, 400, 200, and 100, respectively. The remaining travel distance for lane 1-1 is 200 meters, and the remaining travel distance for lane 1-5 is 3000 meters. Based on the method provided in the above embodiment, the system determines that the target travel trajectory corresponds to lane 1-5. After checking the topological relationships, the system finds that there is a complete topological connection between lanes 1-1 and 1-5, i.e., 1-1 and 1-2 are left-right neighbors, 1-2 and 1-3 are left-right neighbors, 1-3 and 1-4 are left-right neighbors, and 1-4 and 1-5 are left-right neighbors. Based on this, the system determines the intermediate lane sequence as 1-2→1-3→1-4→1-5.

[0068] After determining the middle lane sequence, the system controls the vehicle to change lanes step by step according to this sequence: first, from 1-1 to 1-2; second, from 1-2 to 1-3; third, from 1-3 to 1-4; and fourth, from 1-4 to 1-5, finally reaching the target lane. Throughout the process, after each lane change, the system reconfirms the current position and the next target lane before controlling the vehicle to perform the next lane change, until the target lane 1-5 is reached.

[0069] By determining the intermediate lane sequence based on the topological relationship when it exists and controlling the vehicle to change lanes sequentially, the vehicle can gradually reach the target lane by changing only one lane at a time. This avoids the aggressive lane-changing behavior that would result from directly selecting the trajectory corresponding to the target lane and crossing multiple lanes at once. The complex task of changing multiple lanes at once is decomposed into multiple independent single-lane lane-changing actions. The decision space and safety judgment dimension of each single-lane lane-changing action are simpler and clearer, which greatly reduces the decision complexity and collision risk during the lane-changing process and improves the smoothness, compliance and passenger comfort of the lane-changing operation.

[0070] If it is determined that there is no topological relationship between the lane corresponding to the target driving trajectory and the lane where the vehicle is currently located, a temporary transition trajectory is determined from at least one candidate driving trajectory based on the behavioral attributes of the target driving trajectory. The specific process is as follows: the target lateral direction is determined based on the lateral movement trend represented by the behavioral attributes of the target driving trajectory; the candidate driving trajectory with the highest confidence score among the at least one candidate driving trajectory, whose behavioral attributes are consistent with the behavioral attributes of the target driving trajectory, whose corresponding lane is located on one side of the target lateral direction of the lane where the vehicle is currently located, is determined as the temporary transition trajectory.

[0071] In practical applications, when there is no topological relationship between the lane corresponding to the target driving trajectory and the lane the vehicle is currently in, the system cannot plan a complete lane-changing path from the current lane to the target lane by analyzing the topological relationship. In this case, the system activates a temporary transition trajectory determination mechanism based on behavioral attributes, using the "target direction" as a guide, to find a temporary path in the currently passable lane that can move towards the target direction.

[0072] In practice, the system first reads the behavioral attributes of the target driving trajectory, whose values ​​include at least left lane change, straight ahead, and right lane change. Since the behavioral attributes themselves represent the vehicle's intention to move laterally to the left or right relative to the current lane, the system can determine the target lateral direction that the vehicle needs to execute in order to approach the target lane. If the behavioral attribute of the target driving trajectory is left lane change, the target lateral direction is left; if it is right lane change, the target lateral direction is right.

[0073] After determining the target's lateral direction, the system filters through all candidate driving trajectories to obtain a temporary transition trajectory. The filtering criteria include the following dimensions: First, the behavioral attributes of the candidate driving trajectory must be consistent with those of the target driving trajectory. That is, if the target driving trajectory requires a left lane change, then the candidate driving trajectory must also be a left lane change. This condition is significant in ensuring that the movement intention of the temporary transition trajectory is consistent with the final target, thus preventing the vehicle from executing an incorrect lateral movement direction.

[0074] Second, the lane corresponding to the candidate driving trajectory must be located on the side of the target lateral direction of the vehicle's current lane, and it must be an adjacent lane. This condition ensures that the vehicle only moves one lane laterally in the target direction each time, and will not make a one-time aggressive lane change that crosses multiple lanes.

[0075] Third, among the candidate driving trajectories that meet the above two conditions, the trajectory with the highest confidence score is selected as the temporary transition trajectory, because the confidence score reflects the safety and smoothness of the trajectory in the local environment, and selecting the trajectory with the highest score can ensure the safety of this lane change.

[0076] It's important to note that the confidence score's filtering scope is limited to candidate driving trajectories that already meet the spatial and behavioral conditions of "consistent behavioral attributes" and "adjacent lanes." In other words, its role is to "select the best from the qualified candidate set," rather than choosing the highest-scoring candidate from the entire candidate set. The reason both "consistent behavioral attributes" and "adjacent lanes" are needed to narrow down the candidate set is that they each have their own emphasis: consistent behavioral attributes ensure that the vehicle's movement intention matches the target direction, while adjacent lanes ensure that the vehicle does not cross multiple lanes at once. Both are indispensable. If only consistent behavioral attributes are present but the lanes are not adjacent, a trajectory separated by two lanes might be selected; if only adjacent lanes are present but inconsistent behavioral attributes, incorrect directions such as going straight or changing right might be selected. Only trajectories that simultaneously meet both conditions can achieve a safe one-step lane change while ensuring correct direction. The confidence score then selects the best among these, ensuring that the selected trajectory is safe and feasible in the local environment.

[0077] To facilitate understanding, the following will be combined with... Figure 3 For example: exist Figure 3 There are 10 lanes in total, with lane numbers 1-1, 1-2, 1-3, 1-4, 1-5, 2-1, 2-2, 2-3, 2-4, and 2-5. The vehicle is currently in lane 1-1. The confidence scores of the candidate driving trajectories for each lane are 800, 600, 400, 200, 100, 640, 480, 320, 160, and 80, respectively. The remaining travel distance for lane 2-1 is 200 meters, and the remaining travel distance for lane 2-5 is 3000 meters. Based on the method provided in the above embodiment, it can be determined that the lane corresponding to the target driving trajectory is 2-5. After checking the topology, the system found that there is a 20-meter blank area between lane 1-1 and lane 2-5, making it impossible to establish any preceding / following or left / right adjacent connections; that is, the topology does not exist.

[0078] Faced with this situation, the system first reads the behavioral attributes of the target driving trajectory and finds that the behavioral attribute of the target driving trajectory is "left lane change," thus determining that the target's lateral direction is to the left. Subsequently, the system filters all candidate driving trajectories: First, it filters candidate driving trajectories whose behavioral attributes are consistent with the target driving trajectory (i.e., "left lane change"); second, among the candidate driving trajectories with consistent behavioral attributes, it further filters candidate driving trajectories whose corresponding lanes are located in the adjacent position to the left of the current lane 1-1, i.e., the candidate driving trajectory corresponding to lane 1-2; third, among the candidate driving trajectories that meet the above two conditions, it selects the one with the highest confidence score. For example... Figure 3As shown, the candidate driving trajectory corresponding to lane 1-2 has a confidence score of 640, which is the highest score under this condition, and is therefore determined to be the temporary transition trajectory.

[0079] The system outputs the temporary transition trajectory and controls the vehicle to perform a single lane change from 1-1 to 1-2. Upon reaching 1-2, the system re-executes the above steps to determine the target driving trajectory (the target is still 2-5). Finding that the topology does not exist again, the system repeats the above process: the target behavior attribute remains left lane change, and the target lateral direction remains left. Among the left-adjacent lanes of the current lane 1-2, the candidate trajectory with the highest score and "left lane change" behavior attribute is selected. At this point, the trajectory corresponding to lane 1-3 may be selected as the next temporary transition trajectory. Through multiple iterations, the vehicle gradually approaches from 1-1 to 1-2, then to 1-3, 1-4, and 1-5, eventually reaching the target lane 2-5.

[0080] By determining the target's lateral direction based on the behavioral attributes of the target's driving trajectory when the topological relationship is absent, and then selecting temporary transitional trajectories from candidate trajectories based on the dual conditions of "consistent behavioral attributes" and "adjacent lanes," the vehicle can still obtain an executable driving path even in scenarios with broken or missing lane topologies. This completely eliminates the dependence on complete high-precision map topological relationships, significantly expanding the applicability of end-to-end autonomous driving systems, especially suitable for complex scenarios with incomplete topologies, such as areas without maps, road construction detours, and temporary failures of perception systems. By strictly limiting the spatial range of each lane change to adjacent lanes, the vehicle gradually approaches the target direction by moving only one lane at a time. This effectively avoids aggressive lane-changing behavior that crosses multiple lanes at once due to directly selecting the target lane trajectory. The complex multi-lane lane-changing problem is decomposed into multiple independent single-lane lane-changing decisions. Each single-lane lane change only needs to judge the traffic conditions of adjacent lanes, greatly reducing the decision-making dimensionality and significantly improving safety. By further selecting the trajectory with the highest confidence score from the candidate set that meets the spatial and behavioral conditions, each temporary transition trajectory is the optimal choice in the local environment, ensuring the safety and smoothness of each lane change.

[0081] After that, the vehicle can be controlled to drive according to the temporary transition trajectory, and the operations of determining the target driving trajectory and determining the temporary transition trajectory can be repeated until the vehicle reaches the lane corresponding to the target driving trajectory. Specifically, the operations of obtaining navigation attribute information and trajectory attribute information, determining the target driving trajectory and determining the temporary transition trajectory can be performed cyclically at set time intervals, and in each cycle, the target driving trajectory and the temporary transition trajectory are re-determined based on the updated current position of the vehicle.

[0082] It should be noted that during the lane change process, the vehicle's position is constantly changing, for example, moving from the starting point of lane 1-1 to the ending point of lane 1-2. To ensure the accuracy of the next decision, the system does not plan the trajectory for all subsequent steps before the lane change begins. Instead, after the lane change is completed, the system re-acquires the latest perception and navigation attribute information based on the vehicle's new position, and then re-determines the target driving trajectory and the temporary transition trajectory.

[0083] In practice, the system periodically triggers operations at preset time intervals (e.g., 100 milliseconds) to acquire navigation and trajectory attribute information, determine the target driving trajectory, and determine a temporary transition trajectory. At the beginning of each cycle, the system first acquires the vehicle's latest location information, then acquires the current perception information (surrounding lane lines, obstacles, etc.), navigation and trajectory attribute information based on this updated location, and then re-executes the operations to determine the target driving trajectory and the temporary transition trajectory based on this latest information. When the vehicle reaches the lane corresponding to the target driving trajectory, the cycle ends, and the system exits lane-changing mode.

[0084] By cyclically executing the aforementioned steps at set time intervals, and re-determining the target driving trajectory and temporary transition trajectory based on the updated vehicle position in each cycle, the system eliminates the need to pre-plan the complete lane-changing path before the lane change begins. Instead, it breaks down the lane-changing process into multiple independent single-step decision cycles, with each cycle deciding the next course of action based solely on the latest position. If an unexpected situation occurs during the lane change (such as an obstacle vehicle suddenly appearing in an adjacent lane), the system can reselect a temporary transition trajectory in the next cycle based on updated perception information, avoiding the risk of collision caused by following the preset path. Simultaneously, the cyclical mechanism with set time intervals ensures decision continuity and fault tolerance; even if a suitable temporary transition trajectory cannot be selected in a certain cycle due to perception anomalies, the vehicle can still retry in the next cycle.

[0085] Additionally, it should be noted that this embodiment introduces a backtracking mechanism in the aforementioned loop. That is, after executing a temporary transition trajectory, if the vehicle finds it cannot continue moving towards the target direction, it is controlled to retreat to the lane it was in the previous step, preventing the vehicle from getting stuck in the middle of a lane change and being unable to move forward or backward. Specifically, after controlling the vehicle to travel according to the temporary transition trajectory, the traffic condition information of the lane the vehicle is currently in is obtained. If the traffic condition information indicates that the vehicle cannot continue changing lanes to the lane corresponding to the target travel trajectory, the vehicle is controlled to travel in a direction away from the lane corresponding to the target travel trajectory, so as to return to the lane the vehicle was in before executing this temporary transition trajectory.

[0086] In practical applications, after a vehicle completes a lane change (e.g., from lane 1-1 to lane 1-2) by determining a temporary transition trajectory, the system prepares to execute the next cycle. This involves re-determining the target driving trajectory and a new temporary transition trajectory to continue approaching the target lane. However, before executing the next lane change, the system first obtains the traffic conditions information of the vehicle's current lane (i.e., 1-2). This traffic conditions information is used to assess whether it is safe and feasible to continue changing lanes from the current lane towards the target direction. Traffic conditions information may include: whether there are obstacles in adjacent lanes in the target direction, whether there is sufficient safety clearance for lane changes, and whether the remaining passable distance in the current lane is sufficient to support the completion of subsequent lane changes.

[0087] If the traffic information indicates that the vehicle can safely continue changing lanes towards the target lane, the system will proceed with the next cycle according to the normal procedure, determine a new temporary transition trajectory, and control the vehicle to continue moving forward. If the traffic information indicates that the vehicle cannot continue changing lanes towards the target direction (e.g., an approaching obstacle vehicle is detected in an adjacent lane at high speed, there is insufficient safety clearance, or the current lane is about to be exhausted but the target direction is still not open), the system will not force a lane change, nor will it wait in place. Instead, it will control the vehicle to move away from the lane corresponding to the target trajectory, i.e., it will perform a reverse lane change, returning the vehicle to the lane it was in before executing this temporary transition trajectory.

[0088] To illustrate, consider this example: Suppose a vehicle is performing a gradual lane change from lane 1-1 to the target lane 2-5. The vehicle has completed the first lane change, moving from 1-1 to 1-2. The system then enters the next cycle, preparing to change lanes from 1-2 to 1-3 (or 2-3). However, before executing the lane change, the system obtains traffic conditions information for lane 1-2 and detects a high-speed approaching obstacle vehicle in lane 1-3, with insufficient safety clearance for a safe lane change. Instead of waiting or abruptly changing lanes, the system controls the vehicle to change lanes left from 1-2 back to 1-1, returning to the lane it was in before executing the temporary transition trajectory. After the return, the system re-executes the target trajectory determination and temporary transition trajectory determination based on the new position, potentially planning a different approach path or attempting to approach the target lane again once conditions improve.

[0089] By acquiring current lane clearance information after controlling the vehicle to travel along a temporary transition trajectory, and controlling the vehicle to reverse back to the previous lane when this information indicates that it cannot continue changing lanes in the target direction, the system allows the vehicle to proactively exit the current lane change process and return to a known safe lane if it encounters sudden obstacles or insufficient safety clearance during a lane change. This effectively avoids the vehicle getting stuck in a dilemma halfway through a lane change. Furthermore, after reversing, the system can replan the subsequent path based on the new position, allowing the vehicle to continue approaching the target lane while prioritizing safety, thus balancing safety and target accessibility during the lane change process.

[0090] Based on the above, the vehicle control method provided in this application uses both navigation attribute information and trajectory attribute information as the basis for determining the target driving trajectory. This ensures that the selection of the target driving trajectory takes into account both the driving quality in the local environment and the directional accuracy of the global navigation, avoiding the path deviation problem caused by the confidence score only reflecting local safety and ignoring navigation requirements. When there is no topological relationship between the lane corresponding to the target driving trajectory and the current lane, a temporary transition trajectory is determined based on the behavioral attributes of the target driving trajectory. This allows the vehicle to obtain an executable lane-changing path without relying on a complete topological path in scenarios with lane topology breaks. By repeatedly executing the determination operations of the target driving trajectory and the temporary transition trajectory, the vehicle gradually approaches the target lane by changing only one lane at a time. This avoids the safety risks caused by abandoning lane changing due to the inability to construct a complete path due to topological breaks, or by aggressively changing lanes across multiple lanes at once. This ensures that the vehicle can safely and compliantly reach the target lane even in scenarios with incomplete topological relationships.

[0091] Figure 4 This is a schematic diagram of the structure of a vehicle control device provided in an embodiment of this application, as shown below. Figure 4 As shown, the device includes: an acquisition module 41, a determination module 42, and a control module 43.

[0092] The acquisition module 41 is used to acquire navigation attribute information and trajectory attribute information corresponding to at least one candidate driving trajectory. The navigation attribute information is used to indicate the ability of the lane to support the vehicle to continue driving along the global navigation path. The trajectory attribute information includes behavioral attributes corresponding to each candidate driving trajectory. The behavioral attributes represent the lateral movement trend of the vehicle when driving along the corresponding candidate driving trajectory.

[0093] The determining module 42 is configured to determine a target driving trajectory from the at least one candidate driving trajectory based on the navigation attribute information and the trajectory attribute information, and to determine a temporary transition trajectory from the at least one candidate driving trajectory based on the behavioral attributes of the target driving trajectory when it is determined that the lane corresponding to the target driving trajectory does not have a topological relationship with the lane where the vehicle is currently located.

[0094] The control module 43 is used to control the vehicle to travel according to the temporary transition trajectory, and to repeatedly execute the operations of determining the target driving trajectory and determining the temporary transition trajectory until the vehicle reaches the lane corresponding to the target driving trajectory.

[0095] In one possible implementation, the navigation attribute information includes the passable distance corresponding to at least one lane, where the passable distance represents the distance a vehicle can continue to travel in the current lane along the direction of travel. The trajectory attribute information also includes the confidence score corresponding to each candidate driving trajectory. The determining module 42 is specifically used to: determine the navigation adaptation score corresponding to each candidate driving trajectory based on the passable distance, where the navigation adaptation score is used to characterize the degree of matching between the candidate driving trajectory and the global navigation path; and determine the target driving trajectory from the at least one candidate driving trajectory based on the navigation adaptation score and the confidence score corresponding to each candidate driving trajectory.

[0096] In one possible implementation, the determining module 42 is further configured to: determine a comprehensive score corresponding to each candidate driving trajectory based on the navigation adaptation score and the confidence score corresponding to each candidate driving trajectory; and determine the candidate driving trajectory with the highest comprehensive score as the target driving trajectory. Additionally, it determines the target lateral direction based on the lateral movement trend represented by the behavioral attributes of the target driving trajectory; and determines the candidate driving trajectory with the highest confidence score among the at least one candidate driving trajectory whose behavioral attributes are consistent with those of the target driving trajectory, whose corresponding lane is located on one side of the target lateral direction of the lane currently occupied by the vehicle, as the temporary transition trajectory.

[0097] In one possible implementation, the device further includes: a judgment module, configured to enter a lane-changing mode when the lateral movement trend characterized by the behavioral attributes of the target driving trajectory does not match the expected lateral movement trend indicated by the navigation requirements, and the passable distance is less than a set distance threshold; the lane-changing mode is used to determine a temporary transition trajectory from the at least one candidate driving trajectory.

[0098] In one possible implementation, the determining module 42 is further configured to: obtain lane attribute information corresponding to at least one lane, the lane attribute information including lane type and / or lane usage rules; determine the target lane corresponding to each candidate driving trajectory, and determine the availability correction coefficient corresponding to each target lane based on the matching degree between the lane attribute information and the vehicle type attribute of the vehicle; and determine the navigation adaptation score corresponding to each candidate driving trajectory based on the passable distance and availability correction coefficient corresponding to each target lane.

[0099] In one possible implementation, the control module 43 is specifically used to: cyclically execute the operations of acquiring navigation attribute information and trajectory attribute information, determining the target driving trajectory, and determining the temporary transition trajectory at a set time interval, and in each cycle, redetermine the target driving trajectory and the temporary transition trajectory based on the updated current position of the vehicle.

[0100] In one possible implementation, the control module 43 is further configured to: acquire traffic condition information of the lane currently occupied by the vehicle; if the traffic condition information indicates that the vehicle cannot continue to change lanes to the lane corresponding to the target travel trajectory, control the vehicle to travel away from the lane corresponding to the target travel trajectory, so as to return to the lane the vehicle was in before executing this temporary transition trajectory. Additionally, if it is determined that there is a topological relationship between the lane corresponding to the target travel trajectory and the lane currently occupied by the vehicle, determine the sequence of intermediate lanes traversed from the lane currently occupied by the vehicle to the lane corresponding to the target travel trajectory based on the topological relationship; and control the vehicle to travel sequentially along the sequence of intermediate lanes to the lane corresponding to the target travel trajectory.

[0101] It should be noted that the vehicle control device provided in the above embodiments is only illustrated by the division of the above functional modules when executing the vehicle control method. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the vehicle control device and the vehicle control method embodiments provided in the above embodiments belong to the same concept. Therefore, for details not disclosed in the device embodiments of this application, please refer to the embodiments of the vehicle control method of this application, which will not be repeated here.

[0102] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0103] This application also provides a vehicle. Figure 5 This is a schematic diagram of the structure of a vehicle provided in an embodiment of this application.

[0104] Typically, vehicle 500 includes one or more processors 501 and one or more memories 502.

[0105] Processor 501 may include one or more processing cores, such as a quad-core processor, an octa-core processor, etc. Processor 501 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). Processor 501 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, processor 501 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, processor 501 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0106] Memory 502 may include one or more computer-readable storage media, which may be non-transitory. Memory 502 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, the non-transitory computer-readable storage media in memory 502 is used to store at least one computer program, which is executed by processor 501 to implement the power margin determination method provided in the method embodiments of this application.

[0107] Those skilled in the art will understand that Figure 5 The structure shown does not constitute a limitation on vehicle 500 and may include more or fewer components than shown, or combine certain components, or use different component arrangements.

[0108] In addition, the device provided in the embodiments of this application may specifically be a chip, component or module. The chip may include a connected processor and a memory. The memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute the vehicle control method provided in the above embodiments.

[0109] This embodiment also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the vehicle control method provided in the above embodiment.

[0110] This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to implement the vehicle control method provided in the above embodiment.

[0111] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0112] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0113] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0114] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A vehicle control method, characterized in that, include: Obtain navigation attribute information and trajectory attribute information corresponding to at least one candidate driving trajectory, wherein the navigation attribute information is used to indicate the ability of the lane to support the vehicle to continue driving along the global navigation path, and the trajectory attribute information includes behavioral attributes corresponding to each candidate driving trajectory, wherein the behavioral attributes represent the lateral movement trend of the vehicle when driving along the corresponding candidate driving trajectory; Based on the navigation attribute information and the trajectory attribute information, a target driving trajectory is determined from the at least one candidate driving trajectory; If it is determined that the lane corresponding to the target driving trajectory has no topological relationship with the lane where the vehicle is currently located, a temporary transition trajectory is determined from the at least one candidate driving trajectory based on the behavioral attributes of the target driving trajectory. The vehicle is controlled to travel along the temporary transition trajectory, and the operations of determining the target travel trajectory and determining the temporary transition trajectory are repeated until the vehicle reaches the lane corresponding to the target travel trajectory.

2. The method according to claim 1, characterized in that, The navigation attribute information includes the passable distance corresponding to at least one lane, where the passable distance represents the distance that the vehicle can continue to travel in the current lane along the direction of travel. The trajectory attribute information also includes the confidence score corresponding to each candidate driving trajectory. Determining the target driving trajectory from the at least one candidate driving trajectory based on the navigation attribute information and the trajectory attribute information includes: Based on the passable distance, a navigation adaptation score is determined for each candidate driving trajectory. The navigation adaptation score is used to characterize the degree of matching between the candidate driving trajectory and the global navigation path. The target driving trajectory is determined from the at least one candidate driving trajectory based on the navigation adaptation score and the confidence score corresponding to each candidate driving trajectory.

3. The method according to claim 2, characterized in that, The step of determining the target driving trajectory from the at least one candidate driving trajectory based on the navigation adaptation score and the confidence score corresponding to each candidate driving trajectory includes: Based on the navigation adaptation score and the confidence score corresponding to each candidate driving trajectory, the comprehensive score corresponding to each candidate driving trajectory is determined; The candidate driving trajectory with the highest comprehensive score is determined as the target driving trajectory.

4. The method according to claim 2, characterized in that, The step of determining a temporary transition trajectory from the at least one candidate driving trajectory based on the behavioral attributes of the target driving trajectory includes: The lateral direction of the target is determined based on the lateral movement trend represented by the behavioral attributes of the target's driving trajectory. The candidate driving trajectory with the highest confidence score among the at least one candidate driving trajectory is determined as the temporary transition trajectory, provided that the behavioral attributes are consistent with those of the target driving trajectory, the corresponding lane is located on one side of the target lateral direction of the lane where the vehicle is currently located, and the candidate driving trajectory has the highest confidence score.

5. The method according to claim 2, characterized in that, After determining the target driving trajectory from the at least one candidate driving trajectory, the method further includes: If the lateral movement trend characterized by the behavioral attributes of the target driving trajectory does not match the expected lateral movement trend indicated by the navigation requirements, and the passable distance is less than a set distance threshold, a lane change mode is entered. The lane change mode is used to determine a temporary transition trajectory from the at least one candidate driving trajectory.

6. The method according to claim 2, characterized in that, The step of determining the navigation adaptation score corresponding to each candidate driving trajectory based on the passable distance includes: Obtain lane attribute information corresponding to each of the at least one lane, wherein the lane attribute information includes lane type and / or lane usage rules; The target lane corresponding to each candidate driving trajectory is determined, and the availability correction coefficient corresponding to each target lane is determined based on the degree of matching between the lane attribute information and the vehicle type attribute of the vehicle. Based on the passable distance and availability correction coefficient corresponding to each target lane, the navigation adaptation score corresponding to each candidate driving trajectory is determined.

7. The method according to claim 1, characterized in that, The process of repeating the operations of determining the target driving trajectory and determining the temporary transition trajectory until the vehicle reaches the lane corresponding to the target driving trajectory includes: The system repeatedly performs the operations of obtaining navigation attribute information and trajectory attribute information, determining the target driving trajectory, and determining the temporary transition trajectory at set time intervals, and in each cycle, the target driving trajectory and the temporary transition trajectory are re-determined based on the updated current position of the vehicle.

8. The method according to claim 1, characterized in that, After the controlled vehicle travels according to the temporary transition trajectory, the method further includes: Obtain the traffic conditions information of the lane where the vehicle is currently located; If the traffic condition information indicates that the vehicle cannot continue to change lanes to the lane corresponding to the target driving trajectory, the vehicle is controlled to drive away from the lane corresponding to the target driving trajectory in order to return to the lane where the vehicle was before executing this temporary transition trajectory.

9. The method according to claim 1, characterized in that, Also includes: If it is determined that there is a topological relationship between the lane corresponding to the target driving trajectory and the lane where the vehicle is currently located, the sequence of intermediate lanes from the lane where the vehicle is currently located to the lane corresponding to the target driving trajectory is determined according to the topological relationship. The vehicle is controlled to travel sequentially along the middle lane sequence towards the lane corresponding to the target driving trajectory.

10. A vehicle, characterized in that, The vehicles include: Memory, used to store executable program code; A processor for calling and running the executable program code from the memory, causing the vehicle to perform the method as described in any one of claims 1 to 9.

Images