Vehicle lane change decision system, method, and vehicle

Abstract

The application discloses a vehicle lane-changing decision system, method and vehicle. The system comprises a data input layer, an algorithm configuration layer and a decision processing layer. The data input layer is responsible for acquiring lane-changing scene perception data related to lane-changing decision and transmitting the data to the decision processing layer. The algorithm configuration layer stores a plurality of candidate lane-changing decision algorithms adapted to different vehicle performance attributes. The decision processing layer first analyzes data quality attributes of the lane-changing scene perception data, judges vehicle performance attributes according to the data quality attributes, and then calls a target lane-changing decision algorithm matched with the vehicle performance attributes from the algorithm configuration layer. Finally, a lane-changing decision result is generated based on the target lane-changing decision algorithm according to the lane-changing scene perception data, which is used for vehicle lane-changing control. In this way, the system can flexibly adapt to different types of vehicles to better realize lane-changing decision based on the effective decoupling and execution of the corresponding functions of the aforementioned data input, algorithm configuration and decision processing functional layers.

Description

Technical Field

[0001] This application relates to the field of driver assistance technology, and in particular to a vehicle lane change decision system, method and vehicle. Background Technology

[0002] Current lane change decision-making systems for assisted driving vehicles typically suffer from rigid architectures and poor scalability. Systems designed for low-cost vehicles are usually based on fixed rules and simple perception-prediction models, resulting in limited performance. Even adding sensors to collect more lane change-related perception information fails to effectively improve decision-making quality. Systems designed for high-performance vehicles, on the other hand, are often deeply coupled with complex perception-prediction models, making them difficult to decouple and port to low-computing-power platforms. In other words, different types of vehicles require different lane change decision-making systems. This system fragmentation due to the separation of computing power and performance leads to several problems. First, it results in redundant investment of R&D resources in lane change decision-making technology, leading to wasted resources and impacting R&D costs and efficiency. Second, it hinders the unified evolution of lane change decision-making technology systems and may also affect the accuracy and efficiency of lane change decisions. Therefore, there is an urgent need for a vehicle lane change decision-making system and method that can flexibly adapt to different types of vehicles to better achieve lane change decisions. Summary of the Invention

[0003] The purpose of this application is to address how to implement a vehicle lane change decision system and method that can more flexibly adapt to different types of vehicles to better achieve lane change decisions.

[0004] To address the aforementioned technical problems, in a first aspect, this application discloses a vehicle lane change decision system applied in a vehicle. The system includes a data input layer, an algorithm configuration layer, and a decision processing layer. The data input layer determines lane change scenario perception data related to lane change decisions during vehicle operation and transmits this data to the decision processing layer. The algorithm configuration layer stores a set of lane change decision algorithms, including multiple candidate lane change decision algorithms applicable to different vehicle performance attributes. The decision processing layer determines the data quality attributes of the lane change scenario perception data, determines the vehicle performance attributes based on the data quality attributes, selects a target lane change decision algorithm corresponding to the vehicle performance attributes from the set of lane change decision algorithms stored in the algorithm configuration layer, and obtains a lane change decision result based on the target lane change decision algorithm and the lane change scenario perception data for use in vehicle lane change control.

[0005] By adopting the above system, through the layered architecture design of several functional layers, namely the data input layer, algorithm configuration layer, and decision processing layer, the various functions in the vehicle lane change decision system are effectively decoupled. Each functional layer is also a corresponding functional module, which realizes functional modularization. This solves the problems of rigidity and poor scalability of traditional system architecture, and makes the system have good modular flexibility and scalability.

[0006] The lane change decision is initiated by the data input layer, which collects perception data of the lane change scenario, providing a data foundation for flexible and scalable decision-making.

[0007] The algorithm configuration layer stores and manages the set of lane change decision algorithms, allowing these algorithms to be updated, optimized, or expanded independently of the entire system. When vehicle sensors are upgraded or computing power is increased, there is no need to reconstruct the entire system; only the corresponding candidate algorithms need to be added or upgraded to the algorithm set, thereby reducing R&D complexity and cost.

[0008] The decision processing layer analyzes the data quality attributes of the perceived data in lane-change scenarios, such as dimensionality and confidence level, to infer the vehicle's actual performance attributes, such as computing power and hardware capabilities. Based on this, it dynamically selects the most suitable target lane-change decision algorithm from the algorithm configuration layer. This allows low-cost vehicles to use lightweight, efficient algorithms that match limited perceived data, while high-performance vehicles can utilize more complex and precise algorithms. Consequently, system performance can smoothly scale with the quality of the input source and the level of computing power, rather than being limited by fixed rules or deeply bound to specific complex models.

[0009] Furthermore, the selected best-matching lane change decision algorithm is used to process real-time lane change scenario perception data, generating the final lane change decision result for vehicle lane change control. This seamlessly links the lane change decision result to the vehicle control system, completing a full closed loop from perception and computation to execution. This enables the deployment and operation of lane change decision schemes for different types of vehicles (i.e., vehicles with different performance attributes) within a unified technical framework. This allows the vehicle lane change system to flexibly adapt to different types of vehicles, improving the effectiveness of lane change decisions, saving R&D resources, reducing R&D costs, improving R&D efficiency, and promoting the unified development of the technology system.

[0010] Furthermore, for different types of vehicles, corresponding lane-change decision algorithms can be selected to obtain corresponding lane-change decision results. This means that a more suitable lane-change decision algorithm can be used to obtain the corresponding lane-change decision results more quickly and accurately, effectively ensuring or improving the accuracy and efficiency of lane-change decisions. Additionally, this layered architecture design can also effectively ensure or improve the accuracy and efficiency of lane-change decisions.

[0011] In summary, the vehicle lane change decision-making scheme provided by the embodiments of this application can flexibly adapt to different types of vehicles, improve the effectiveness of lane change decision-making, save R&D resources, reduce R&D costs, improve R&D efficiency, promote the unified development of the technology system, and effectively guarantee or improve the accuracy and efficiency of lane change decision-making, thus enabling better lane change decision-making.

[0012] Secondly, the embodiments of this application disclose a vehicle lane change decision method, applied to the aforementioned vehicle lane change decision system in a vehicle. The method includes: a data input layer determining lane change scenario perception data related to lane change decision during vehicle operation, and transmitting the lane change scenario perception data to a decision processing layer; the decision processing layer determining the data quality attributes of the lane change scenario perception data, determining the vehicle performance attributes based on the data quality attributes, selecting a target lane change decision algorithm corresponding to the vehicle performance attributes from a set of lane change decision algorithms stored in the algorithm configuration layer based on the vehicle performance attributes, and obtaining a lane change decision result based on the target lane change decision algorithm and the lane change scenario perception data for use in vehicle lane change control. The set of lane change decision algorithms includes multiple candidate lane change decision algorithms applicable to different vehicle performance attributes.

[0013] Using the above method, the functions corresponding to the layered architecture of the system are transformed into specific executable process steps. This method is initiated by the data input layer, which collects lane-change scenario perception data, providing a data foundation for flexible and scalable decision-making. The decision processing layer analyzes the data quality attributes of the lane-change scenario perception data to infer the actual performance attributes of the vehicle, and dynamically selects the most suitable target lane-change decision algorithm from the algorithm configuration layer accordingly. This allows low-cost vehicles to use lightweight, efficient algorithms that match limited perception data, while high-performance vehicles can call more complex and precise algorithms. This enables the system performance to scale smoothly with the quality of the input source and the level of computing power, rather than being limited by fixed rules or deeply bound to specific complex models. Finally, the selected most suitable target lane-change decision algorithm processes the real-time lane-change scenario perception data to generate the final lane-change decision result for vehicle lane-change control. This seamlessly links the lane-change decision result to the vehicle control system, completing a complete closed loop from perception, computation to execution. This enables the deployment and operation of different types of vehicles within a unified technical framework. This method enables the vehicle lane-changing system to flexibly adapt to different types of vehicles, improving the effectiveness of lane-changing decisions, saving R&D resources, reducing R&D costs, improving R&D efficiency, and promoting the unified development of the technology system. Furthermore, it effectively guarantees or improves the accuracy and efficiency of lane-changing decisions. Therefore, this method can flexibly adapt to different types of vehicles to better achieve lane-changing decisions.

[0014] In one possible implementation of the second aspect above, based on the target lane change decision algorithm, a lane change decision result is obtained according to the lane change scenario perception data, including: predicting whether the vehicle meets the overtaking trigger condition according to the lane change scenario perception data, wherein meeting the overtaking trigger condition means that the vehicle can overtake the target vehicle or can be overtaken by the target vehicle; if the overtaking trigger condition is met, predicting whether there is a first safe lane change space in the target lane where the target vehicle is located that meets the corresponding lane change requirements according to the lane change scenario perception data; if there is a first safe lane change space, determining the lane change decision result according to the lane change scenario perception data.

[0015] Using the above method, based on lane-change scenario perception data, it predicts whether a vehicle meets the overtaking trigger conditions. The trigger logic for lane-change decisions is clearly defined as overtaking demand. By first determining whether there is a vehicle to be overtaken and whether there is a vehicle that can be overtaken, meaningless calculations of lane-change space are avoided, improving the efficiency of the lane-change decision-making process. The overtaking trigger conditions are further subdivided into two scenarios: the vehicle actively overtaking and the vehicle being overtaken, covering a more comprehensive range of real-world driving interactions and enhancing adaptability to complex traffic scenarios. After determining that there is an overtaking demand, the immediate safety of the target lane is further assessed. Feasibility and safety are judged step by step, reducing the risk of collisions caused by misjudgments of space. After completing the demand and safety verification, a specific decision result is finally generated. This step-by-step, progressive lane-change decision-making process ensures the stability and reliability of lane-change decisions, thereby improving the accuracy and efficiency of lane-change decisions.

[0016] In one possible implementation of the second aspect above, based on the target lane change decision algorithm and the lane change scenario perception data, a lane change decision result is obtained, including: when a vehicle has a lane change intention, determining whether there is a second safe lane change space that meets the corresponding lane change requirements in the lane corresponding to the lane change intention; if there is a second safe lane change space, obtaining a lane change decision result that directly executes the lane change; if there is no second safe lane change space, obtaining a lane change decision result based on the target lane change decision algorithm and the lane change scenario perception data.

[0017] By employing the above method, when a vehicle intends to change lanes, it determines whether a second safe lane-changing space exists in the lane corresponding to the lane-changing intention, such as an active lane-changing intention triggered by the driver or navigation, thus making the application scenario more complete. For active lane changes with a second safe lane-changing space, a lane-changing decision result is obtained by directly executing the lane change, achieving rapid response in lane-changing decisions and improving the system's execution efficiency. When a direct lane change does not exist with a second safe lane-changing space, a lane-changing decision result is obtained based on the target lane-changing decision algorithm and the lane-changing scenario perception data, making the method more adaptable to different scenarios. This allows the vehicle lane-changing method to flexibly adapt to different types of vehicles, improving the accuracy and efficiency of lane-changing decisions, saving R&D resources, and promoting the unified development of the technology system.

[0018] In one possible implementation of the second aspect above, the lane change decision algorithm includes a vehicle safety verification algorithm, a vehicle control space prediction algorithm, and a vehicle control strategy algorithm.

[0019] Using the above method, the lane change decision algorithm is deconstructed into three functionally defined and clearly demarcated sub-algorithm modules. This modular definition makes the construction, updating, and maintenance of the algorithm set more systematic, and also facilitates the independent optimization or replacement of a sub-algorithm for different types of vehicles, further enhancing the system's configurability and scalability, thus enabling the vehicle lane change method to flexibly adapt to different types of vehicles.

[0020] In one possible implementation of the second aspect above, the vehicle safety verification algorithm includes an overtaking calculation equation to predict whether the vehicle meets the overtaking trigger condition based on lane change scenario perception data. This includes: solving the overtaking calculation equation based on the lane change scenario perception data to determine whether there is a feasible positive time solution for the corresponding overtaking relationship; if a positive time solution exists, the corresponding overtaking relationship is determined to be feasible, and the vehicle is determined to meet the overtaking trigger condition, and the smaller the value of the positive time solution, the higher the overtaking efficiency; if no positive time solution exists, the corresponding overtaking relationship is determined to be infeasible, and the vehicle is determined not to meet the overtaking trigger condition.

[0021] By employing the above method, the judgment of whether overtaking is permissible is concretized into a computable mathematical equation, transforming the decision-making process from rule-based empirical judgment to equation-based quantitative calculation, thus improving the objectivity and accuracy of the decision. By solving the equation to find a positive time-varying solution, the feasibility of overtaking is transformed into a mathematical optimization problem. This allows for comprehensive consideration of various kinematic parameters such as the speed, acceleration, and relative position of both the driver and the target vehicle, enabling accurate future state predictions and making the decision more scientific. The existence of a positive time-varying solution determines the feasibility of the corresponding overtaking relationship, not only determining whether overtaking is possible but also quantifying the overtaking efficiency through the magnitude of the solution. This provides crucial quantitative evidence for selecting the optimal strategy from multiple feasible options, contributing to more efficient driving behavior. The absence of a positive time-varying solution determines the infeasibility of the corresponding overtaking relationship, avoiding forced lane changes in scenarios with excessive risk or physical impossibility, ensuring driving safety. This improves the accuracy and efficiency of lane-change decisions.

[0022] In one possible implementation of the second aspect above, the vehicle control space prediction algorithm includes a lane change space calculation equation. Based on lane change scenario perception data, it predicts whether there is a first safe lane change space in the target lane where the target vehicle is located that meets the corresponding lane change requirements. This includes: solving the lane change space calculation equation based on the lane change scenario perception data; calculating the predicted distance between the target vehicle and the corresponding lane change reference vehicle within the prediction time, wherein the lane change reference vehicle is either a vehicle in front of the target vehicle or a vehicle behind the target vehicle; comparing the predicted distance with a preset lane change safety distance threshold; and if the predicted distance is greater than or equal to the lane change safety distance threshold, determining that there is a first safe lane change space in the target lane that meets the corresponding lane change requirements.

[0023] By employing the above method, the determination of whether a safe space exists is concretized into another calculable mathematical equation, ensuring the accuracy of the safety assessment. Solving the lane change space calculation equation is used to predict changes in traffic flow gaps in the target lane over a future period, dynamically evaluating lane change decisions rather than solely considering the current instantaneous distance, which better aligns with the actual need for space prediction in driving. By comparing the dynamic prediction results with static safety thresholds, a clear binary judgment of safe or unsafe is made. This method standardizes the complex safety assessment, allowing it to be implemented and adjusted across different lane change decision algorithms. This enables vehicle lane change methods to flexibly adapt to different vehicle types, improving the accuracy and efficiency of lane change decisions, saving R&D resources, and promoting the unified development of the technology system.

[0024] In one possible implementation of the second aspect above, the vehicle control strategy algorithm includes a lane spacing calculation equation and determines a lane change decision based on lane change scenario perception data. This includes: solving the lane spacing calculation equation based on the lane change scenario perception data to determine whether there is a feasible positive lane spacing solution for the corresponding lane distance relationship; if a positive lane spacing solution exists, the corresponding lane distance relationship is determined to be feasible, the vehicle is determined to perform a lane change, and the lane change type is determined. A lane change decision result is obtained based on the lane change type, which is either an acceleration lane change or a deceleration lane change. The lane change decision result includes the corresponding acceleration lane change command or deceleration lane change command; if no positive lane spacing solution exists, the corresponding lane distance relationship is determined to be infeasible, the vehicle is determined not to perform a lane change, and a corresponding lane change decision result is obtained, which includes the corresponding lane keeping command.

[0025] Using the above method, after confirming feasibility and safety, the lane spacing calculation equation is used to determine how to change lanes, which is crucial for generating specific control commands. This step involves finding an entry point that ensures a safe following distance from vehicles in front and behind after changing lanes. By solving for the existence of such a positive lane spacing, the strategy of accelerating or decelerating to enter lanes is determined. The feasible mathematical solution is directly mapped to specific driving action commands, achieving a seamless connection from prediction to decision. This allows the vehicle to perform the most appropriate dynamic operations to complete the lane change, improving driver comfort during the lane change process. Finally, a safety safeguard is established. Even if the preliminary feasibility and spatial predictions are successful, if a safe entry strategy cannot be found, the lane change is abandoned, ensuring driving safety. This improves the accuracy and efficiency of lane change decisions.

[0026] In one possible implementation of the second aspect above, the lane change scenario perception data includes at least one of the following: vehicle status information of the vehicle, vehicle status information of vehicles surrounding the vehicle, and traffic environment information of the traffic scenario in which the vehicle is located. The vehicle status information includes vehicle kinematic parameters. The data quality attributes include at least one of the following: data dimension, data type, data confidence, and data completeness. The vehicle performance attributes include at least one of the following: vehicle computing power, hardware capabilities, and functional requirements that the vehicle must meet in lane change decision-making.

[0027] Using the above method, the three key elements of information required for decision-making—the vehicle itself, other vehicles, and the environment—are clearly defined, ensuring the comprehensiveness and systematic nature of the decision-making basis and providing a data foundation for constructing accurate perception data quality attributes. The data quality attributes are specifically enumerated, making data quality an assessable and calculable indicator, providing a clear criterion for the performance adaptation layer. The composition of vehicle performance attributes is clarified, including not only objective capabilities such as hardware computing power but also subjective settings such as the functional requirements that the vehicle must meet in lane-changing decisions. This allows the vehicle lane-changing method to flexibly adapt to different types of vehicles, improving the effectiveness of lane-changing decisions and saving R&D resources.

[0028] Thirdly, embodiments of this application disclose a vehicle including the aforementioned vehicle lane change decision system for implementing the aforementioned vehicle lane change decision method.

[0029] The relevant beneficial effects of the third aspect mentioned above can be found in the relevant descriptions of the first and second aspects mentioned above, and will not be repeated here. Attached Figure Description

[0030] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below.

[0031] Figure 1 A schematic diagram of a vehicle lane change decision system provided in an embodiment of this application;

[0032] Figure 2 A schematic flowchart of a vehicle lane change decision-making method provided in an embodiment of this application;

[0033] Figure 3 Another schematic flowchart of the vehicle lane change decision method provided in the embodiments of this application;

[0034] Figure 4 A schematic diagram illustrating the vehicle lane change decision logic provided in this application embodiment;

[0035] Figure 5 A schematic diagram of a scenario for predicting whether a vehicle can overtake a target vehicle, provided in an embodiment of this application;

[0036] Figure 6 A schematic diagram of a scenario for predicting whether a vehicle can be overtaken by a target vehicle, provided in an embodiment of this application;

[0037] Figure 7 A schematic diagram illustrating a scenario for predicting whether there is space in front of or behind a target vehicle to perform a lane change, as provided in an embodiment of this application.

[0038] Figure 8This is a schematic diagram of a scenario for selecting acceleration or deceleration lane change, provided in an embodiment of this application. Detailed Implementation

[0039] As mentioned earlier, current lane change decision-making systems for assisted driving vehicles adopt a design approach tailored to specific vehicle models or computing platforms, resulting in inherent flaws such as rigid architecture and poor scalability. Specifically, in applications targeting low-cost vehicles, the system typically relies on pre-defined fixed rules and simplified perception and prediction models. Due to the simplicity and closed nature of its core logic, such systems have a low performance ceiling. Even by increasing the number of sensors to acquire richer environmental perception information, the decision-making logic cannot effectively integrate and utilize multi-source heterogeneous data, making it difficult to substantially improve decision quality. In applications targeting high-performance vehicles, in pursuit of optimal decision-making performance, the core lane change decision-making algorithm is often deeply coupled with highly complex perception and prediction models based on deep learning, forming a closed technology tightly dependent on high-computing-power hardware platforms. This deep binding makes it difficult to separate and port the entire system to low-computing-power platforms.

[0040] Using the aforementioned technologies presents several challenges. First, OEMs or suppliers must develop and maintain two or more completely independent lane-change decision-making systems for vehicle models with different market positioning, resulting in redundant investment of R&D resources and significant resource waste. Second, the fragmented technological approaches hinder the sharing and accumulation of algorithms, data, and experience, preventing performance improvements and functional iterations from being conducted on a unified basis, thus impeding the unified evolution and progress of the assisted driving technology system. Third, this system fragmentation caused by the separation of computing power and performance may also affect the accuracy and efficiency of lane-change decisions.

[0041] Therefore, there is an urgent need for a vehicle lane change decision system and method that can flexibly adapt to different types of vehicles, that is, can flexibly adapt to the hardware configuration, computing power level and performance of different types of driver assistance systems.

[0042] Based on the above issues, such as Figure 1As shown, this application discloses a vehicle lane change decision system applied in a vehicle. The system includes a data input layer, an algorithm configuration layer, and a decision processing layer. The data input layer is used to determine lane change scenario perception data related to lane change decision during vehicle operation and transmits the lane change scenario perception data to the decision processing layer. The algorithm configuration layer is used to store a set of lane change decision algorithms, which includes multiple candidate lane change decision algorithms applicable to different vehicle performance attributes. The decision processing layer is used to determine the data quality attributes of the lane change scenario perception data, determine the vehicle performance attributes based on the data quality attributes, select a target lane change decision algorithm corresponding to the vehicle performance attributes from the set of lane change decision algorithms stored in the algorithm configuration layer, and obtain a lane change decision result based on the target lane change decision algorithm and the lane change scenario perception data for vehicle lane change control.

[0043] By adopting the above system, through the layered architecture design of several functional layers, namely the data input layer, algorithm configuration layer, and decision processing layer, the various functions in the vehicle lane change decision system are effectively decoupled. Each functional layer is also a corresponding functional module, which realizes functional modularization. This solves the problems of rigidity and poor scalability of traditional system architecture, and makes the system have good modular flexibility and scalability.

[0044] The lane change decision is initiated by the data input layer, which collects perception data of the lane change scenario, providing a data foundation for flexible and scalable decision-making.

[0045] The algorithm configuration layer stores and manages the set of lane change decision algorithms, allowing these algorithms to be updated, optimized, or expanded independently of the entire system. When vehicle sensors are upgraded or computing power is increased, there is no need to reconstruct the entire system; only the corresponding candidate algorithms need to be added or upgraded to the algorithm set, thereby reducing R&D complexity and cost.

[0046] The decision processing layer analyzes the data quality attributes of the perceived data in lane-change scenarios, such as dimensionality and confidence level, to infer the vehicle's actual performance attributes, such as computing power and hardware capabilities. Based on this, it dynamically selects the most suitable target lane-change decision algorithm from the algorithm configuration layer. This allows low-cost vehicles to use lightweight, efficient algorithms that match limited perceived data, while high-performance vehicles can utilize more complex and precise algorithms. Consequently, system performance can smoothly scale with the quality of the input source and the level of computing power, rather than being limited by fixed rules or deeply bound to specific complex models.

[0047] Furthermore, the selected best-matching lane change decision algorithm is used to process real-time lane change scenario perception data, generating the final lane change decision result for vehicle lane change control. This seamlessly links the lane change decision result to the vehicle control system, completing a full closed loop from perception and computation to execution. This enables the deployment and operation of lane change decision schemes for different types of vehicles (i.e., vehicles with different performance attributes) within a unified technical framework. This allows the vehicle lane change system to flexibly adapt to different types of vehicles, improving the effectiveness of lane change decisions, saving R&D resources, reducing R&D costs, improving R&D efficiency, and promoting the unified development of the technology system.

[0048] Furthermore, for different types of vehicles, corresponding lane-change decision algorithms can be selected to obtain corresponding lane-change decision results. This means that a more suitable lane-change decision algorithm can be used to obtain the corresponding lane-change decision results more quickly and accurately, effectively ensuring or improving the accuracy and efficiency of lane-change decisions. Moreover, this layered architecture design can also effectively ensure or improve the accuracy and efficiency of lane-change decisions.

[0049] Furthermore, vehicle lane change control can be executed by the decision processing layer, or an additional lane change control layer can be set up in the vehicle lane change decision system and executed by the lane change control layer, or it can be executed by the lane change control system in the vehicle that is independent of the vehicle lane change decision system, which can be set as needed.

[0050] like Figure 2 As shown, this application also discloses a vehicle lane change decision method, applied to the aforementioned vehicle lane change decision system in a vehicle. The method includes:

[0051] In step S100, the data input layer determines the lane change scenario perception data related to lane change decision during vehicle driving, and transmits the lane change scenario perception data to the decision processing layer.

[0052] Lane change scenario perception data can be referred to as vehicle perception data or a set of input sources composed of vehicle perception data. Lane change scenario perception data includes at least one of the following: vehicle status information, vehicle status information of vehicles surrounding the vehicle, and traffic environment information of the traffic scenario in which the vehicle is located. Of course, lane change scenario perception data can also be other types of vehicle status-related information.

[0053] Furthermore, vehicle status information includes vehicle kinematic parameters. These include, for example, the vehicle's current speed, acceleration, heading angle, and yaw rate, as well as the speed, acceleration, and relative distance to the target vehicle and lane-changing reference vehicles. Of course, vehicle kinematic parameters can also be other kinematic parameters.

[0054] Furthermore, traffic environment information includes, for example, lane markings, traffic light status, road type (e.g., highway, urban road), weather conditions (e.g., rain, fog), and static obstacle information. Of course, traffic environment information can also be other types of traffic environment information.

[0055] Furthermore, the lane-change scenario perception data typically differs for different types of vehicles. For vehicles designed for low-cost or low-computing-power platforms, with limited sensor configurations and computing capabilities, their lane-change scenario perception data usually consists of a basic set of kinematic parameters derived from the fusion of millimeter-wave radar and cameras, such as the relative distance, speed, and acceleration of the vehicle and nearby vehicles. For vehicles designed for high-performance platforms, with richer sensor configurations and more powerful computing units, their lane-change scenario perception data can include high-precision vehicle contours and attitudes, as well as perception information predicting vehicle intent.

[0056] In step S200, the decision processing layer determines the data quality attributes of the lane change scenario perception data, determines the vehicle performance attributes based on the data quality attributes, selects the target lane change decision algorithm corresponding to the vehicle performance attributes from the lane change decision algorithm set stored in the algorithm configuration layer, and obtains the lane change decision result based on the lane change scenario perception data according to the target lane change decision algorithm for vehicle lane change control. The lane change decision algorithm set includes multiple candidate lane change decision algorithms applicable to different vehicle performance attributes.

[0057] Data quality attributes include, for example, the data quality and completeness of the input source set; vehicle performance attributes include, for example, computing power, such as low, medium, and high-performance intelligent driving platforms and functional requirement levels, as well as hardware capabilities. Further, data quality attributes include at least one of the following: data dimension, data type, data confidence, and data completeness; of course, other data quality attributes are also possible. Vehicle performance attributes include at least one of the following: vehicle computing power, hardware capabilities, and functional requirement levels that the vehicle must meet in lane-changing decisions; of course, other vehicle performance attributes are also possible.

[0058] Furthermore, the target lane change decision algorithm can be called the target decision processing algorithm.

[0059] Furthermore, lane change decisions can result in, for example, acceleration lane change instructions, deceleration lane change instructions, lane keeping instructions, etc.

[0060] Specifically, determining the data quality attributes of lane change scenario perception data at the decision processing layer means that after receiving the raw lane change scenario perception data, the layer does not directly use its numerical values. Instead, it first analyzes and evaluates the data, generating quality indicators from multiple aspects such as data dimensions, confidence level, completeness, and real-time performance to quantify the data quality. For example, a data quality attribute might be a comprehensive quality score or a feature vector containing ratings for each dimension, thus providing an objective basis for subsequent performance inference. Of course, other methods can also be used to determine the data quality attributes of lane change scenario perception data.

[0061] Determining vehicle performance attributes based on data quality attributes means inferring the vehicle's current hardware and software capability level, i.e., vehicle performance attributes, based on the data quality attributes obtained above. This can be achieved through a pre-defined mapping rule or evaluation model. High-performance perception and computing platforms typically output higher-dimensional, higher-confidence, and more complete data. Therefore, a lookup table mapping method can be used. For example, when the data quality attribute is evaluated as high-dimensional, high-confidence, and complete, it is directly mapped to a high-performance attribute; if it is evaluated as only basic kinematic parameters with moderate confidence, it is mapped to a low-performance attribute. Of course, other methods can also be used to determine vehicle performance attributes based on data quality attributes.

[0062] Based on vehicle performance attributes, the system selects a target lane change decision algorithm corresponding to the vehicle performance attributes from the set of lane change decision algorithms stored in the algorithm configuration layer. This could be a query and matching process based on vehicle configuration. The algorithm configuration layer pre-stores a set of lane change decision algorithms, where each candidate algorithm is labeled with its minimum required performance attributes or applicable performance attribute range. The decision processing layer uses the determined vehicle performance attributes as query conditions to search for all candidate algorithms in the algorithm set that match the current vehicle performance attributes. For example, if the vehicle performance attribute is determined to be low, the system will select an algorithm with low computational complexity that requires only basic data input; if the attribute is high, the system can select a computationally complex but more accurate algorithm. Finally, the system loads one or more target lane change decision algorithms that precisely correspond to the current vehicle's capabilities, thus completing the configuration of algorithm resources. Alternatively, other methods can be used to select a target lane change decision algorithm corresponding to the vehicle performance attributes from the set of lane change decision algorithms stored in the algorithm configuration layer.

[0063] Using the above method, the functions corresponding to the layered architecture of the system are transformed into specific executable process steps. This method is initiated by the data input layer, which collects lane-change scenario perception data, providing a data foundation for flexible and scalable decision-making. The decision processing layer analyzes the data quality attributes of the lane-change scenario perception data to infer the actual performance attributes of the vehicle, and dynamically selects the most suitable target lane-change decision algorithm from the algorithm configuration layer accordingly. This allows low-cost vehicles to use lightweight, efficient algorithms that match limited perception data, while high-performance vehicles can call more complex and precise algorithms. This enables the system performance to scale smoothly with the quality of the input source and the level of computing power, rather than being limited by fixed rules or deeply bound to specific complex models. Finally, the selected most suitable target lane-change decision algorithm processes the real-time lane-change scenario perception data to generate the final lane-change decision result for vehicle lane-change control. This seamlessly links the lane-change decision result to the vehicle control system, completing a complete closed loop from perception, computation to execution. This enables the deployment and operation of different types of vehicles within a unified technical framework. This allows the vehicle lane-changing system to flexibly adapt to different types of vehicles, improving the effectiveness of lane-changing decisions, saving R&D resources, reducing R&D costs, improving R&D efficiency, and promoting the unified development of the technology system. Furthermore, it effectively guarantees or improves the accuracy and efficiency of lane-changing decisions. Therefore, it can flexibly adapt to different types of vehicles to better achieve lane-changing decisions.

[0064] In one possible implementation of the above method, the decision processing layer obtains the lane change decision result based on the target lane change decision algorithm and the lane change scenario perception data, including: predicting whether the vehicle meets the overtaking trigger condition based on the lane change scenario perception data, wherein meeting the overtaking trigger condition means that the vehicle can overtake the target vehicle or can be overtaken by the target vehicle; if the overtaking trigger condition is met, predicting whether there is a first safe lane change space in the target lane where the target vehicle is located that meets the corresponding lane change requirements based on the lane change scenario perception data; if the first safe lane change space exists, determining the lane change decision result based on the lane change scenario perception data.

[0065] The first safe lane-changing space refers to the space in front of and behind the target vehicle that meets the requirements for lane changing. That is, the space needed to overtake the vehicle in front or avoid the vehicle behind. The specific size (e.g., length) of the first safe lane-changing space can be set as needed.

[0066] Furthermore, if the overtaking trigger condition is not met, the subsequent lane change decision will not be executed, and the vehicle can continue driving based on the current driving status. The system can also determine in real time or periodically whether the overtaking trigger condition is met.

[0067] Furthermore, if there is no first safe lane change space, no subsequent lane change decision processing will be performed. The vehicle can continue driving based on the current driving state, and can determine whether there is a first safe lane change space in real time or periodically.

[0068] Using the above method, based on lane-change scenario perception data, it predicts whether a vehicle meets the overtaking trigger conditions. The trigger logic for lane-change decisions is clearly defined as overtaking demand. By first determining whether there is a vehicle to be overtaken and whether there is a vehicle that can be overtaken, meaningless calculations of lane-change space are avoided, improving the efficiency of the lane-change decision-making process. The overtaking trigger conditions are further subdivided into two scenarios: the vehicle actively overtaking and the vehicle being overtaken, covering a more comprehensive range of real-world driving interactions and enhancing adaptability to complex traffic scenarios. After determining that there is an overtaking demand, the immediate safety of the target lane is further assessed. Feasibility and safety are judged step by step, reducing the risk of collisions caused by misjudgments of space. After completing the demand and safety verification, a specific decision result is finally generated. This step-by-step, progressive lane-change decision-making process ensures the stability and reliability of lane-change decisions, thereby improving the accuracy and efficiency of lane-change decisions.

[0069] In one possible implementation of the above method, the decision processing layer obtains the lane change decision result based on the target lane change decision algorithm and the lane change scenario perception data. It also includes: when a vehicle has a lane change intention, determining whether there is a second safe lane change space that meets the corresponding lane change requirements in the lane corresponding to the lane change intention; if there is a second safe lane change space, obtaining the lane change decision result of directly executing the lane change; if there is no second safe lane change space, obtaining the lane change decision result based on the target lane change decision algorithm and the lane change scenario perception data.

[0070] The second safe lane-change space refers to the space in the lane corresponding to the intended lane change that directly meets the lane-change requirements, such as navigation instructions or driver levers. In other words, it is the space required to achieve a predetermined route change. The specific size (e.g., length) of the second safe lane-change space can be set as needed, and it is usually larger than the first safe lane-change space. That is, if the lane-change space is large enough or safe enough, the lane change is executed directly.

[0071] When a vehicle intends to change lanes, determining whether a second safe lane-changing space exists in the lane corresponding to the intention to change lanes, satisfying the corresponding lane-changing requirements, involves the system first identifying two vehicles in the target lane that define a pre-cut-in gap: the target vehicle (the vehicle in front of the vehicle after it cuts in) and the reference vehicle (the vehicle behind the vehicle after it cuts in). Then, the system invokes a lane-changing decision algorithm, calculating the predicted distance between the two vehicles within a predicted timeframe based on current lane-changing scenario perception data such as speed, acceleration, and relative distance. Finally, this predicted distance is compared to a preset lane-changing safety distance threshold. If the predicted distance is greater than or equal to the preset safety distance threshold, a second safe lane-changing space is deemed to exist. For example, if the calculated predicted distance between the two vehicles after 3 seconds is 48 meters, and the system's safety threshold is 40 meters, since 48 meters > 40 meters, the space is deemed sufficient, and a safe lane-changing is permitted.

[0072] Furthermore, in the presence of a second safe lane-changing space, the lane-changing decision result can be obtained for direct execution. For example, corresponding lane-changing decision result data or instructions can be generated and sent to the vehicle's control layer or control system, thereby driving the vehicle's steering, driving, and braking actuators to complete the lane change.

[0073] By employing the above method, when a vehicle intends to change lanes, it determines whether a second safe lane-changing space exists in the lane corresponding to the lane-changing intention, such as an active lane-changing intention triggered by the driver or navigation, thus making the application scenario more complete. For active lane changes with a second safe lane-changing space, a lane-changing decision result is obtained by directly executing the lane change, achieving rapid response in lane-changing decisions and improving the system's execution efficiency. When a direct lane change does not exist with a second safe lane-changing space, a lane-changing decision result is obtained based on the target lane-changing decision algorithm and the lane-changing scenario perception data, making the method more adaptable to different scenarios. This allows the vehicle lane-changing method to flexibly adapt to different types of vehicles, improving the accuracy and efficiency of lane-changing decisions, saving R&D resources, and promoting the unified development of the technology system.

[0074] In one possible implementation of the above method, the lane change decision algorithm includes a vehicle safety verification algorithm, a vehicle control space prediction algorithm, and a vehicle control strategy algorithm. The vehicle safety verification algorithm can be referred to as a safety verification method, the vehicle control space prediction algorithm as a vehicle control space prediction method, and the vehicle control strategy algorithm as a vehicle control strategy.

[0075] Using the above method, the lane change decision algorithm is deconstructed into three functionally defined and clearly demarcated sub-algorithm modules. This modular definition makes the construction, updating, and maintenance of the algorithm set more systematic, and also facilitates the independent optimization or replacement of a sub-algorithm for different types of vehicles, further enhancing the system's configurability and scalability, thus enabling the vehicle lane change method to flexibly adapt to different types of vehicles.

[0076] Of course, lane change decision algorithms can also include other algorithms.

[0077] In one possible implementation of the above method, the vehicle safety verification algorithm includes an overtaking calculation equation. Based on the lane change scenario perception data, it predicts whether the vehicle meets the overtaking trigger condition. This includes: solving the overtaking calculation equation based on the lane change scenario perception data to determine whether there is a feasible positive time solution for the corresponding overtaking relationship; if a positive time solution exists, the corresponding overtaking relationship is determined to be feasible, and the vehicle is determined to meet the overtaking trigger condition. The smaller the value of the positive time solution, the higher the overtaking efficiency; if no positive time solution exists, the corresponding overtaking relationship is determined to be infeasible, and the vehicle is determined not to meet the overtaking trigger condition.

[0078] The overtaking calculation equation includes the overtaking target calculation formula and the overtaken calculation formula. When the overtaking trigger condition is that the vehicle can overtake the target vehicle, the corresponding overtaking calculation equation is the overtaking target calculation formula; when the overtaking trigger condition is that the vehicle can be overtaken by the target vehicle, the corresponding overtaking calculation equation is the overtaken overtaking calculation formula.

[0079] By employing the above method, the judgment of whether overtaking is permissible is concretized into a computable mathematical equation, transforming the decision-making process from rule-based empirical judgment to equation-based quantitative calculation, thus improving the objectivity and accuracy of the decision. By solving the equation to find a positive time-varying solution, the feasibility of overtaking is transformed into a mathematical optimization problem. This allows for comprehensive consideration of various kinematic parameters such as the speed, acceleration, and relative position of both the driver and the target vehicle, enabling accurate future state predictions and making the decision more scientific. The existence of a positive time-varying solution determines the feasibility of the corresponding overtaking relationship, not only determining whether overtaking is possible but also quantifying the overtaking efficiency through the magnitude of the solution. This provides crucial quantitative evidence for selecting the optimal strategy from multiple feasible options, contributing to more efficient driving behavior. The absence of a positive time-varying solution determines the infeasibility of the corresponding overtaking relationship, avoiding forced lane changes in scenarios with excessive risk or physical impossibility, ensuring driving safety. This improves the accuracy and efficiency of lane-change decisions.

[0080] In one possible implementation of the above method, the vehicle control space prediction algorithm includes a lane change space calculation equation. Based on lane change scenario perception data, it predicts whether there is a first safe lane change space in the target lane where the target vehicle is located that meets the corresponding lane change requirements. This includes: solving the lane change space calculation equation based on the lane change scenario perception data; calculating the predicted distance between the target vehicle and the corresponding lane change reference vehicle within the prediction time, where the lane change reference vehicle is either a vehicle in front of the target vehicle or a vehicle behind the target vehicle; comparing the predicted distance with a preset lane change safety distance threshold; and if the predicted distance is greater than or equal to the lane change safety distance threshold, determining that there is a first safe lane change space in the target lane that meets the corresponding lane change requirements.

[0081] The equation for calculating lane change space can be called the formula for predicting the space in front of and behind the target vehicle. The lane change safety distance threshold can be set as needed, and is usually less than the lane change safety distance threshold corresponding to the aforementioned second safe lane change space.

[0082] In addition, the reference vehicle for lane changing is either the vehicle in front of the target vehicle or the vehicle behind the target vehicle. For example, when the overtaking trigger condition is met and the vehicle can overtake the target vehicle, the corresponding reference vehicle for lane changing is the vehicle in front of the target vehicle; when the overtaking trigger condition is met and the vehicle can be overtaken by the target vehicle, the corresponding reference vehicle for lane changing is the vehicle behind the target vehicle.

[0083] By employing the above method, the determination of whether a safe space exists is concretized into another calculable mathematical equation, ensuring the accuracy of the safety assessment. Solving the lane change space calculation equation is used to predict changes in traffic flow gaps in the target lane over a future period, dynamically evaluating lane change decisions rather than solely considering the current instantaneous distance, which better aligns with the actual need for space prediction in driving. By comparing the dynamic prediction results with static safety thresholds, a clear binary judgment of safe or unsafe is made. This method standardizes the complex safety assessment, allowing it to be implemented and adjusted across different lane change decision algorithms. This enables vehicle lane change methods to flexibly adapt to different vehicle types, improving the accuracy and efficiency of lane change decisions, saving R&D resources, and promoting the unified development of the technology system.

[0084] In one possible implementation of the above method, the vehicle control strategy algorithm includes a lane spacing calculation equation and, based on lane change scenario perception data, determines a lane change decision, including: solving the lane spacing calculation equation based on the lane change scenario perception data to determine whether there is a feasible positive lane spacing solution for the corresponding lane distance relationship; if a positive lane spacing solution exists, the corresponding lane distance relationship is determined to be feasible, the vehicle is determined to execute a lane change, and the lane change type is determined; a lane change decision result is obtained based on the lane change type, which is either an acceleration lane change or a deceleration lane change, and the lane change decision result includes the corresponding acceleration lane change command or deceleration lane change command; if no positive lane spacing solution exists, the corresponding lane distance relationship is determined to be infeasible, the vehicle is determined not to execute a lane change, and a corresponding lane change decision result is obtained, which includes the corresponding lane keeping command.

[0085] The equation for calculating the passage distance can be called the formula for calculating the passable distance in a vehicle acceleration scenario.

[0086] Furthermore, if a positive distance solution exists, the corresponding travel distance relationship is deemed feasible, the vehicle is instructed to change lanes, and the lane change type is determined. Based on the lane change type, a lane change decision is obtained. Specifically, first, the system obtains a positive distance solution by solving the travel distance calculation equation. This solution mathematically proves that there exists at least one future moment in which the vehicle can maintain a safe distance from the target lane vehicle, thus confirming the basic feasibility of the lane change. Subsequently, the system selects the optimal solution from all feasible solutions, such as the one corresponding to the shortest access time. By comparing the vehicle's predicted trajectory based on the optimal solution with the trajectory maintaining the current speed, the system determines the operation required to achieve the optimal solution. If the predicted trajectory needs to lead the current trajectory, it is determined to be an acceleration lane change; if it needs to lag, it is determined to be a deceleration lane change. Finally, the system converts the determined lane change type into an executable instruction.

[0087] Using the above method, after confirming feasibility and safety, the lane spacing calculation equation is used to determine how to change lanes, which is crucial for generating specific control commands. This step involves finding an entry point that ensures a safe following distance from vehicles in front and behind after changing lanes. By solving for the existence of such a positive lane spacing, the strategy of accelerating or decelerating to enter lanes is determined. The feasible mathematical solution is directly mapped to specific driving action commands, achieving a seamless connection from prediction to decision. This allows the vehicle to perform the most appropriate dynamic operations to complete the lane change, improving driver comfort during the lane change process. Finally, a safety safeguard is established. Even if the preliminary feasibility and spatial predictions are successful, if a safe entry strategy cannot be found, the lane change is abandoned, ensuring driving safety. This improves the accuracy and efficiency of lane change decisions.

[0088] Furthermore, after obtaining the lane change decision result, the aforementioned lane change control layer or lane change control system can execute vehicle lane change control based on the lane change decision result.

[0089] The following example, using a vehicle acceleration / deceleration lane-changing decision architecture and method based on a scalable input source configured on a low-computing-power platform with only a basic input set as the input source, further illustrates the vehicle lane-changing decision scheme provided in this application.

[0090] The basic input source set consists only of kinematic parameters, such as the speed / position of the vehicle, the preceding vehicle, and the following vehicle. The vehicle acceleration / deceleration lane-changing decision-making architecture based on scalable input sources, configured on the aforementioned low-computing-power platform, includes an input interface module, an algorithm configuration module, and a decision processing module. The input interface module is the data input layer, the algorithm configuration module is the algorithm configuration layer, and the decision processing module is the decision processing layer.

[0091] like Figure 3 As shown, the lane change decision-making method under this framework includes the following steps.

[0092] Step S101: The input interface module receives vehicle perception data, which constitutes an input source set. The input source set is then sent to the decision processing module.

[0093] Vehicle perception data can be, for example, the aforementioned kinematic parameters collected by cameras, sensors, etc. in the vehicle.

[0094] Step S102: The decision processing module determines the computing power level based on the quality and completeness of the input source set, and selects the corresponding target decision processing algorithm from the multiple decision processing algorithms stored in the algorithm configuration module for the vehicle control strategy, vehicle control space prediction method, and safety verification method according to the computing power level.

[0095] Step S103: The decision processing module performs calculations and judgments based on the selected target decision processing algorithm and the vehicle perception data to obtain the lane change decision result.

[0096] like Figure 4As shown, the logical flow of a vehicle lane-changing decision can be as follows: After the lane-changing intention takes effect, the system first determines whether there is directly usable lane-changing space in the target lane, i.e., the second safe lane-changing space. If there is directly usable lane-changing space, the corresponding lane-changing process is performed directly. If there is no directly usable lane-changing space, an overtaking-based strategy is initiated. This strategy is divided into two parallel branches, and each branch includes three steps. The first step of the first branch is to predict whether the vehicle can overtake the target vehicle in the target lane. If so, the corresponding second step is executed to predict or check whether there is corresponding lane-changing space in front of the target vehicle, i.e., the first safe lane-changing space. If not, the previous step is executed to determine whether there is directly usable lane-changing space in the target lane. The first step of the second branch is to predict whether the vehicle will be overtaken by the target vehicle in the target lane. If so, the corresponding second step is executed to predict or check whether there is corresponding lane-changing space behind the target vehicle, i.e., the first safe lane-changing space. If not, the previous step is executed to determine whether there is directly usable lane-changing space in the target lane. If available space is found in the corresponding direction of any branch, i.e., there is corresponding lane-changing space in front of or behind the target vehicle, this information will be transmitted to the acceleration / deceleration selector to execute the third step to determine the specific lane-changing decision. If there is no corresponding lane-changing space in front of or behind the target vehicle, the previous step will be further executed in real time or periodically to determine whether the vehicle can overtake or be overtaken by the target vehicle. If so, the same process as described above will be followed. If not, it will be determined whether there is directly usable lane-changing space in the target lane. Further processing is as described above and will not be repeated here.

[0097] Furthermore, if none of the above strategy branches find available lane-changing space, the system may determine that lane changing is not currently possible and output the corresponding decision result, such as keeping the lane.

[0098] Specifically, regarding the first step mentioned above: predicting whether the vehicle can overtake the target vehicle or be overtaken by the target vehicle, this can be achieved through the following process.

[0099] Based on lane change scenario perception data, a safety verification method is executed. This method includes formulas for calculating the overtaking target and the overtaken vehicle. By solving these formulas, it is determined whether a feasible positive-time solution exists for the corresponding overtaking relationship. If a positive-time solution exists, the overtaking relationship is deemed feasible, satisfying the overtaking trigger condition, allowing the vehicle to overtake the preceding vehicle or be overtaken by the following vehicle. A smaller solution value indicates higher overtaking efficiency. If no positive-time solution exists, the overtaking relationship is deemed infeasible, and the overtaking trigger condition is not met.

[0100] The formula for calculating the target exceedance is as follows:

[0101]

[0102] in, The speed of the vehicle behind the overtaking window (as an example of a target vehicle); The current speed of the vehicle (i.e., the vehicle itself). To accelerate the vehicle ,For example The initial suggested value is 2; dr represents the relative distance between the vehicle and the vehicle behind the overtaking window at the current moment, and r represents the vehicle behind the overtaking window. The acceleration of the vehicle behind the overtaking window; Set a time scale, for example, 0.3s; The desired speed for the car to accelerate and overtake. The solution is a positive time value.

[0103] If t has no positive value, it means the vehicle is predicted to be unable to overtake the target vehicle. If t has a positive value, it means the vehicle is predicted to be able to overtake the target vehicle, and the shorter the time, the higher the overtaking efficiency.

[0104] An example scenario is a prediction of whether the vehicle can overtake a target vehicle, i.e., a diagram illustrating whether the vehicle can overtake the target vehicle, such as... Figure 5 As shown in the diagram, car A represents the driver's vehicle, car B represents the vehicle behind the overtaking window (i.e., the target vehicle), and car C represents the vehicle in front of the overtaking window (as an example of a reference vehicle for lane changing). The current time in the diagram represents the starting point for calculating all distances and positional relationships shown. v_r is the aforementioned... , where v_0 is the speed of the vehicle behind the overtaking window (as an example of a target vehicle). dr represents the current speed of the vehicle. That is, the aforementioned dtp_r represents the relative distance between the current vehicle and the vehicle behind the overtaking window at the current moment. dtp_r represents the predicted distance traveled by the vehicle behind the overtaking window relative to the starting point at a certain prediction time T_P. dtp_0 represents the distance traveled at prediction time T_P. At that time, the predicted distance traveled by the vehicle relative to the starting point. T_P corresponds to the positive time solution obtained from the overtaking calculation equation. ds_r is the safe distance after a lane change, and v_T*tG is a fixed safe buffer distance preset by the system, where v_T is the aforementioned safe buffer distance. Let t_G be the desired speed at which the vehicle accelerates to overtake, and t_G is the aforementioned speed. The time calibrator is a fixed time interval preset by the system. This distance represents the distance traveled by the vehicle at the target speed v_T for t_G time, and is used as part of the minimum safety margin to be reserved when changing lanes. Alternatively, ds_r could be, for example, v_T*tG+10.

[0105] The formula for calculating what is surpassed is as follows:

[0106]

[0107] in, The speed of the vehicle in front of the waiting window; Acceleration of the vehicle in front of the waiting window. The desired speed of the vehicle is generally determined by the vehicles behind it at the waiting window; The relative distance between the current vehicle and the vehicle in front of the waiting window is dr, where dr represents the relative distance and f represents the vehicle in front of the waiting window; other parameters are the same as above and will not be repeated.

[0108] If t has no positive value, it means the predicted vehicle cannot be overtaken by the target vehicle. If t has a positive value, it means the predicted vehicle can be overtaken by the target vehicle, and the shorter the time, the higher the overtaking efficiency.

[0109] An example scenario for predicting whether a vehicle can be overtaken by a target vehicle, i.e., a diagram illustrating the overtaking of the vehicle, is shown below. Figure 6 As shown in the diagram, car A represents the driver's vehicle, car B represents the vehicle behind the waiting window (as an example of a reference vehicle for lane changing), and car C represents the vehicle in front of the waiting window (as the target vehicle). The current time in the diagram represents the starting point for calculating all distances and positional relationships shown. v_f is the aforementioned... , where v_0 represents the speed of the vehicle in front of the waiting window. v_0 is the aforementioned... dr_f is the current speed of the vehicle. dtp_f is the current relative distance between the vehicle and the vehicle in front of the waiting window. dtp_f is the predicted distance traveled by the vehicle in front of the waiting window after T_P seconds. dtp_0 is the predicted distance traveled by the vehicle after T_P seconds, at which point the vehicle, as the rear vehicle in the overtaking scenario, is the vehicle that was originally leading in the target lane and will be overtaken by the vehicle C behind it. v_T is the aforementioned... ds_f is the desired speed of the vehicle. ds_f is the minimum safe distance between the vehicle and the vehicle in front of the waiting window to allow for lane changing. For example, a suggested value for this distance is v_T*TG+10.

[0110] Regarding the second step mentioned above: predicting whether there is space in front of or behind the target vehicle to perform a lane change, i.e. whether the overtaking window or waiting window is larger than the lane change space, can be achieved through the following process.

[0111] After the overtaking trigger condition is met, a vehicle control space prediction method is executed. This method includes a formula for predicting the space in front of and behind the target vehicle. Based on the perception data of the lane change scenario, this formula is solved to calculate the predicted distance between the vehicle in front (the target vehicle) and the vehicle behind (the lane change reference vehicle) in the target lane within the prediction time. The predicted distance is compared with a preset lane change safety distance threshold. If the predicted distance is greater than or equal to the lane change safety distance threshold, it is determined that the space in front of and behind the target vehicle meets the lane change requirement.

[0112] The formula for predicting the front and rear space of the target vehicle is as follows:

[0113]

[0114] in, The relative distance between the vehicles in front and behind at the overtaking window; To meet the spatial requirements for lane changing; This refers to the speed of the vehicle ahead of the overtaking window; This refers to the speed of vehicles behind the overtaking window. Accelerate the vehicle in front of the overtaking window; This represents the acceleration of vehicles behind the overtaking window. Other parameters are the same as above and will not be repeated.

[0115] When the above formula holds true, it means that after t seconds, the space between the two target vehicles (i.e., the target vehicle and the corresponding lane-changing reference vehicle) is sufficient for lane changing.

[0116] For example, one scenario is predicting whether there is space in front of or behind the target vehicle to perform a lane change, i.e., a diagram showing the lane change space within the overtaking window, such as... Figure 7 As shown in the diagram, car A represents the driver's vehicle, car B represents the vehicle behind the overtaking window (i.e., the target vehicle), and car C represents the vehicle in front of the overtaking window (as an example of a reference vehicle for lane changing). The current time in the diagram represents the starting point for calculating all distances and positional relationships shown. v_r is the aforementioned... , where v_0 is the speed of the vehicle behind the overtaking window. v_0 is the aforementioned... d0_fr represents the vehicle's current speed. dtp_fr represents the relative distance between the vehicles in front and behind at the current moment in the overtaking window. dtp_fr represents the relative distance between the vehicles in front and behind in the overtaking window after T_P seconds. v_T is the aforementioned... , is the expected speed for a car to accelerate and overtake.

[0117] Similarly, if you need to slow down and wait to be overtaken, you need to calculate whether the distance of the waiting window is greater than the lane change space. The calculation method is the same as that for the overtaking window, so I will not repeat it here.

[0118] Regarding the third step above: choosing between accelerating or decelerating when changing lanes based on efficiency and safety considerations can be achieved through the following process.

[0119] Once there is sufficient space in front of and behind the target vehicle to meet the lane-changing requirement, a vehicle control strategy is executed. This strategy includes a formula for calculating the passable distance in the vehicle's acceleration scenario. Based on the perceived data of the lane-changing scenario, this formula is solved to determine whether there exists a feasible positive distance solution for the corresponding passable distance relationship. If a positive distance solution exists, the passable distance relationship is deemed feasible, and the specific lane-changing type is determined, generating a lane-changing decision result that includes either an acceleration lane-changing command or a deceleration lane-changing command. If no positive distance solution exists, the passable distance relationship is deemed infeasible, and a lane-changing decision result that includes a lane-keeping command is generated.

[0120] The formula for calculating the passable distance in the vehicle acceleration scenario is as follows:

[0121]

[0122] in, It is a fixed time parameter representing the buffer time for the vehicle's reaction; This is a time parameter, related to the time to collision, used to ensure safety; The speed of the nearest vehicle in front of the current vehicle in the current lane; This represents the acceleration of the nearest vehicle ahead of the current vehicle in the lane. Other parameters are the same as above and will not be repeated.

[0123] When the above formula holds true, it means that the vehicle has room to accelerate to the target lane change area.

[0124] For example, choosing to accelerate and change lanes requires considering the distance relationship with the vehicle in front, i.e., predicting the distance between the vehicle in front and the time needed to satisfy the overtaking and lane-changing maneuver, as shown in the diagram. Figure 8 As shown in the diagram, car A represents the driver's vehicle, car B represents the vehicle behind the overtaking window (i.e., the target vehicle), and car C represents the vehicle in front of the overtaking window (as an example of a reference vehicle for lane changing). Car D represents the vehicle in front of the driver's vehicle (i.e., the closest vehicle in front of the driver's vehicle in the current lane). The current time in the diagram represents the starting point for calculating all distances and positional relationships shown. v_r is the aforementioned... , where v_0 is the speed of the vehicle behind the overtaking window. v_0 is the aforementioned... dr_cipv is the current speed of the vehicle. dr_cipv is the relative distance between the vehicle and the nearest preceding vehicle (cipv) in the current lane at the current moment. dtp_cipv is the distance traveled by the nearest preceding vehicle (cipv) in the current lane after T_P seconds. dtpr_cipv is the predicted relative distance between the vehicle and the preceding vehicle just before the vehicle crosses the line after a lane change; this relative distance should be greater than the current following distance. dtp_0 is the distance traveled by the vehicle after T_P seconds. d_S1S2 are the values ​​in the above formula for calculating the passable distance. The distance corresponding to the buffer time during the vehicle's acceleration process represents the distance the vehicle travels while maintaining that speed after accelerating from its current state to v_T. v_T is the distance mentioned above. This refers to the desired speed at which the vehicle accelerates to overtake.

[0125] Furthermore, once it's confirmed that there's space to change lanes in the target lane, the acceleration / deceleration selector receives one or two prediction times. If only one prediction time is received, it means the vehicle can only accelerate or decelerate. If it's an acceleration prediction time, it calculates whether there are other vehicles inhibiting the vehicle's acceleration. The formula for calculating the passable distance is the same as the formula for calculating the passable distance in the vehicle acceleration scenario described above. If the formula holds true, it means the vehicle has space to accelerate to the target area. If two prediction times are received simultaneously, acceleration and deceleration, in addition to considering whether the vehicle can accelerate or decelerate, efficiency must also be considered, selecting the shorter time option. That is, if two prediction times (acceleration and deceleration) are received simultaneously, it's not only necessary to consider whether the vehicle can accelerate or decelerate, but also efficiency. For example, if the acceleration time is 10 seconds and the deceleration time is 5 seconds, then selecting the shorter time option is more efficient.

[0126] Subsequently, the decision processing module, or the corresponding lane change control module, lane change control system, etc., can control the vehicle's steering, drive and braking systems to perform corresponding driving control actions based on the lane change decision results, thus completing the lane change decision and execution.

[0127] Similarly, if you need to slow down and wait to be overtaken, you need to choose to slow down and change lanes, taking into account the distance relationship with the vehicles behind you. The calculation method is the same as the overtaking window, so I will not repeat it here.

[0128] The lane change decision scheme provided in this application can be described as a vehicle acceleration / deceleration lane change decision architecture and method based on scalable input sources. By constructing a layered and modular unified system architecture, the unchanging lane change decision logic is separated from the variable algorithm implementation. On the one hand, it provides a unified technical framework for vehicles of different price points and positioning, reducing R&D complexity and costs, and solving the problems of R&D duplication and resource waste caused by fragmented architecture. On the other hand, it achieves performance scalability, with system performance smoothly increasing as the available input sources and computing power increase. This not only meets the needs of low-cost, low-computing-power vehicles to implement core functions through basic input sets and simple algorithms, but also supports high-performance, high-computing-power vehicles to improve performance through rich input sources and upgraded algorithms, breaking the dilemma of low performance limits for traditional low-computing-power platforms and the inability of high-computing-power systems to adapt downwards. Furthermore, this layered architecture design can also guarantee or improve the efficiency of lane change decisions. Furthermore, the robustness and effectiveness of the system architecture were verified through a low-computing-power platform example, and a clear technical evolution path was provided for the upgrade of functions from L2 to L3, namely, enriching the input sources and upgrading the algorithm without overturning and reconstructing the entire lane-changing decision system, thereby promoting the unified development of the technology system.

[0129] Furthermore, this solution is applicable to all intelligent driving platforms with low, medium, and high-level algorithms. It does not have strong requirements for input sources, but the richer the input sources, the more accurate the output results.

[0130] This application also discloses a vehicle including the above-described vehicle lane change decision system for implementing the above-described vehicle lane change decision method.

[0131] It should be noted that, in addition to the specific embodiments described above, those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Although the description of this application is presented in conjunction with preferred embodiments, this does not mean that the features of this application are limited to these embodiments. On the contrary, the purpose of describing the application in conjunction with the embodiments is to cover other options or modifications that may be derived based on the claims of this application. To provide a thorough understanding of this application, many specific details are included in the above description, and this application may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of this application, some specific details will be omitted in the description. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0132] It should be noted that in this specification, similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0133] The terms “first”, “second”, etc., are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0134] Although this application has been illustrated and described with reference to certain preferred embodiments, those skilled in the art should understand that the above description is a further detailed explanation of the application in conjunction with specific embodiments, and should not be construed as limiting the specific implementation of the application to these descriptions. Those skilled in the art can make various changes in form and detail, including some simple deductions or substitutions, without departing from the spirit and scope of this application.

Claims

1. A vehicle lane-change decision-making system, characterized in that, Applied to vehicles, the system includes a data input layer, an algorithm configuration layer, and a decision processing layer, wherein... The data input layer is used to determine lane change scenario perception data related to lane change decision during the vehicle's driving process, and to transmit the lane change scenario perception data to the decision processing layer. The algorithm configuration layer is used to store a set of lane change decision algorithms, which includes multiple candidate lane change decision algorithms suitable for different vehicle performance attributes. The decision processing layer is used to determine the data quality attributes of the lane change scenario perception data, determine the vehicle performance attributes of the vehicle based on the data quality attributes, select the target lane change decision algorithm corresponding to the vehicle performance attributes from the set of lane change decision algorithms stored in the algorithm configuration layer based on the vehicle performance attributes, and obtain the lane change decision result based on the target lane change decision algorithm and the lane change scenario perception data for use in vehicle lane change control.

2. A vehicle lane-changing decision-making method, characterized in that, A vehicle lane change decision-making system applied in vehicles, the system comprising a data input layer, an algorithm configuration layer, and a decision processing layer, the method comprising: The data input layer determines the lane change scenario perception data related to lane change decision during the vehicle's driving process, and transmits the lane change scenario perception data to the decision processing layer; The decision processing layer determines the data quality attributes of the lane change scenario perception data, determines the vehicle performance attributes of the vehicle based on the data quality attributes, selects a target lane change decision algorithm corresponding to the vehicle performance attributes from the lane change decision algorithm set stored in the algorithm configuration layer, and obtains a lane change decision result based on the target lane change decision algorithm and the lane change scenario perception data for use in vehicle lane change control. The lane change decision algorithm set includes multiple candidate lane change decision algorithms applicable to different vehicle performance attributes.

3. The method according to claim 2, characterized in that, Based on the target lane change decision algorithm, and according to the lane change scenario perception data, the lane change decision result is obtained, including: Based on the lane change scenario perception data, predict whether the vehicle meets the overtaking trigger condition. Meeting the overtaking trigger condition means that the vehicle can overtake the target vehicle or can be overtaken by the target vehicle. If the overtaking triggering conditions are met, based on the lane change scenario perception data, it is predicted whether there is a first safe lane change space in the target lane where the target vehicle is located that meets the corresponding lane change requirements. If the first safe lane-changing space exists, the lane-changing decision result is determined based on the lane-changing scenario perception data.

4. The method according to claim 2 or 3, characterized in that, Based on the target lane change decision algorithm, and according to the lane change scenario perception data, the lane change decision result is obtained, including: When the vehicle intends to change lanes, determine whether there is a second safe lane-changing space in the lane corresponding to the lane-changing intention that meets the corresponding lane-changing requirements; In the case of the existence of the second safe lane change space, the lane change decision result of directly executing the lane change is obtained; In the absence of the second safe lane change space, the lane change decision result is obtained based on the target lane change decision algorithm and the lane change scenario perception data.

5. The method according to claim 4, characterized in that, The lane change decision algorithm includes a vehicle safety verification algorithm, a vehicle control space prediction algorithm, and a vehicle control strategy algorithm.

6. The method according to claim 5, characterized in that, The vehicle safety verification algorithm includes an overtaking calculation equation, which predicts whether the vehicle meets the overtaking trigger conditions based on the lane change scenario perception data, including: Based on the lane change scenario perception data, the overtaking calculation equation is solved to determine whether there is a feasible positive time solution for the corresponding overtaking relationship; If the positive time solution exists, the corresponding overtaking relationship is determined to be feasible, the vehicle is determined to meet the overtaking triggering condition, and the smaller the value of the positive time solution, the higher the overtaking efficiency. If no positive time solution exists, the corresponding overtaking relationship is determined to be infeasible, and the vehicle is determined not to meet the overtaking trigger condition.

7. The method according to claim 6, characterized in that, The vehicle control space prediction algorithm includes a lane change space calculation equation, which, based on the lane change scenario perception data, predicts whether there is a first safe lane change space in the target lane where the target vehicle is located that meets the corresponding lane change requirements, including: Based on the lane change scenario perception data, the lane change space calculation equation is solved to calculate the predicted distance between the target vehicle and the lane change reference vehicle corresponding to the target vehicle within the prediction time. The lane change reference vehicle is either a vehicle in front of the target vehicle or a vehicle behind the target vehicle. The predicted distance is compared with a preset lane change safety distance threshold. If the predicted distance is greater than or equal to the lane change safety distance threshold, it is determined that there is a first safe lane change space in the target lane that meets the corresponding lane change requirements.

8. The method according to claim 7, characterized in that, The vehicle control strategy algorithm includes a traffic spacing calculation equation, and determines the lane change decision result based on the lane change scenario perception data, including: Based on the lane change scenario perception data, the traffic spacing calculation equation is solved to determine whether there is a feasible positive spacing solution for the corresponding traffic distance relationship; If the positive distance solution exists, the corresponding passage distance relationship is determined to be feasible, the vehicle is determined to change lanes, and the lane change type is determined. The lane change decision result is obtained according to the lane change type, which is either an acceleration lane change or a deceleration lane change. The lane change decision result includes the corresponding acceleration lane change command or deceleration lane change command. If no positive distance solution exists, the corresponding travel distance relationship is determined to be infeasible, the vehicle is determined not to change lanes, and the corresponding lane change decision result is obtained, the lane change decision result including the corresponding lane keeping instruction.

9. The method according to claim 8, characterized in that, The lane change scenario perception data includes at least one of the following: vehicle status information of the vehicle, vehicle status information of vehicles surrounding the vehicle, and traffic environment information of the traffic scenario in which the vehicle is located, and the vehicle status information includes vehicle kinematic parameters. The data quality attributes include at least one of the following: data dimension, data type, data confidence level, and data completeness. The vehicle performance attributes include at least one attribute from the vehicle's computing power, hardware capabilities, and functional requirements that the vehicle must meet in lane change decisions.

10. A vehicle, characterized in that, The system includes the vehicle lane change decision system as described in claim 1, used to implement the vehicle lane change decision method as described in any one of claims 2-9.

Images