Decision control integrated automatic driving vehicle control instruction generation method and device

Abstract

The embodiment of the invention discloses a decision-making and control integrated automatic driving vehicle control instruction generation method and device. The method comprises the following steps: acquiring predicted motion trails of one or more traffic participants around the vehicle; determining hard security constraint information of one or more traffic objects in a near-end time domain according to the predicted motion trail; determining interaction potential field information between the vehicle and the one or more traffic objects in a far-end time domain according to the predicted motion trail; and determining a control instruction of the vehicle according to the rigid safety constraint information and the interaction potential field information. According to the scheme, decision control tasks are modeled into OCP in a unified mode, immediate safety and traffic rule following are guaranteed through near-end time domain hard constraints, long-term behavior optimization is achieved through far-end time domain potential field guidance, the problems that layered architecture targets are difficult to coordinate, and a data driving method is weak in interpretation are effectively solved, and the method is suitable for being applied to the field of behavior optimization. And the track generation robustness and adaptability in a complex dynamic traffic scene are improved.

Description

Technical Field

[0001] This application relates to the field of autonomous driving, and in particular to a method and apparatus for generating control commands for autonomous vehicles that integrates decision-making and control. Background Technology

[0002] In autonomous driving systems, the decision-making and control processes directly determine vehicle safety, passenger comfort, and traffic efficiency. Vehicles need to make reasonable driving judgments based on real-time environmental conditions and traffic rules, and translate these judgments into specific steering, acceleration, or braking commands.

[0003] However, in real-world urban traffic, traffic scenarios are dynamic and ever-changing, and the behavior of surrounding vehicles and pedestrians is often unpredictable. Systems must quickly achieve a balance between multiple objectives, including safety, comfort, and efficiency, which significantly increases the difficulty of decision-making and control. Therefore, achieving coordinated optimization of decision-making and control in complex and highly uncertain scenarios remains a key challenge in current autonomous driving research. Summary of the Invention

[0004] To address the aforementioned shortcomings, this application provides a method and apparatus for generating control commands for autonomous vehicles that integrates decision-making and control. This approach divides the prediction time domain of model predictive control (MPC) into a near-end time domain and a far-end time domain, and integrates hard safety constraints and interactive potential fields. The decision-making and control tasks are unified into a single optimal control problem (OCP), which is then solved online within the MPC framework. Furthermore, while ensuring driving safety and rule compliance, the vehicle achieves trajectory tracking of the reference path. Ultimately, this method effectively overcomes the coordination difficulties of hierarchical architectures and the poor interpretability of data-driven methods, significantly improving the decision-making and control performance of autonomous driving systems in complex dynamic traffic environments, and enhancing the system's safety, comfort, and traffic efficiency.

[0005] Firstly, a method for generating control commands for autonomous vehicles integrating decision-making and control is provided. The method includes: acquiring the predicted motion trajectories of one or more traffic participants surrounding the vehicle, wherein the predicted motion trajectory includes the planar position and state of the one or more traffic participants at multiple discrete moments after the current moment; and determining, based on the predicted motion trajectories of the one or more traffic participants, hard safety constraint information of one or more traffic objects in the near-end time domain, wherein the near-end time domain includes a time domain range of a first time length after the current moment, and the one or more traffic objects include the one or more traffic participants, lane boundaries, and red light stop lines. Hard safety constraint information is used to indicate the real-time safety boundaries of one or more traffic objects; based on the predicted motion trajectories of one or more traffic participants, the interaction potential field information between the vehicle and one or more traffic objects in the far time domain is determined, wherein the time domain range of the far time domain includes the time domain range of a second time length after the near time domain; based on the hard safety constraint information of one or more traffic objects and the interaction potential field information between the vehicle and one or more traffic objects, the control command of the vehicle is determined, the control command including the front wheel steering angle and acceleration of the vehicle.

[0006] This scheme clearly distinguishes between hard safety constraints in the near-term time domain and flexible interactive potential field guidance in the far-term time domain. While ensuring strict adherence to driving safety and traffic rules (such as collision avoidance, yielding to pedestrians, stopping at red lights, and lane keeping) in the short term, it provides an optimization direction for the vehicle's driving tendencies over a longer time scale. This achieves a balance between multiple objectives such as safety, comfort, and efficiency, effectively improving the overall performance of the autonomous driving system in complex and ever-changing urban traffic scenarios.

[0007] In conjunction with the first aspect, in a possible implementation of the first aspect, obtaining the predicted motion trajectory of one or more traffic participants around the vehicle includes: obtaining environmental information and a global static reference path, wherein the environmental information includes the historical trajectory of the one or more participants, and the global static reference path serves as a reference path for trajectory tracking; and determining the predicted motion trajectory of the one or more traffic participants based on the environmental information, a zero-mean Gaussian function, and a radial basis function as a covariance function.

[0008] This scheme uses the Gaussian process regression (GPR) model to predict the short-term trajectories of traffic participants. Based on limited historical observation data, it can better characterize the continuous change pattern of traffic participants' movements and has good computational efficiency.

[0009] In conjunction with the first aspect, in a possible implementation of the first aspect, determining the vehicle's control command based on the hard safety constraint information of the one or more traffic objects and the interaction potential field information between the vehicle and the one or more traffic objects includes: constructing a nonlinear vehicle dynamics model, wherein the input of the nonlinear vehicle dynamics model includes the vehicle's current motion state information, and the output of the nonlinear vehicle dynamics model includes the vehicle's future motion state information; constructing an Objective Function Point (OCP) based on the nonlinear vehicle dynamics model, the hard safety constraint information, and the interaction potential field information; and determining the vehicle's control command by solving the OCP, wherein the objective function of the OCP includes the hard safety constraint information and the interaction potential field information.

[0010] This scheme integrates vehicle dynamics, near-term hard constraints, and far-term interactive potential fields into a single operational programming (OCP), and solves it online using the rolling optimization framework of dynamic programming (MPC). Ultimately, this scheme can output vehicle control commands in real time based on the global static reference path, conforming to vehicle dynamics, satisfying hard safety constraints, and progressing towards the optimization goal (achieving long-term reasonable behavior).

[0011] In conjunction with the first aspect, in a possible implementation of the first aspect, the hard safety constraint information of the one or more traffic objects includes the control barrier function (CBF) corresponding to each of the one or more traffic objects.

[0012] In conjunction with the first aspect, in a possible implementation of the first aspect, the CBF corresponding to the one or more traffic participants is determined according to the elliptical envelope parameters, the CBF corresponding to the lane boundary is used to limit the front and rear envelope centers of the vehicle from crossing the lane boundary, and the CBF corresponding to the red light stop line is used to limit the distance between the vehicle and the red light stop line to be greater than or equal to a distance threshold.

[0013] In conjunction with the first aspect, in possible implementations of the first aspect, the interactive potential field information includes the interactive potential field between the vehicle and the one or more traffic participants, strong repulsion trend information indicating that the vehicle cannot cross the lane boundary, weak repulsion trend information indicating that the vehicle crosses the lane boundary, and advance deceleration and stopping guidance information for the vehicle at the red light stop line.

[0014] This scheme uses artificial potential field (APF) technology in the remote time domain to abstract various traffic elements (such as background vehicles, lane lines, and stop lines at red lights) into "potential fields" with clear physical or semantic meanings. These potential fields are not mandatory constraints, but rather flexibly guide vehicles to form smooth and reasonable future motion control by optimizing objectives.

[0015] Secondly, an integrated decision-making and control device for generating control commands for autonomous vehicles is provided. The device includes: an acquisition module for acquiring environmental information and a global static reference path, the environmental information including the historical trajectories of one or more participants, and the global static reference path serving as a reference path for trajectory tracking; a processing module for determining the predicted motion trajectories of one or more traffic participants around the vehicle based on the environmental information, a zero-mean Gaussian function, and a radial basis function as a covariance function, wherein the predicted motion trajectories include the planar positions and states of the one or more traffic participants at multiple discrete moments after the current moment; and determining, based on the predicted motion trajectories of the one or more traffic participants, hard safety constraint information of one or more traffic objects in the near-terminal time domain, wherein the near-terminal time domain includes the time range of the current moment. The time domain range of the first time length following the near time domain includes the one or more traffic objects, which include the one or more traffic participants, lane boundaries, and red light stop lines. The hard safety constraint information is used to indicate the real-time safety boundaries of the one or more traffic objects. Based on the predicted motion trajectory of the one or more traffic participants, the interaction potential field information between the vehicle and the one or more traffic objects in the far time domain is determined, wherein the time domain range of the far time domain includes the time domain range of the second time length following the near time domain. Based on the hard safety constraint information of the one or more traffic objects and the interaction potential field information between the vehicle and the one or more traffic objects, the control command of the vehicle is determined, which includes the front wheel steering angle and acceleration of the vehicle.

[0016] Thirdly, an electronic device is provided. The electronic device includes one or more processors; one or more memories; the one or more memories storing one or more computer programs, the one or more computer programs including instructions that, when executed by the one or more processors, cause a method as described in the first aspect or any possible implementation thereof to be performed.

[0017] Fourthly, a computer-readable storage medium is provided. This computer-readable storage medium stores computer instructions that, when executed on a computer, cause a method as described in the first aspect or any possible implementation thereof to be performed.

[0018] Fifthly, a computer program product is provided. When the computer program product is run on a computer, it causes the computer to perform the methods as described in the first aspect or any possible implementation thereof. Attached Figure Description

[0019] Figure 1 This is a schematic flowchart of a method for generating control commands for an autonomous vehicle that integrates decision-making and control, provided in an embodiment of this application.

[0020] Figure 2 This is a flowchart of a method for generating control commands for an automated driving vehicle that integrates decision-making and control, provided in an embodiment of this application.

[0021] Figure 3 This is a schematic diagram of a dual-time-domain secure boot collaboration strategy provided in an embodiment of this application;

[0022] Figure 4 This is a structural schematic diagram of a device provided in an embodiment of this application;

[0023] Figure 5 This is a structural schematic diagram of a system on a chip (SoC) provided in an embodiment of this application. Detailed Implementation

[0024] The technical solution of this application is described below with reference to the accompanying drawings.

[0025] As mentioned in the background section, in scenarios with complex interactions and high uncertainty, how to achieve coordinated optimization of decision-making and control remains a key challenge in current autonomous driving research.

[0026] For example, autonomous driving systems in highly interactive urban scenarios face two main challenges. First, traditional layered architectures design and optimize behavioral decisions and underlying control in separate stages, which can easily lead to decision-making and control biases, delayed responses to sudden interactions, and difficulty in achieving a stable and consistent trade-off between safety, traffic efficiency, and passenger comfort. Second, while end-to-end data-driven integrated methods can reduce module coupling and reliance on manual rules, they generally suffer from insufficient interpretability, difficulty in explicitly modeling and formally verifying traffic rules and key safety constraints, and uncertainty in generalization capabilities for long-tail scenarios, thus limiting their reliable application in complex, highly interactive urban environments.

[0027] To address the aforementioned issues, this application proposes a method 100 for generating control commands for autonomous vehicles that integrates decision-making and control. Figure 1 A schematic flowchart of an integrated decision-making and control method for generating control commands for autonomous vehicles, as provided in an embodiment of this application, is shown. Figure 1As shown, method 100 includes steps S110 to S140. In method 100, by dividing the prediction time domain of MPC into a near-end time domain and a far-end time domain, and integrating hard safety constraints and interactive potential fields, the decision-making task and control task are uniformly modeled as an OCP, and solved online under the MPC framework. Under the premise of ensuring driving safety and rule compliance, the collaborative optimization generation of vehicle control commands is realized.

[0028] Step S110: Obtain the predicted motion trajectories of one or more traffic participants around the vehicle.

[0029] Specifically, the predicted trajectory includes the planar position and state of one or more traffic participants at multiple discrete moments after the current moment. For example, in embodiments of this application, traffic participants can refer to other dynamic objects around the vehicle that may affect its movement, such as other vehicles, pedestrians, or non-motorized vehicles. As another example, in embodiments of this application, the planar position includes, for example, coordinate values ​​in a two-dimensional coordinate system, and the state can include kinematic parameters such as velocity, acceleration, and heading angle.

[0030] Optionally, to address the poor adaptability of existing trajectory prediction models in complex interactive scenarios, embodiments of this application may also use a GPR model for trajectory prediction of traffic participants. Specifically, obtaining the predicted motion trajectories of one or more traffic participants around the vehicle includes: acquiring environmental information and a global static reference path, wherein the environmental information includes the historical trajectories of the one or more participants, and the global static reference path serves as a reference path for trajectory tracking; and determining the predicted motion trajectory of the one or more traffic participants based on the environmental information, a zero-mean Gaussian function, and a radial basis function as a covariance function.

[0031] It should be understood that, in the embodiments of this application, GPR is able to learn the movement patterns of traffic participants through historical state sequences. Radial basis functions are used to describe the correlation of states in the time dimension, and they can output a mean trajectory aligned with the MPC prediction time domain as the prediction result.

[0032] Step S120: Based on the predicted motion trajectory of one or more traffic participants, determine the hard safety constraint information of one or more traffic objects in the near time domain.

[0033] Specifically, the near-end time domain includes a time range of a first time length after the current moment, such as 0 to 2 seconds after the current moment. The one or more traffic objects include the one or more traffic participants, lane boundaries, and red light stop lines. The hard safety constraint information is used to indicate the real-time safety boundaries of the one or more traffic objects. It should be understood that, in the embodiments of this application, hard safety constraints refer to conditions that must be met to ensure that vehicles do not collide or violate traffic rules in the short term. In the embodiments of this application, the near-end time domain focuses on immediate safety, therefore employing a mandatory constraint form.

[0034] Optionally, in embodiments of this application, the hard safety constraint information of the one or more traffic objects includes the CBF corresponding to each of the one or more traffic objects. For example, in embodiments of this application, the CBF corresponding to the one or more traffic participants is determined based on elliptical envelope parameters. The CBF corresponding to the lane boundary is used to restrict the front and rear envelope centers of the vehicle from crossing the lane boundary, and the CBF corresponding to the red light stop line is used to restrict the distance between the vehicle and the red light stop line of the traffic light to be greater than or equal to a distance threshold. The following embodiments will describe the CBF corresponding to each traffic object.

[0035] Step S130: Based on the predicted motion trajectory of the one or more traffic participants, determine the interaction potential field information between the vehicle and the one or more traffic objects in the remote time domain.

[0036] Specifically, the time domain range of the far-end time domain includes a second time length following the near-end time domain, such as 2 to 5 seconds after the current moment. In the embodiments of this application, the far-end time domain focuses on long-term behavior guidance, and therefore a flexible cost function can be used instead of a hard constraint.

[0037] Optionally, in embodiments of this application, the interactive potential field information includes the interactive potential field between the vehicle and one or more traffic participants, strong repulsion trend information indicating that the vehicle cannot cross the lane boundary, weak repulsion trend information indicating that the vehicle crosses the lane boundary, and early deceleration and stopping guidance information for the vehicle at the red light stop line. In embodiments of this application, the interactive potential field (such as a distance-based repulsion potential field) can be used to guide the vehicle to maintain a safe following distance from other traffic participants. The lane boundary potential field can be used to impose a slight penalty on behaviors that approach the boundary or frequently hug the edge when crossing is possible, improving trajectory smoothness and driving comfort. The red light stop line potential field can be designed to consist of three distance terms from the stop line and the boundaries of the left and right stop areas, used to form a deceleration-stop trend in advance in the far-end time domain. The following embodiments will introduce the interactive potential field information corresponding to each traffic object.

[0038] Step S140: Determine the control command for the vehicle based on the hard safety constraint information of the one or more traffic objects and the interactive potential field information between the vehicle and the one or more traffic objects.

[0039] Specifically, in the embodiments of this application, step S140 can construct a nonlinear vehicle dynamics model, wherein the input of the nonlinear vehicle dynamics model includes the motion state information of the vehicle at the current moment, and the output of the nonlinear vehicle dynamics model includes the motion state information of the vehicle at future moments. Then, based on the nonlinear vehicle dynamics model, the hard safety constraint information, and the interactive potential field information, an Objective Function (OCP) is constructed, wherein the objective function of the OCP includes the hard safety constraint information and the interactive potential field information. Finally, the control command for the vehicle is determined by solving the OCP. In the embodiments of this application, the control command may include the front wheel steering angle and acceleration of the vehicle.

[0040] It should be understood that the inputs to a nonlinear vehicle dynamics model can include the physical relationship between vehicle states (such as position, velocity, and heading angle) and control inputs (such as front wheel steering angle and acceleration). The objective function can include the CBF inequality representing hard safety constraints and the APF term representing behavioral guidance, and can also include trajectory tracking error, energy consumption, and driving comfort constraints, thus allowing for rolling optimization within the MPC framework. For example, in the MPC solution process, the combination of the target path and the vehicle's current and future states can serve as the path tracking component of the OCP objective function; CBF constraints and APF guidance can serve as the traffic rule component of the OCP objective function; and the OCP objective function can also include comfort and energy consumption.

[0041] Method 100 achieves dual-time-domain partitioning and integrates CBF hard constraints and APF guidance, unifying safety rules and behavioral tendencies into the OCP model. This enables integrated optimization of decision-making and control, ensuring real-time security and behavioral rationality in complex dynamic environments.

[0042] The following describes a detailed implementation of method 100. Figure 2 A flowchart of a method for generating control commands for an autonomous vehicle that integrates decision-making and control, as provided in an embodiment of this application, is shown.

[0043] Among them, such as Figure 2As shown, the embodiments of this application first, based on step S110, after acquiring real-time environmental information and a global reference path, estimate the short-term motion trajectories of surrounding traffic participants such as vehicles and pedestrians based on GPR trajectory prediction technology. Here, the global reference path is the reference path for the vehicle's trajectory tracking. The prediction result will be aligned with the time domain length of MPC to ensure that the prediction information can be directly used for subsequent optimization calculations.

[0044] After that, as Figure 2 As shown, based on the dual-time-domain safety guidance coordination strategy corresponding to steps S120 and S130, traffic rules are modeled using CBF and APF to construct hard constraints and cost functions. Specifically, after completing trajectory prediction, the system executes this dual-time-domain safety guidance coordination strategy. This strategy divides the prediction period of MPC into two parts: the near-terminal time domain and the far-terminal time domain.

[0045] In the near-terminal time domain, the system uses CBF to transform traffic safety rules into hard constraints, including collision avoidance constraints for dynamic obstacles (such as vehicles and pedestrians), boundary crossing prevention constraints for lane boundaries, and stopping constraints for red light stop lines. These constraints directly limit the feasible range of vehicle states and control commands in the form of mathematical inequalities, ensuring short-term driving safety from a mechanistic perspective. Figure 3 This diagram illustrates a dual-time-domain secure boot collaboration strategy provided in an embodiment of this application. Figure 3 As shown, the near-time domain CBF can include pedestrian CBF, background vehicle CBF, traffic light CBF, and non-crossable solid line CBF. The following embodiments will provide a detailed description of the CBF of each traffic object.

[0046] In the remote time domain, the system constructs a cost function using the Advanced Driver Function (APF) to transform traffic rules such as lane keeping and following distance into a physically meaningful potential energy field. This function guides the vehicle to form a smooth and reasonable long-term driving trajectory in the form of an optimization objective, rather than imposing mandatory constraints. Figure 3 As shown, the far-time domain APF can include pedestrian APF, background vehicle APF, traffic light APF, lane line APF, and time-to-collision (TCC) APF. The following examples will provide a detailed description of the APF for each traffic object.

[0047] Finally, corresponding to step S140, the OCP is constructed based on the vehicle dynamics model and solved in real time, outputting the final control command. Specifically, in this embodiment, the decision-making task that satisfies traffic rules and the control task of high-precision path tracking are uniformly modeled as an OCP, and the optimization problem is solved in real time within each control cycle, directly outputting control commands such as front wheel steering angle and acceleration. For example, as Figure 2As shown, in the MPC solution process, the vehicle state can be used as the input to the vehicle model, and the future state can be used as the output of the vehicle model. The target path, combined with the vehicle's current and future states, can be used as the path tracking part in the OCP objective function. CBF constraints and APF guidance can be used as the traffic rule part in the OCP objective function. The OCP objective function can also include comfort and energy consumption, etc. Figure 2 The integrated decision-making and control architecture shown achieves collaborative processing of decision-making and control tasks within the optimization framework while ensuring hard safety constraints.

[0048] The following is an introduction Figure 2 Detailed implementation examples of each step in the architecture.

[0049] First, in the embodiments of this application, corresponding to step S110, the trajectories of traffic participants around the vehicle can be predicted based on GPR. For example, the integrated decision-control method used in the embodiments of this application mainly relies on the prediction of the short-term movement behavior of surrounding traffic participants. Therefore, a trajectory prediction method based on GPR can be used to estimate the movement trends of vehicles and pedestrians over a future period. Furthermore, compared to prediction models that rely on fixed motion assumptions, GPR can better characterize the continuous change patterns of traffic participant movement based on limited historical observation data, while also possessing good computational efficiency, making it suitable for urban traffic environments with variable behavior and frequent interactions.

[0050] For example, for each traffic participant, this embodiment can calculate their trajectory within a prediction time range matching the MPC based on their recent state sequence. The prediction target is their planar position and state at several discrete future moments, and the prediction duration is consistent with the time domain length of the MPC. This embodiment can employ a zero-mean Gaussian process and select a radial basis function as the covariance function to describe the motion correlation of traffic participants in the time dimension. The mean trajectory of the prediction output can be used as an estimation of the future motion trend of traffic participants to construct hard constraints and cost functions based on CBF and APF. Furthermore, by aligning the short-term trajectory prediction with the time domain of the MPC, this embodiment can explicitly consider the expected motion of surrounding traffic participants in the optimization solution, thereby enhancing the system's forward-looking decision-making capability in complex dynamic scenarios.

[0051] Then, in the embodiments of this application, corresponding to steps S120 and S130, the following can be constructed: Figure 2 and Figure 3The illustrated dual-time-domain safety guidance coordination strategy is as follows. Specifically, to ensure strict safety of the autonomous driving system in the short term and reasonable behavioral guidance in the long term, while also taking into account the accuracy and adaptability of traffic rule expression, this application embodiment designs a dual-time-domain safety guidance coordination strategy. This strategy does not add an independent decision-making module, but directly performs a temporally structured design of safety constraints and guidance objectives within the prediction time domain of MPC.

[0052] Specifically, the embodiments of this application can predict the complete prediction period of MPC. Divided into near-terminal time domain With remote terminal time domain .

[0053] Among them, in the near-terminal time domain Within the framework, the strategy prioritizes ensuring immediate safety and compliance with regulations. Key traffic rules such as collision avoidance, pedestrian yielding, stop lines at red lights, and lane keeping are all translated into hard safety constraints based on CBF (Carrier Flow Factor). These constraints act directly on vehicle status and control commands in the form of inequalities, mathematically limiting the vehicle's range of actions to ensure short-term driving safety and compliance with traffic regulations.

[0054] In the distant time domain, the strategy focuses on characterizing the vehicle's driving tendencies and interaction habits over a longer time scale. To this end, this invention designs a cost function based on APF, which abstracts various traffic elements (such as lanes, background vehicles, etc.) into "potential fields" with clear physical or semantic meanings. This function does not act as a mandatory constraint on the vehicle, but rather flexibly guides the vehicle to form smooth and reasonable future motion control by optimizing the objective.

[0055] Corresponding to step S120, to ensure that the safety and traffic rules that must be followed in the urban road environment can be met within a limited time, this embodiment of the application can set the core task of the near-terminal time domain as transforming these mandatory requirements into feasible domain constraints of OCP, and through the rolling optimization and execution mechanism of MPC, ensure that the short-term behavior of vehicles always meets verifiable safety standards. Specifically, in the near-terminal time domain, this embodiment of the application constructs corresponding CBFs (i.e., the hard safety constraint information of step S120) for the four types of objects that have the most direct impact on safety in the urban scenario. The four types of objects include other vehicles, pedestrians, lane boundaries, and red light stop lines.

[0056] For example, for dynamic obstacles (such as other vehicles and pedestrians), in order to strike a balance between fully covering their safety area and controlling computational complexity, embodiments of this application may employ differentiated envelope representation strategies. For instance, for vehicles, a dual-centered safety envelope is used to cover the critical areas in front of and behind the vehicle, respectively; for external traffic participants, a parameterized elliptical envelope is used, modeling the safety area according to its actual shape and expanding it appropriately.

[0057] Suppose the state representation of the background vehicle at the τ-th step after the current time (i.e., the predicted trajectory of the background vehicle) is as follows: ,in , Let represent the number of background vehicles. Based on the normalized distance between the center of the vehicle and the elliptical envelope of the background vehicles, the CBF of the background vehicles at the τ-th step after the current time can be constructed as shown in equation (1):

[0058]

[0059] in and Let the lengths of the major and minor semi-axises of the elliptical safety envelope of the background vehicle be given. This represents the Hadamard product. The boundary value of the background vehicle CBF. and The front and rear circular safety envelopes of the vehicle are respectively the first and second circular safety envelopes after the current moment. The center position of the circle under the step is calculated using the following formula (2):

[0060]

[0061] in The radius of the front and rear centers of the vehicle.

[0062] This modeling approach can simultaneously cover potential collision risks at both the front and rear of the vehicle while maintaining the simplicity of the constraint form. The pedestrian's CBF constraint is isomorphic to that of the background vehicle, reflecting the difference in safety distance requirements between pedestrians and vehicles only in the elliptical envelope parameters and boundary values. This constructs the model for the current time step. The pedestrian CBF of the step is shown in the following formula (3):

[0063]

[0064] in and Let the lengths of the major and minor semi-axises of the pedestrian's elliptical safety envelope be given. The pedestrian's current moment after the first The relevant states of the step, ,in For the number of pedestrians, This represents the boundary value of the pedestrian CBF.

[0065] To ensure that vehicles remain within the lane during driving, this application embodiment establishes corresponding safety constraints for lane boundaries. Specifically, firstly, in a local coordinate system aligned with the lane tangent, the lateral positions of the vehicle's front and rear safety centers are calculated. Simultaneously, based on the current lane centerline position and lane width, the lateral coordinate values ​​corresponding to the left and right lane boundaries are determined. Based on this, the driving rule prohibiting crossing left and right lane lines is transformed into minimum lateral distance constraints that must be maintained between the vehicle's front and rear centers and the lane lines on both sides. In a local coordinate system aligned with the lane tangent, the lateral distances after the vehicle's current moment are... Lateral position of the center of the front and rear circular safety envelope of the step , Lateral position relative to the left and right lane lines and The calculation formulas are shown in equations (4) to (8) below:

[0066]

[0067]

[0068]

[0069]

[0070]

[0071] The coordinates of the lane centerline corresponding to the current position of the vehicle are: The angle between the tangent to the lane centerline and the global x-axis is... , This refers to the road width.

[0072] Furthermore, construct the th time after the current moment. The CBF (Cross-Border Line) that a pedestrian cannot cross is shown in the following formula (9):

[0073]

[0074] in , This is a solid line marking the left lane for vehicles. The solid line marking the right lane line for vehicles is set to 1 when the lane line is solid and 0 otherwise. The CBF boundary value represents the lane line that cannot be crossed. Through this modeling method, the lane line boundary that cannot be crossed is directly represented as a hard feasible domain constraint in the near-terminal time domain, strictly prohibiting the behavior of crossing the solid line.

[0075] For another example, in the case of traffic light control, the CBF constraint of the red light stop area is activated only when the traffic light is red and the vehicle is at the corresponding intersection during the current control cycle, and the CBF constraint of the stop area after the current time is constructed. The CBF of the red light stop line is shown in the following formula (10):

[0076]

[0077] in , , and These represent the longitudinal distance from the vehicle's coordinates to the stop line ahead and the lateral distance from the left and right lane lines, respectively. , These are the corresponding CBF boundary values. This constraint simultaneously restricts the vehicle's longitudinal position in front of the stop line and its lateral feasible region within the lane, preventing it from crossing the line or deviating from the lane during braking.

[0078] By combining the various CBFs described in formulas (1), (3), (9), and (10), the embodiments of this application can be used for the CBFs after the current time. A unified CBF was constructed and related constraints were designed, as shown in equations (11) and (12) below:

[0079]

[0080]

[0081] in For the current traffic environment in the first The relevant status after the step, The first time after the current time of the vehicle The relevant states of the step, It is a column vector of all zeros. For the whole Column vector, It is a column vector of all 1s. This is a column vector of CBF trend coefficients, where each component is in... between, , , , The dimensions are all with The same applies. All the above-mentioned CBF-based constraints collectively constitute the safe and feasible domain in the near-terminal time domain, and serve as a partial constraint set for OCP.

[0082] Corresponding to step S130, the core objective of the remote terminal time domain is to provide clear and reasonable driving trend guidance for the vehicle over a longer time span, thereby improving the smoothness of the driving trajectory and avoiding potential interaction risks in advance. To achieve this objective, the embodiments of this application can uniformly design the interaction behavior preferences in the remote terminal time domain as a cost function based on APF, and use it as an additional term of the OCP objective function in that time period for superposition calculation. The OCP objective function is shown in the following formula (13):

[0083]

[0084] in and These represent the nth time after the current time. The interaction potential field between the bicycle and background vehicles and pedestrians during walking; Indicates the time after the current moment for the vehicle. A strong repulsive tendency to not cross lane lines while walking; Indicates the time after the current moment for the vehicle. A weak repulsive tendency to cross lane lines while walking; Indicates the time after the current moment for the vehicle. Pre-stop guidance in areas where red lights are in effect; Indicates the time after the current moment for the vehicle. Risks associated with TTC-based timing.

[0085] For example, for dynamic obstacles (such as surrounding vehicles and pedestrians), embodiments of this application can construct the APF using a geometrically simplified model of the vehicle's double-centered envelope and the target's elliptical envelope. The vehicle uses a front and rear double-centered envelope to simultaneously cover the two safety-critical areas of the front and rear; while for other traffic participants, a parameterized elliptical envelope is used, thereby achieving conservative and safe coverage of the target shape with lower computational cost. Based on this geometric representation, the current time after the [missing information]... The cost function of the interaction potential field of the step is shown in the following equation (14):

[0086]

[0087] in Control the overall strength, For the first The background car after the current moment The coordinates of the step. Given the background number of vehicles, the cost function is essentially a distance-inverse penalty based on elliptic metrics: the cost function increases rapidly when the center of the vehicle approaches the target elliptical region, thus creating a tendency to detour or maintain distance at the far end.

[0088] For another example, the interaction potential cost function between the vehicle and the pedestrian and the background vehicle are isomorphic, as shown in the following equation (15):

[0089]

[0090] in For the first The pedestrian is the first person after the current moment. The coordinates of the step. and Let be the axis length parameter of the ellipse envelope. For the number of people.

[0091] For another example, regarding lane boundaries, this application embodiment can distinguish between two types—prohibited crossing and permitted crossing—based on whether vehicles are allowed to cross them, and model them using two different potential field functions: strong repulsion and weak repulsion, respectively. For instance, for lane lines that are prohibited from crossing (such as solid lines), their potential field function is constructed using an inverse distance ratio, causing vehicles to experience a rapidly increasing repulsive force when approaching. The cost function for the uncrossable lane line of the step is shown in equations (16) and (17):

[0092]

[0093]

[0094] in The correlation coefficient, and This is a solid line judgment indicator. It is set to 1 when the corresponding lane line is solid, and 0 otherwise. When the center of the vehicle approaches the lane line, the cost increases rapidly, gradually forming a trend of moving away from the boundary at the far end.

[0095] For another example, for crossable lane lines, the potential cost function adopts a weaker piecewise linear form, where the cost function is the first step after the current time step. The cost function of the potential field of the traversable lane is shown in equations (18) and (19):

[0096]

[0097]

[0098] in As an intensity parameter, the role of this potential cost function is not to prohibit crossing, but to impose a slight penalty on behaviors that are close to the boundary or frequently brush against the edge when crossing is possible, thereby improving trajectory smoothness and driving comfort.

[0099] For another example, for a red light, the embodiments of this application can model the red light stopping region as a potential field cost function, as shown in the following equation (20):

[0100]

[0101] in and As an intensity parameter, the potential cost function consists of three distance terms from the stop line and the boundaries of the left and right stop regions, which are used to form a deceleration-stop trend in advance in the time domain of the far end.

[0102] Optionally, to further enhance the foresight regarding potential rear-end collision risks, embodiments of this application may also introduce a potential field cost function based on TTC, as shown in equations (21) and (22) below:

[0103]

[0104]

[0105] in For strength parameters, For shape parameters, For time threshold, ( For the train that triggered TTC, the train after the current time... The coordinates of the step. For its speed, ( For the vehicle after the current moment The coordinates of the step. Its speed. The TTC trigger flag is set to 1 when the vehicle detects a vehicle ahead, and 0 otherwise. The potential cost function increases rapidly in an exponential manner with the accumulation of risk, raising the risk cost in advance before the near-end hard constraint is triggered, prompting the vehicle to adopt a more conservative driving strategy.

[0106] Furthermore, through the aforementioned potential field cost function design, the key traffic semantics are uniformly incorporated into the OCP in the form of a cost function in the far-end time domain. Based on the immediate safety and compliance with traffic regulations guaranteed by the near-end hard constraints, driving trends are further guided.

[0107] Finally, under urban driving conditions, vehicles are often in a state of frequent start-stop, acceleration, deceleration, and steering. In this situation, rapid changes in longitudinal velocity significantly affect the vehicle's lateral motion response. Using a lateral dynamics model based on the assumption of constant longitudinal velocity will struggle to accurately characterize the vehicle's actual dynamics; while using only a kinematic model will neglect key physical processes such as tire lateral forces and vehicle yaw dynamics. Neither of these models can fully and accurately describe the real-time motion behavior of vehicles in complex urban traffic environments.

[0108] For the above considerations, corresponding to step S140, this embodiment of the application adopts a numerically stable nonlinear vehicle dynamics model as the model basis for MPC. This model can more realistically reflect the relationship between the forces and motion of the vehicle, and is used to describe the dynamic behavior of the vehicle and predict its future state. The vehicle state is defined as follows: ,in and Let x and y be the vehicle's x and y coordinates. The vehicle's heading angle. and Let x be the lateral and longitudinal velocities of the vehicle. For the vehicle's yaw rate, the control input is defined as follows: ,in For the vehicle's acceleration, This refers to the front wheel steering angle. The continuous-time dynamic model is based on the sampling period. The discretization is performed using the backward Euler method, and its discretization form is shown in equation (23) below:

[0109]

[0110] The specific expansion form of the discrete model is shown in equation (24) below:

[0111]

[0112] in Indicates the mass of the vehicle. and These are the distances from the center of gravity to the front and rear axles, respectively. and These are the lateral stiffnesses of the front and rear wheels, respectively. It refers to the vehicle's lateral stiffness.

[0113] Furthermore, based on the aforementioned vehicle dynamics model, the embodiments of this application can construct a finite-time nonlinear OCP. Specifically, within each control cycle, the system solution length is... The future control sequence is determined, and the first control input obtained is executed in a rolling manner to form a closed-loop control.

[0114] like Figure 2 As described in the partial description, in the MPC solution process, the target path combined with the vehicle's current and future states can be used as the path tracking part of the OCP objective function, while CBF constraints and APF guidance can be used as the traffic rule part of the OCP objective function. The OCP objective function can also include comfort and energy consumption, etc.

[0115] For example, the objective function of OCP is composed of trajectory tracking error, control input penalty, control smoothness penalty, and APF cost in the far-end time domain. Among these, the potential field cost function... Only in the remote terminal time domain The cumulative sum is used to shape future behavioral trends. In the near-terminal time domain, immediate safety and compliance with traffic regulations can be ensured through CBF-based safety constraints. The corresponding OCP objective function is shown in equation (25) below:

[0116]

[0117] in , and These are the positive semidefinite diagonal weight matrices for trajectory tracking, energy consumption, and driving comfort, respectively. For the corresponding reference trajectory, , and , For relevant upper and lower limit constraints, It is a column vector of all zeros. For the whole Column vector, It is a column vector of all 1s. This is the column vector of CBF trend coefficients. , , , The dimensions are all with The same applies. By solving the OCP, the final control quantity can be obtained. .

[0118] The technical solutions described above clearly distinguish between hard safety constraints in the near-terminal time domain and flexible interactive potential field guidance in the far-terminal time domain. While ensuring that driving safety and traffic rules (such as collision avoidance, yielding to pedestrians, stopping at red lights, and lane keeping) are strictly followed in the short term, they provide an optimized direction for the vehicle's driving tendency over a longer time scale. This achieves a balance between multiple objectives such as safety, comfort, and efficiency, effectively improving the overall performance of the autonomous driving system in complex and ever-changing urban traffic scenarios.

[0119] Furthermore, using the GPR model for short-term trajectory prediction of traffic participants can effectively characterize the continuous changes in their movements based on limited historical observation data, while also demonstrating good computational efficiency. In addition, vehicle dynamics characteristics, hard constraints in the near-end time domain, and the interactive potential field in the far-end time domain are all incorporated into an OCP (Optimal Coding Process), and solved online using the rolling optimization framework of MPC (Multi-Process Optimization). Ultimately, this scheme can output vehicle control commands in real time based on the global static reference path that conform to vehicle dynamics characteristics, satisfy hard safety constraints, and evolve towards the optimization goal (achieving long-term rational behavior).

[0120] This application also provides a computer program product that, when run on an electronic device, causes the electronic device to execute the technical solutions described in the above embodiments. Its implementation principle and technical effects are similar to those of the related embodiments described above, and will not be repeated here.

[0121] This application also provides a readable storage medium containing instructions that, when executed by an electronic device, cause the electronic device to perform the technical solutions described in the above embodiments. The implementation principle and technical effects are similar and will not be repeated here.

[0122] This application also provides a chip for executing instructions. When the chip is running, it executes the technical solutions described in the above embodiments. Its implementation principle and technical effects are similar and will not be repeated here.

[0123] The hardware module of this application is described below, which can be used to implement the aforementioned method 100.

[0124] Now for reference Figure 4 The diagram shows a block diagram of a device 400 according to one embodiment of this application. Device 400 may include one or more processors 401 coupled to a controller hub 403. In at least one embodiment, the controller hub 403 communicates with the processor 401 via a multi-branch bus such as a front side bus (FSB), a point-to-point interface such as a quickpath interconnect (QPI), or a similar connection 410. The processor 401 executes instructions controlling general types of data processing operations. In one embodiment, the controller hub 403 includes, but is not limited to, a graphics memory controller hub (GMCH) (not shown) and an input / output hub (IOH) (which may be on a separate chip) (not shown), wherein the GMCH includes memory and a graphics controller and is coupled to the IOH.

[0125] Device 400 may also include a coprocessor 402 and a memory 404 coupled to a controller hub 403. Alternatively, one or both of the memory and the GMCH may be integrated within the processor, with memory 404 and coprocessor 402 directly coupled to processor 401 and controller hub 403, which resides on a single chip with the IOH. Memory 404 may be, for example, dynamic random access memory (DRAM), phase change memory (PCM), or a combination of both. In one embodiment, coprocessor 402 is a dedicated processor, such as, for example, a high-throughput many integrated core (MIC) processor, a network or communication processor, a compression engine, a graphics processor, a general-purpose computing on GPU (GPGPU), or an embedded processor, etc. Optional properties of coprocessor 402 are indicated by dashed lines. Figure 4 middle.

[0126] Memory 404, as a computer-readable storage medium, may include one or more tangible, non-transitory computer-readable media for storing data and / or instructions. For example, memory 404 may include any suitable non-volatile memory such as flash memory and / or any suitable non-volatile storage device such as one or more hard-disk drives (HDD(s)), one or more compact disc (CD) drives, and / or one or more digital versatile disc (DVD) drives.

[0127] In one embodiment, device 400 may further include a network interface controller (NIC) 406. NIC 406 may include a transceiver for providing a radio interface to device 400, thereby enabling communication with any other suitable device, such as a front-end module, antenna, etc. In various embodiments, NIC 406 may be integrated with other components of device 400. NIC 406 can implement the functionality of the communication unit in the above embodiments.

[0128] Device 400 may further include input / output (I / O) device 405. I / O 405 may include: a user interface designed to enable a user to interact with device 400; a peripheral component interface designed to enable peripheral components to also interact with device 400; and / or sensors designed to determine environmental conditions and / or location information related to device 400.

[0129] It is worth noting that, Figure 4 This is merely an example. That is, although... Figure 4 The diagram shows that device 400 includes multiple devices such as processor 401, controller hub 403, and memory 404. However, in actual applications, devices using the methods of this application may include only a portion of the devices in device 400. For example, it may include only processor 401 and NIC 406. Figure 4 The properties of the optional devices are shown in dashed lines. According to some embodiments of this application, the memory 404, which is a computer-readable storage medium, stores instructions that, when executed on a computer, cause the device 400 to perform the methods according to the above embodiments. Specific details can be found in the methods of the above embodiments, and will not be repeated here.

[0130] Now for reference Figure 5 The diagram shown is a block diagram of a SoC 500 according to an embodiment of this application. Figure 5 In the diagram, similar components share the same reference numerals. Additionally, dashed boxes are an optional feature for more advanced SoCs. Figure 5 In this SoC 500, the following are included: an interconnect unit 550 coupled to an application processor 510; a system proxy unit 580; a bus controller unit 590; an integrated memory controller unit 540; a group or one or more coprocessors 520, which may include integrated graphics logic, an image processor, an audio processor, and a video processor; a static random access memory (SRAM) unit 530; and a direct memory access (DMA) unit 560. In one embodiment, the coprocessor 520 includes a dedicated processor, such as, for example, a network or communication processor, a compression engine, a GPGPU, a high-throughput MIC processor, or an embedded processor.

[0131] The static random-access memory (SRAM) unit 530 may include one or more computer-readable media for storing data and / or instructions. The computer-readable storage medium may store instructions, specifically, temporary and permanent copies of those instructions. These instructions may include, when executed by at least one unit in the processor, causing the SoC 500 to perform the attention training method according to the above embodiments, as detailed in the methods described above, which will not be repeated here.

[0132] Various embodiments of the mechanisms disclosed in this application can be implemented in hardware, software, firmware, or combinations of these implementation methods. Embodiments of this application can be implemented as computer programs or program code executable on a programmable system, the programmable system including at least one processor, a storage system (including volatile and non-volatile memory and / or storage elements), at least one input device, and at least one output device.

[0133] Program code can be applied to input instructions to execute the functions described in this application and generate output information. The output information can be applied to one or more output devices in a known manner. For the purposes of this application, the processing system includes any system having a processor such as, for example, a digital signal processor (DSP), a microcontroller, an application-specific integrated circuit (ASIC), or a microprocessor.

[0134] The program code can be implemented using a high-level procedural language or an object-oriented programming language to communicate with the processing system. Assembly language or machine language can also be used when needed. In fact, the mechanisms described in this application are not limited to any particular programming language. In either case, the language can be a compiled language or an interpreted language.

[0135] In some cases, the disclosed embodiments may be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions carried or stored thereon on one or more temporary or non-temporary machine-readable (e.g., computer-readable) storage media, which may be read and executed by one or more processors. For example, the instructions may be distributed via a network or through other computer-readable media. Therefore, machine-readable media may include any mechanism for storing or transmitting information in a machine-readable (e.g., computer-readable) form, including but not limited to floppy disks, optical disks, CD-ROMs, compact disc read-only memory (CD-ROMs), magneto-optical disks, read-only memory (ROM), random-access memory (RAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic cards or optical cards, flash memory, or tangible machine-readable storage for transmitting information (e.g., carrier waves, infrared signals, digital signals, etc.) using the Internet in the form of electrical, optical, acoustic, or other forms of propagated signals. Therefore, machine-readable media includes any type of machine-readable medium suitable for storing or transmitting electronic instructions or information in a machine-readable (e.g., computer-readable) form.

[0136] In the accompanying drawings, some structural or methodological features may be shown in a specific arrangement and / or order. However, it should be understood that such a specific arrangement and / or order may not be necessary. Rather, in some embodiments, these features may be arranged in a manner and / or order different from that shown in the accompanying drawings. Furthermore, including structural or methodological features in a particular figure does not imply that such features are required in all embodiments, and in some embodiments, these features may be omitted or may be combined with other features.

[0137] It should be noted that all units / modules mentioned in the device embodiments of this application are logical units / modules. Physically, a logical unit / module can be a physical unit / module, a part of a physical unit / module, or a combination of multiple physical units / modules. The physical implementation of these logical units / modules themselves is not the most important factor; the combination of functions implemented by these logical units / modules is the key to solving the technical problems proposed in this application. Furthermore, to highlight the innovative aspects of this application, the above-described device embodiments of this application have not introduced units / modules that are not closely related to solving the technical problems proposed in this application. This does not mean that the above-described device embodiments do not contain other units / modules.

[0138] It should be noted that in the examples and description of this patent, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0139] Although this application has been illustrated and described with reference to certain preferred embodiments thereof, those skilled in the art should understand that various changes in form and detail may be made thereto without departing from the spirit and scope of this application.

Claims

1. A method for generating control commands for automated driving vehicles that integrates decision-making and control, characterized in that, include: Obtain the predicted motion trajectory of one or more traffic participants around the vehicle, wherein the predicted motion trajectory includes the planar position and state of the one or more traffic participants at multiple discrete time points after the current time. Based on the predicted motion trajectories of one or more traffic participants, hard safety constraint information of one or more traffic objects in the near-end time domain is determined, wherein the time domain range of the near-end time domain includes the time domain range of the first time length after the current moment, the one or more traffic objects include the one or more traffic participants, lane boundaries and red light stop lines, and the hard safety constraint information is used to indicate the real-time safety boundaries of the one or more traffic objects; Based on the predicted motion trajectories of one or more traffic participants, the interaction potential field information between the vehicle and the one or more traffic objects in the far time domain is determined, wherein the time domain range of the far time domain includes the time domain range of the second time length after the near time domain. Based on the hard safety constraint information of the one or more traffic objects and the interactive potential field information between the vehicle and the one or more traffic objects, the control command of the vehicle is determined, and the control command includes the front wheel steering angle and acceleration of the vehicle.

2. The method according to claim 1, characterized in that, The acquisition of the predicted motion trajectories of one or more traffic participants around the vehicle includes: Obtain environmental information and a global static reference path, wherein the environmental information includes the historical trajectories of one or more participants, and the global static reference path serves as a reference path for trajectory tracking; The predicted motion trajectories of one or more traffic participants are determined based on the environmental information, the zero-mean Gaussian function, and the radial basis function as the covariance function.

3. The method according to claim 1 or 2, characterized in that, Based on the hard safety constraint information of the one or more traffic objects and the interaction potential field information between the vehicle and the one or more traffic objects, the control command of the vehicle is determined, including: A nonlinear vehicle dynamics model is constructed, wherein the input of the nonlinear vehicle dynamics model includes the motion state information of the vehicle at the current moment, and the output of the nonlinear vehicle dynamics model includes the motion state information of the vehicle at future moments; Based on the nonlinear vehicle dynamics model, the hard safety constraint information, and the interactive potential field information, an optimal control problem (OCP) is constructed, wherein the objective function of the OCP includes the hard safety constraint information and the interactive potential field information. The control commands for the vehicle are determined by solving the OCP.

4. The method according to claim 1 or 2, characterized in that, The hard safety constraint information of the one or more traffic objects includes the control barrier function (CBF) corresponding to each of the one or more traffic objects.

5. The method according to claim 4, characterized in that, The CBF corresponding to one or more traffic participants is determined according to the elliptical envelope parameters. The CBF corresponding to the lane boundary is used to limit the front and rear envelope centers of the vehicle from crossing the lane boundary. The CBF corresponding to the red light stop line is used to limit the distance between the vehicle and the red light stop line to be greater than or equal to a distance threshold.

6. The method according to claim 1 or 2, characterized in that, The interactive potential field information includes the interactive potential field between the vehicle and one or more traffic participants, strong repulsion trend information indicating that the vehicle cannot cross the lane boundary, weak repulsion trend information indicating that the vehicle crosses the lane boundary, and early deceleration and stopping guidance information for the vehicle at the red light stop line.

7. A decision-making and control integrated control command generation device for automated driving vehicles, characterized in that, include: The acquisition module is used to acquire environmental information and a global static reference path. The environmental information includes the historical trajectories of one or more participants, and the global static reference path serves as a reference path for trajectory tracking. The processing module is used to determine the predicted motion trajectory of one or more traffic participants around the vehicle based on the environmental information, the zero-mean Gaussian function, and the radial basis function as the covariance function, wherein the predicted motion trajectory includes the planar position and state of the one or more traffic participants at multiple discrete time points after the current time. And based on the predicted motion trajectories of the one or more traffic participants, determine the hard safety constraint information of one or more traffic objects in the near time domain, wherein the time domain range of the near time domain includes the time domain range of the first time length after the current time, the one or more traffic objects include the one or more traffic participants, lane boundaries and red light stop lines, and the hard safety constraint information is used to indicate the real-time safety boundaries of the one or more traffic objects; And based on the predicted motion trajectories of the one or more traffic participants, determine the interaction potential field information between the vehicle and the one or more traffic objects in the far time domain, wherein the time domain range of the far time domain includes the time domain range of the second time length after the near time domain; And based on the hard safety constraint information of the one or more traffic objects and the interactive potential field information between the vehicle and the one or more traffic objects, the control command of the vehicle is determined, the control command including the front wheel steering angle and acceleration of the vehicle.

8. An electronic device, characterized in that, It includes one or more processors; one or more memories; said one or more memories storing one or more computer programs, said one or more computer programs including instructions that, when executed by said one or more processors, cause the method as described in any one of claims 1 to 6 to be performed.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed on a computer, cause the method as described in any one of claims 1 to 6 to be performed.

10. A computer program product, characterized in that, When the computer program product is run on a computer, it causes the computer to perform the method as described in any one of claims 1 to 6.

Images