Automatic driving track generation method and device, equipment and medium

Abstract

The invention discloses an automatic driving track generation method and device, equipment and a medium. The method comprises the steps that target state information of a vehicle is obtained, and the target state information comprises vehicle body state information, obstacle information, road structure information and traffic state information; the target state information is input into a trajectory planning model for trajectory generation, a plurality of candidate driving trajectories and trajectory evaluation results corresponding to the candidate driving trajectories are determined, and the trajectory planning model is constructed based on a deep neural network; and determining a target driving trajectory based on the plurality of candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory. According to the method, the automatic driving track of the vehicle can be automatically and accurately generated, emergencies in dynamic traffic are covered, the generalization ability during cross-scene migration is improved, the track generation flexibility is improved, and therefore the track precision and safety in complex scenes are guaranteed.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to an autonomous driving trajectory generation method, apparatus, device, and medium. Background Technology

[0002] With the rapid iteration of autonomous driving technology and the continuous emergence of complex traffic scenarios, the requirements for the accuracy and dynamic adaptability of vehicle trajectory generation have significantly increased. As a core means to achieve safe and efficient driving, trajectory generation directly affects the reliability and economy of autonomous driving systems.

[0003] Currently, traditional autonomous driving trajectory generation methods often rely on human experience rules or fixed scene data to determine the trajectory through preset path planning algorithms or offline optimized strategy parameters. However, traditional autonomous driving trajectory generation methods are difficult to cover unexpected situations in dynamic traffic, and models driven by fixed scene data have weak generalization ability and low flexibility when migrating across scenes, thus making it difficult to guarantee trajectory accuracy and safety in complex scenarios. Summary of the Invention

[0004] This invention provides an autonomous driving trajectory generation method, apparatus, device, and storage medium to achieve automatic and accurate generation of vehicle autonomous driving trajectories. It can cover unexpected situations in dynamic traffic, improve the generalization ability when migrating across scenarios, and enhance the flexibility of trajectory generation, thereby ensuring trajectory accuracy and safety in complex scenarios.

[0005] According to one aspect of the present invention, an autonomous driving trajectory generation method is provided, the method comprising:

[0006] Obtain the target status information of the vehicle;

[0007] The target state information is input into a pre-trained trajectory planning model to generate a trajectory, thereby determining multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory. The trajectory planning model is constructed based on a deep neural network.

[0008] The target driving trajectory is determined based on multiple candidate driving trajectories and the trajectory evaluation results corresponding to each candidate driving trajectory.

[0009] According to another aspect of the present invention, an autonomous driving trajectory generation apparatus is provided, the apparatus comprising:

[0010] The information acquisition module is used to acquire the target status information of the vehicle;

[0011] The candidate trajectory determination module is used to input the target state information into a pre-trained trajectory planning model to generate a trajectory, determine multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory, wherein the trajectory planning model is constructed based on a deep neural network;

[0012] The target trajectory determination module is used to determine the target driving trajectory based on multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the autonomous driving trajectory generation method according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the autonomous driving trajectory generation method according to any embodiment of the present invention.

[0018] The technical solution of this invention provides a complete decision-making basis for subsequent trajectory planning by acquiring the target state information of the vehicle. The target state information is input into a pre-trained trajectory planning model for trajectory generation. Based on the output of the trajectory planning model, multiple candidate driving trajectories and a trajectory evaluation result corresponding to each candidate driving trajectory are determined. The trajectory planning model is built on a deep neural network, which can dynamically adapt to complex environments, quantitatively evaluate the comprehensive performance of trajectories, provide a scientific basis for trajectory selection, and avoid the limitations of relying on manual rules. Based on the multiple candidate driving trajectories and the trajectory evaluation results corresponding to each candidate driving trajectory, a target driving trajectory is determined, which can adapt to complex traffic scenarios and improve the reliability of autonomous driving. This invention optimizes the generation and evaluation of autonomous driving trajectories by using a trajectory planning model to make trajectory decisions based on target state information. This greatly improves the adaptability and efficiency of trajectory generation in complex traffic scenarios, enhances the generalization ability during cross-scenario migration, and increases the flexibility of trajectory generation, thereby ensuring trajectory accuracy and safety in complex scenarios.

[0019] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of an autonomous driving trajectory generation method provided in Embodiment 1 of the present invention;

[0022] Figure 2 This is a flowchart of an autonomous driving trajectory generation method provided in Embodiment 2 of the present invention;

[0023] Figure 3 This is a schematic diagram of the structure of an autonomous driving trajectory generation device according to Embodiment 3 of the present invention;

[0024] Figure 4 This is a schematic diagram of the structure of an electronic device that implements the autonomous driving trajectory generation method of this invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first," "second," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] Example 1

[0028] Figure 1 This is a flowchart illustrating an autonomous driving trajectory generation method according to Embodiment 1 of the present invention. This embodiment is applicable to the generation of trajectories for autonomous vehicles. The method can be executed by an autonomous driving trajectory generation device, which can be implemented in hardware and / or software. This autonomous driving trajectory generation device can be configured in an electronic device. Figure 1 As shown, the method includes:

[0029] S110, Obtain the target status information of the vehicle.

[0030] Among them, target status information can refer to the comprehensive information on vehicle status and environment after preprocessing such as data cleaning and standardization.

[0031] Specifically, vehicle-mounted sensors (such as lidar, millimeter-wave radar, cameras, IMU inertial sensors, GPS positioning modules, etc.) can collect real-time data on the vehicle's own status (such as position, speed, acceleration, heading angle, yaw rate) and environmental information (such as the positions of surrounding vehicles / pedestrians / obstacles, road geometry, traffic signals, and weather conditions). This data can provide a comprehensive and accurate foundation for subsequent trajectory planning, helping to ensure that subsequent trajectory planning can make decisions under the latest environmental conditions and adapt to dynamic traffic scenarios.

[0032] For example, S110 may include: in response to the user's autonomous driving activation operation, acquiring the vehicle's original state information, wherein the original state information includes vehicle body state information, obstacle information, road structure information, and traffic state information; performing preprocessing operations on the original state information to obtain target state information, wherein the preprocessing operations include at least one of data cleaning and standardization.

[0033] Raw state information refers to the raw data directly collected by onboard sensors (such as LiDAR, cameras, etc.), which may include unprocessed vehicle state, environmental perception, and traffic dynamics information. Vehicle state information refers to the vehicle's own dynamic parameters. For example, vehicle state information may include the vehicle's position, speed, acceleration, heading angle, yaw rate, throttle / brake status, etc. Obstacle information refers to the state information of other vehicles, pedestrians, and static obstacles in the surrounding environment, such as their position, speed, trajectory, and category (e.g., pedestrian / truck). Road structure information refers to the geometric structure of the road where the vehicle is located (e.g., lane lines, curvature, slope), traffic signs (e.g., speed limits, yield signs), and road attributes (e.g., tunnels, toll booths). Traffic state information refers to the traffic flow density and traffic light status at the vehicle's location.

[0034] Specifically, the autonomous driving system can be activated by receiving explicit user commands through the in-vehicle human-machine interface (such as buttons on the central control screen, the voice command "start autonomous driving", or physical buttons). The system must verify the user's identity (e.g., fingerprint recognition, facial recognition) and driving permissions (e.g., whether they hold a valid driver's license) to ensure compliant activation. The vehicle can collect vehicle status information through inertial measurement units, wheel speed sensors, and steering wheel angle sensors; obstacle information about the surrounding environment through LiDAR and cameras; road structure information about the area where the vehicle is located through high-precision maps; and traffic status information about the surrounding area through the vehicle navigation module, thus obtaining the vehicle's raw state information. Preprocessing of this raw state information—for example, noise removal through data cleaning and data quality improvement through standardization—results in target state information that more accurately reflects the real-world environment. This provides precise input for trajectory planning models, facilitating efficient and safe decision-making in complex scenarios.

[0035] S120. Input the target state information into the pre-trained trajectory planning model to generate the trajectory, determine multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory. The trajectory planning model is constructed based on a deep neural network.

[0036] Here, the trajectory planning model can refer to an end-to-end decision-making system built on a deep neural network. The candidate driving trajectory can refer to the possible driving paths generated by the trajectory planning model. The trajectory evaluation result can refer to the comprehensive score output after multi-dimensional evaluation of the candidate driving trajectories, used to guide the final selection of the driving trajectory. The deep neural network can refer to a multi-layered network architecture containing convolutional layers, recurrent layers, fully connected layers, etc.

[0037] Specifically, target state information can be input into a pre-trained trajectory planning model. The trajectory planning model can generate multiple candidate driving trajectories that meet dynamic constraints (such as speed, acceleration, and curvature) based on the conditional variational autoencoder (CVAE) or generative adversarial network (GAN) in a deep neural network. By calculating the safety indicators (such as the minimum distance between the vehicle and obstacles), comfort indicators (such as evaluating lateral / longitudinal acceleration fluctuations), and efficiency indicators (such as travel time) corresponding to each candidate driving trajectory, and by weighting and summing the different indicators, a comprehensive score corresponding to each candidate driving trajectory can be obtained, i.e., the trajectory evaluation result. This allows for dynamic adaptation to complex environments, quantitative evaluation of the comprehensive performance of the trajectory, and provides a scientific basis for trajectory selection, avoiding the limitations of relying on manual rules.

[0038] S130. Based on multiple candidate driving trajectories and the trajectory evaluation results corresponding to each candidate driving trajectory, determine the target driving trajectory.

[0039] The target driving trajectory can refer to the optimal or most suitable driving path for the current scenario selected from multiple candidate driving trajectories, serving as the final execution basis for the autonomous driving system.

[0040] Specifically, the trajectory evaluation results corresponding to each candidate driving trajectory can be sorted, and the candidate driving trajectory with the highest comprehensive score can be used as the target driving trajectory based on the sorting results. This ensures the balance between vehicle driving safety, efficiency and comfort, improves the reliability of autonomous driving, and enhances the user experience.

[0041] In this embodiment, by acquiring the vehicle's target state information, a complete decision-making basis can be provided for subsequent trajectory planning. The target state information is input into a pre-trained trajectory planning model for trajectory generation. Based on the output of the trajectory planning model, multiple candidate driving trajectories and the corresponding trajectory evaluation results for each candidate driving trajectory are determined. The trajectory planning model is built on a deep neural network, which can dynamically adapt to complex environments, quantitatively evaluate the comprehensive performance of trajectories, provide a scientific basis for trajectory selection, and avoid the limitations of relying on manual rules. Based on multiple candidate driving trajectories and the corresponding trajectory evaluation results for each candidate driving trajectory, the target driving trajectory is determined, which can adapt to complex traffic scenarios and improve the reliability of autonomous driving. This invention optimizes autonomous driving trajectory generation and evaluation by using a trajectory planning model to make trajectory decisions based on target state information. This greatly improves the adaptability and efficiency of trajectory generation in complex traffic scenarios, enhances the generalization ability during cross-scenario migration, and increases the flexibility of trajectory generation, thereby ensuring trajectory accuracy and safety in complex scenarios.

[0042] Example 2

[0043] Figure 2 This is a flowchart of an autonomous driving trajectory generation method provided in Embodiment 2 of the present invention. The trajectory planning model includes a trajectory generation module and a trajectory evaluation module. Based on the above embodiments, this embodiment optimizes the step of "inputting the target state information into the pre-trained trajectory planning model to generate the trajectory, determining multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory". The explanations of the same or corresponding terms as in the above embodiments will not be repeated here.

[0044] See Figure 2 The other autonomous driving trajectory generation method provided in this embodiment specifically includes the following steps:

[0045] S210, Obtain the target status information of the vehicle.

[0046] S220. Input the target status information into the trajectory generation module to perform trajectory planning and determine multiple candidate driving trajectories corresponding to the vehicle.

[0047] Specifically, the trajectory generation module can be built based on the Actor network. The target state information can be input into the trajectory generation module. The trajectory generation module generates multiple candidate driving trajectories through latent space sampling, that is, introducing random noise into the hidden layer of the Actor network, and through reparameterization techniques. This can cover different driving environments and styles, adapt to complex scenarios, and improve decision robustness.

[0048] For example, the trajectory generation module includes: an encoding layer, an intermediate layer, and an output layer; S220 may include: inputting target state information into the encoding layer for feature extraction to obtain the target feature vector corresponding to the vehicle; inputting the target feature vector into the intermediate layer for temporal modeling strategy optimization to obtain multiple sets of initial trajectory parameters; inputting the multiple sets of initial trajectory parameters into the output layer for trajectory generation to obtain multiple candidate driving trajectories.

[0049] Here, the target feature vector can refer to a low-dimensional feature representation generated after nonlinear mapping of the target state information. The initial trajectory parameters can refer to the trajectory parameters generated by the intermediate layer through deterministic policy gradients or latent space sampling.

[0050] Specifically, the target state information is input into the encoding layer for data cleaning. The encoding layer extracts features from the cleaned target state information through convolutional and fully connected layers, and enhances the extracted features using activation functions to obtain the target feature vector. The intermediate layer captures the temporal series dependencies of the target feature vector output by the encoding layer and generates multiple sets of initial trajectory parameters based on deterministic policy gradients and latent space sampling. The output layer converts the multiple sets of initial trajectory parameters into specific trajectories through a parameterized trajectory model, obtaining multiple candidate driving trajectories. Through the collaborative mechanism of the encoding layer, intermediate layer, and output layer, feature extraction, temporal modeling, and trajectory generation can be decoupled, ensuring the efficiency, safety, and interpretability of trajectory generation. This facilitates independent module optimization and fault diagnosis, and improves the maintainability of the system.

[0051] For example, the encoding layer is specifically used to: perform multimodal spatiotemporal alignment on the target state data, and extract features from the aligned target state data to obtain a target feature vector, wherein the target feature vector contains spatial features, temporal features and semantic features corresponding to the target state data.

[0052] Spatial features can refer to characteristics describing the geometric relationship and positional distribution of vehicles and their environment in physical space, reflecting static spatial structure and dynamic positional relationships. Temporal features can refer to characteristics describing the dynamic changes in vehicle state and environment over time, reflecting temporal dependence and trend prediction capabilities. Semantic features can refer to characteristics describing the functional attributes, behavioral intentions, and rule constraints of environmental elements, reflecting the semantic understanding and rule compliance capabilities of traffic scenarios.

[0053] Specifically, the encoding layer can perform timestamp synchronization and coordinate system unification processing on the target state data to achieve multimodal spatiotemporal alignment of the target state data. This can eliminate time delay and spatial deviation of multiple sensors, ensuring that the state information truly reflects the vehicle environment in the spatiotemporal dimension. Spatial feature extraction, temporal feature extraction, and semantic feature extraction are then performed on the aligned target state data to obtain the target feature vector. This allows high-dimensional data to be compressed into low-dimensional target feature vectors, reducing subsequent computation and improving efficiency.

[0054] For example, the output layer is specifically used to: perform trajectory smoothing processing on multiple sets of initial trajectory parameters to obtain multiple initial driving trajectories, and perform risk assessment on the initial driving trajectories based on vehicle dynamics models and collision detection algorithms to obtain risk assessment results. In response to the risk assessment results satisfying the preset driving effect, the initial driving trajectory that satisfies the preset driving effect is determined as a candidate driving trajectory.

[0055] The initial driving trajectory can refer to a continuous driving path formed after smoothing the output trajectory, including dynamic parameters such as trajectory point sequences, velocity curves, and acceleration curves. The vehicle dynamics model can refer to a mathematical model describing the vehicle's kinematics and dynamic characteristics, used to verify whether the trajectory conforms to physical constraints. The collision detection algorithm can refer to an algorithm used to detect potential collisions between the vehicle trajectory and obstacles (such as other vehicles, pedestrians, and static objects). The risk assessment result can refer to a comprehensive score of the initial driving trajectory across multiple dimensions such as safety, comfort, and efficiency, used to quantify the trajectory's quality. The preset driving effect can refer to multi-dimensional driving effect thresholds or weights set according to scenario requirements and user preferences, used to filter candidate trajectories that meet the requirements.

[0056] Specifically, the output layer can input multiple sets of initial trajectory parameters output from the intermediate layer into the parameterized trajectory model. Through mathematical transformations, it generates multiple sequences of continuous trajectory points and performs trajectory smoothing on these sequences to obtain multiple initial driving trajectories. These initial driving trajectories are then assessed for risk using a vehicle dynamics model and collision detection algorithms. A risk assessment result is obtained for each initial driving trajectory. If the risk assessment result meets the preset driving effect, the initial driving trajectory that meets the preset driving effect is determined as a candidate driving trajectory. Through a rigorous trajectory processing and risk assessment process, the output layer ensures that the autonomous driving system outputs safe, comfortable, and efficient candidate driving trajectories in complex scenarios, providing core technical support for decision-making and execution in autonomous driving trajectory planning.

[0057] S230. For each candidate driving trajectory, input the target status information and the candidate driving trajectory into the trajectory evaluation module to determine the trajectory evaluation result corresponding to the candidate driving trajectory.

[0058] Specifically, the trajectory generation module can be built based on a Critic network. For each candidate driving trajectory, the target state information and the candidate driving trajectory are input into the trajectory evaluation module. The trajectory evaluation module can evaluate the candidate driving trajectory based on the target state information and multi-dimensional evaluation indicators (such as safety evaluation, efficiency evaluation, and comfort evaluation), and then perform a weighted sum of the evaluation results from different dimensions to obtain the trajectory evaluation result corresponding to the candidate driving trajectory. Quantifying trajectory risk through multi-dimensional evaluation indicators can provide an interpretable basis for decision-making, enhancing system transparency and user trust.

[0059] For example, the trajectory evaluation module includes an input layer, a processing layer, and an evaluation layer; S230 may include: inputting the target state information and candidate driving trajectory into the input layer for feature concatenation to obtain a joint feature vector; inputting the joint feature vector into the processing layer for feature extraction to obtain a trajectory evaluation vector; and inputting the trajectory evaluation vector into the evaluation layer for trajectory evaluation to obtain the trajectory evaluation result corresponding to the candidate driving trajectory.

[0060] The joint feature vector can be a high-dimensional composite vector formed by concatenating target state information and candidate driving trajectory features. It serves as the input basis for the trajectory evaluation module, containing comprehensive information on environmental state and trajectory dynamic attributes. The trajectory evaluation vector can be a high-order abstract vector formed by deep feature extraction from the joint feature vector, containing multi-dimensional evaluation features such as spatial, temporal, and semantic dimensions.

[0061] Specifically, target state information and candidate driving trajectories can be input into the input layer for spatiotemporal alignment, and the aligned data can be fused into a joint feature vector through vector concatenation. The processing layer uses convolutional neural networks, recurrent neural networks, and attention mechanisms to perform deep feature extraction on the joint feature vector, and performs feature dimensionality reduction and enhancement through fully connected layers and activation functions to obtain the trajectory evaluation vector. The trajectory evaluation vector is then input into the evaluation layer, which maps the trajectory evaluation vector to each evaluation dimension and uses weighted summation to generate the trajectory evaluation result corresponding to the candidate driving trajectory, thereby achieving automation and accuracy of trajectory evaluation and reducing human intervention errors.

[0062] S240. Based on multiple candidate driving trajectories and the trajectory evaluation results corresponding to each candidate driving trajectory, determine the target driving trajectory.

[0063] The technical solution of this embodiment inputs target state information into the trajectory generation module for trajectory planning, determining multiple candidate driving trajectories for the vehicle, thereby adapting to complex scenarios and improving decision-making reliability. For each candidate driving trajectory, the target state information and the candidate driving trajectory are input into the trajectory evaluation module to determine the trajectory evaluation result corresponding to the candidate driving trajectory, providing an interpretable basis for subsequent decisions. This invention, through the collaborative mechanism between the trajectory generation module and the trajectory evaluation module, achieves accurate generation and evaluation of autonomous driving trajectories, significantly improving the safety, efficiency, and adaptability of autonomous driving systems in complex scenarios.

[0064] Example 3

[0065] Figure 3 This is a schematic diagram of the structure of an autonomous driving trajectory generation device provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes: an information acquisition module 310, a candidate trajectory determination module 320, and a target trajectory determination module 330;

[0066] The information acquisition module is used to acquire the target status information of the vehicle.

[0067] The candidate trajectory determination module is used to input the target state information into a pre-trained trajectory planning model to generate a trajectory, determine multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory, wherein the trajectory planning model is constructed based on a deep neural network;

[0068] The target trajectory determination module is used to determine the target driving trajectory based on multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory.

[0069] In this embodiment, by acquiring the vehicle's target state information, a complete decision-making basis can be provided for subsequent trajectory planning. The target state information is input into a pre-trained trajectory planning model for trajectory generation. Based on the output of the trajectory planning model, multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory are determined. The trajectory planning model is built based on a deep neural network, which can dynamically adapt to complex environments, quantitatively evaluate the comprehensive performance of trajectories, provide a scientific basis for trajectory selection, and avoid the limitations of relying on manual rules. Based on the multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory, the target driving trajectory is determined, which can adapt to complex traffic scenarios and improve the reliability of autonomous driving. This invention optimizes autonomous driving trajectory generation and evaluation by using a trajectory planning model to perform trajectory decision-making and evaluation based on target state information. This greatly improves the adaptability and efficiency of trajectory generation in complex traffic scenarios, enhances the generalization ability during cross-scenario migration, and increases the flexibility of trajectory generation, thereby ensuring trajectory accuracy and safety in complex scenarios.

[0070] Optionally, the information acquisition module 310 is specifically used to: in response to the user's autonomous driving activation operation, acquire the vehicle's original state information, wherein the original state information includes vehicle body state information, obstacle information, road structure information, and traffic state information; perform preprocessing operations on the original state information to obtain target state information, wherein the preprocessing operations include at least one of data cleaning and standardization.

[0071] Optionally, the trajectory planning model includes a trajectory generation module and a trajectory evaluation module, and the candidate trajectory determination module 320 includes:

[0072] The trajectory generation unit is used to input the target state information into the trajectory generation module for trajectory planning and to determine multiple candidate driving trajectories corresponding to the vehicle.

[0073] The trajectory evaluation unit is used to input the target state information and the candidate driving trajectory into the trajectory evaluation module for each candidate driving trajectory, and determine the trajectory evaluation result corresponding to the candidate driving trajectory.

[0074] Optionally, the trajectory generation module includes an encoding layer, an intermediate layer, and an output layer; the trajectory generation unit is specifically used for: inputting the target state information into the encoding layer for feature extraction to obtain the target feature vector corresponding to the vehicle; inputting the target feature vector into the intermediate layer for temporal modeling strategy optimization to obtain multiple sets of initial trajectory parameters; and inputting the multiple sets of initial trajectory parameters into the output layer for trajectory generation to obtain multiple candidate driving trajectories.

[0075] Optionally, the encoding layer is specifically used to: perform multimodal spatiotemporal alignment on the target state data, and extract features from the aligned target state data to obtain a target feature vector, wherein the target feature vector includes spatial features, temporal features, and semantic features corresponding to the target state data.

[0076] Optionally, the output layer is specifically used to: perform trajectory smoothing processing on multiple sets of initial trajectory parameters to obtain multiple initial driving trajectories, and perform risk assessment on the initial driving trajectories based on vehicle dynamics models and collision detection algorithms to obtain risk assessment results; and in response to the risk assessment results satisfying preset driving effects, determine the initial driving trajectories that satisfy preset driving effects as candidate driving trajectories.

[0077] Optionally, the trajectory evaluation module includes an input layer, a processing layer, and an evaluation layer; the trajectory evaluation unit is specifically used to: input the target state information and the candidate driving trajectory into the input layer for feature concatenation to obtain a joint feature vector; input the joint feature vector into the processing layer for feature extraction to obtain a trajectory evaluation vector; and input the trajectory evaluation vector into the evaluation layer for trajectory evaluation to obtain the trajectory evaluation result corresponding to the candidate driving trajectory.

[0078] The above-described device can execute the autonomous driving trajectory generation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the autonomous driving trajectory generation method.

[0079] Example 4

[0080] Figure 4This is a schematic diagram of the structure of an electronic device implementing the autonomous driving trajectory generation method of this invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0081] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded into the RAM 13 from storage unit 18. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0082] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0083] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as autonomous driving trajectory generation methods.

[0084] In some embodiments, the autonomous driving trajectory generation method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the autonomous driving trajectory generation method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the autonomous driving trajectory generation method by any other suitable means (e.g., by means of firmware).

[0085] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication unit 19, or installed from storage unit 18, or installed from ROM 12. When the computer program is executed by processor 11, it performs the functions defined in the methods of the embodiments of the present invention.

[0086] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0087] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0088] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0089] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0090] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0091] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0092] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0093] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for generating autonomous driving trajectories, characterized in that, include: Obtain the target status information of the vehicle; The target state information is input into a pre-trained trajectory planning model to generate a trajectory, thereby determining multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory. The trajectory planning model is constructed based on a deep neural network. The target driving trajectory is determined based on multiple candidate driving trajectories and the trajectory evaluation results corresponding to each candidate driving trajectory.

2. The method according to claim 1, characterized in that, The acquisition of the vehicle's target status information includes: In response to the user's autonomous driving activation operation, the vehicle's original state information is obtained, including vehicle body state information, obstacle information, road structure information, and traffic state information. The original state information is preprocessed to obtain the target state information, wherein the preprocessing operation includes at least one of data cleaning and standardization.

3. The method according to claim 1, characterized in that, The trajectory planning model includes a trajectory generation module and a trajectory evaluation module. The step of inputting the target state information into the pre-trained trajectory planning model for trajectory generation, determining multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory, includes: The target state information is input into the trajectory generation module for trajectory planning to determine multiple candidate driving trajectories for the vehicle. For each candidate driving trajectory, the target state information and the candidate driving trajectory are input into the trajectory evaluation module to determine the trajectory evaluation result corresponding to the candidate driving trajectory.

4. The method according to claim 3, characterized in that, The trajectory generation module includes: an encoding layer, an intermediate layer, and an output layer; The step of inputting the target state information into the trajectory generation module for trajectory planning to determine multiple candidate driving trajectories corresponding to the vehicle includes: The target state information is input into the coding layer for feature extraction to obtain the target feature vector corresponding to the vehicle; The target feature vector is input into the intermediate layer for strategy optimization of temporal modeling to obtain multiple sets of initial trajectory parameters; Multiple sets of initial trajectory parameters are input into the output layer to generate a trajectory, thereby obtaining multiple candidate driving trajectories.

5. The method according to claim 4, characterized in that, The encoding layer is specifically used to: perform multimodal spatiotemporal alignment on the target state data, and extract features from the aligned target state data to obtain a target feature vector, wherein the target feature vector includes spatial features, temporal features and semantic features corresponding to the target state data.

6. The method according to claim 4, characterized in that, The output layer is specifically used to: perform trajectory smoothing processing on multiple sets of initial trajectory parameters to obtain multiple initial driving trajectories, and perform risk assessment on the initial driving trajectories based on vehicle dynamics models and collision detection algorithms to obtain risk assessment results. In response to the risk assessment results satisfying preset driving effects, the initial driving trajectory that satisfies the preset driving effects is determined as a candidate driving trajectory.

7. The method according to claim 3, characterized in that, The trajectory evaluation module includes: an input layer, a processing layer, and an evaluation layer; The step of inputting the target state information and the candidate driving trajectory into the trajectory evaluation module to determine the trajectory evaluation result corresponding to the candidate driving trajectory includes: The target state information and the candidate driving trajectory are input into the input layer for feature concatenation to obtain a joint feature vector. The joint feature vector is input into the processing layer for feature extraction to obtain the trajectory evaluation vector; The trajectory evaluation vector is input into the evaluation layer for trajectory evaluation to obtain the trajectory evaluation result corresponding to the candidate driving trajectory.

8. An autonomous driving trajectory generation device, characterized in that, include: The information acquisition module is used to acquire the target status information of the vehicle; The candidate trajectory determination module is used to input the target state information into a pre-trained trajectory planning model to generate a trajectory, determine multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory, wherein the trajectory planning model is constructed based on a deep neural network; The target trajectory determination module is used to determine the target driving trajectory based on multiple candidate driving trajectories and the trajectory evaluation result corresponding to each candidate driving trajectory.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the autonomous driving trajectory generation method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the autonomous driving trajectory generation method according to any one of claims 1-7.

Images