Automatic driving vehicle path planning method, vehicle and storage medium

Abstract

The embodiment of the application provides a kind of automatic driving vehicle path planning method, vehicle and storage medium, the method comprises: according to the current state of vehicle in state space and action space is searched, obtains candidate action sequence set;At least one driving trajectory of vehicle under current state is obtained based on candidate action sequence set prediction;At least one driving trajectory is optimized using novelty atom, and initial driving trajectory set after optimization is obtained;The initial driving trajectory set is filtered by semantic sketch rule, and target driving trajectory set is obtained;The driving path of vehicle is planned based on target driving trajectory set.This application solves the technical problems of low efficiency of path planning of automatic driving vehicle in complex traffic environment in the prior art.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and more specifically, to an autonomous vehicle path planning method, a vehicle, and a storage medium. Background Technology

[0002] In the field of autonomous driving technology, existing path planning and behavior decision-making methods mainly rely on template matching or offline generated policy libraries. Although these methods can achieve autonomous navigation of vehicles to a certain extent, their performance in complex traffic environments is significantly inadequate. For example, when faced with sudden situations or constantly changing road conditions, most path planning and behavior decision-making algorithms have slow response speeds and poor adaptability, resulting in low path planning efficiency for autonomous vehicles in complex traffic environments.

[0003] There is currently no good solution to the above problems. Summary of the Invention

[0004] This application provides an autonomous vehicle path planning method, vehicle, and storage medium to at least solve the technical problem of low path planning efficiency of autonomous vehicles in complex traffic environments in the prior art.

[0005] According to one aspect of the embodiments of this application, an autonomous vehicle path planning method is provided. The method includes: searching in a state space and an action space based on the current state of the vehicle to obtain a candidate action sequence set, wherein the candidate action sequence set represents a set of driving strategies of the vehicle in the current state, the state space represents a combination of vehicle state information and dynamic obstacle state information, the action space represents the driving strategy adopted by the vehicle under the state combination, the vehicle state information represents the motion state of the vehicle at any time, and the dynamic obstacle state information represents the motion state of dynamic obstacles around the vehicle at any time; predicting at least one driving trajectory of the vehicle in the current state based on the candidate action sequence set; optimizing the at least one driving trajectory using novelty atoms to obtain an optimized initial driving trajectory set, wherein the novelty atoms are used to determine the state features corresponding to the novelty atoms in the at least one driving trajectory; filtering the initial driving trajectory set through semantic sketch rules to obtain a target driving trajectory set, wherein the semantic sketch rules are used to provide decision guidance for the driving behavior of the vehicle; and planning the driving path of the vehicle based on the target driving trajectory set.

[0006] Furthermore, the novelty atom includes: the lane where the vehicle is located, the lane where the dynamic obstacle is located, and the speed range. The autonomous vehicle path planning method also includes: constructing a novelty table based on the lane where the vehicle is located, the lane where the dynamic obstacle is located, and the speed range. The novelty table is used to store novelty atom features, and the novelty atom features are used to represent the state features that appear for the first time in at least one driving trajectory and correspond to the novelty atom.

[0007] Furthermore, at least one driving trajectory is optimized using novel atoms to obtain an optimized initial driving trajectory set, including: obtaining the state features corresponding to the novel atoms in any driving trajectory; pruning the duplicated state features in response to the duplication of state features and novel atom features; storing the non-duplicated state features in a novelty table in response to the non-duplication of state features and novel atom features; and determining the initial driving trajectory set based on the novel atom features in the novelty table.

[0008] Furthermore, the autonomous vehicle path planning method also includes: acquiring the vehicle's executed actions, the number of blocked lanes, the number of dynamic obstacles blocking the vehicle, and traffic rules; and determining at least one semantic sketch rule based on whether the vehicle performs lane changing or merging operations, the number of blocked lanes, the number of dynamic obstacles blocking the vehicle, and whether the vehicle violates traffic rules.

[0009] Furthermore, the initial driving trajectory set is filtered through semantic sketch rules to obtain the target driving trajectory set, including: in response to any driving trajectory in the initial driving trajectory set not conforming to the semantic sketch rules, driving trajectories that do not conform to the semantic sketch rules are removed; in response to any driving trajectory in the initial driving trajectory set conforming to the semantic sketch rules, driving trajectories that conform to the semantic sketch rules are retained; and the target driving trajectory set is determined based on at least one driving trajectory that conforms to the semantic sketch rules.

[0010] Furthermore, the vehicle's driving path is planned based on the target driving trajectory set, including: evaluating multiple driving trajectories in the target driving trajectory set based on a preset trajectory cost function to obtain the driving trajectory to be driven, wherein the preset trajectory cost function is obtained by weighted summation of driving time cost, collision risk cost, ride comfort cost, and traffic rule violation cost; decomposing the driving trajectory to be driven into multiple control commands; and controlling the vehicle's driving according to the multiple control commands to determine the vehicle's driving path.

[0011] Furthermore, based on the candidate action sequence set, at least one driving trajectory of the vehicle in the current state is predicted, including: simulating any action sequence in the candidate action sequence set using the vehicle's dynamics model to obtain the vehicle's first position state information at the next moment, wherein the first position state information is used to represent the vehicle's position, heading angle, and speed; determining the second position state information of the dynamic obstacle at the next moment using an environmental prediction model, wherein the second position state information is used to represent the dynamic obstacle's position, direction, and behavioral intention; and evaluating the feasibility of the first position state information based on the second position state information to obtain at least one driving trajectory.

[0012] According to another aspect of the embodiments of this application, an autonomous vehicle path planning device is also provided, comprising: a search module, configured to search in a state space and an action space based on the current state of the vehicle to obtain a set of candidate action sequences, wherein the set of candidate action sequences represents a set of driving strategies of the vehicle in the current state, the state space represents a combination of vehicle state information and dynamic obstacle state information, the action space represents the driving strategy adopted by the vehicle under the state combination, the vehicle state information represents the motion state of the vehicle at any time, and the dynamic obstacle state information represents the motion state of dynamic obstacles around the vehicle at any time; a prediction module, configured to predict at least one driving trajectory of the vehicle in the current state based on the set of candidate action sequences; an optimization module, configured to optimize the at least one driving trajectory using novelty atoms to obtain an optimized initial driving trajectory set, wherein the novelty atoms are used to determine the state features corresponding to the novelty atoms in the at least one driving trajectory; a filtering module, configured to filter the initial driving trajectory set through semantic sketch rules to obtain a target driving trajectory set, wherein the semantic sketch rules are used to provide decision guidance for the driving behavior of the vehicle; and a planning module, configured to plan the driving path of the vehicle based on the target driving trajectory set.

[0013] According to another aspect of the embodiments of this application, a vehicle is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0014] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0015] According to another aspect of the embodiments of this application, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the methods in various embodiments of this application.

[0016] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0017] In this embodiment, a search is first performed in the state space and action space based on the vehicle's current state to obtain a set of candidate action sequences. Based on this set, at least one driving trajectory for the vehicle in its current state is predicted. Next, novelty atoms are used to optimize this trajectory, resulting in an optimized initial trajectory set. Then, semantic sketch rules are used to filter this initial trajectory set, yielding a target trajectory set. These semantic sketch rules provide decision-making guidance for the vehicle's driving behavior. Finally, the vehicle's driving path is planned based on the target trajectory set. By identifying and eliminating trajectories highly similar to known states using novelty atoms, and filtering the initial trajectory set using semantic sketch rules, the aim is to reduce collision risk and complete path planning. This improves the real-time performance and efficiency of path planning, thus solving the technical problem of low path planning efficiency for autonomous vehicles in complex traffic environments in existing technologies. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0019] Figure 1 This is a flowchart of an autonomous vehicle path planning method according to an embodiment of this application;

[0020] Figure 2 This is a flowchart of an optional vehicle routing and behavior joint planning according to an embodiment of this application;

[0021] Figure 3 This is a schematic diagram of an optional multi-parent node and orphan chain processing mechanism according to an embodiment of this application;

[0022] Figure 4 This is a schematic diagram of an optional novelty table repair mechanism according to an embodiment of this application;

[0023] Figure 5 This is a schematic diagram of an autonomous vehicle path planning device according to an embodiment of this application. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application 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 the embodiments of this application 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 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.

[0026] According to an embodiment of this application, an embodiment of an autonomous vehicle path planning method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0027] This method embodiment can be executed in an electronic device or similar computing device that includes memory and a processor. Taking operation on a computer terminal as an example, the computer terminal may include one or more processors (processors may include, but are not limited to, central processing units (CPUs), graphics processing units (GPUs), digital signal processing (DSP) chips, microcontroller units (MCUs), field-programmable gate arrays (FPGAs), neural network processors (NPUs), tensor processors (TPUs), artificial intelligence (AI) type processors, etc.) and memory for storing data. Optionally, the computer terminal may also include transmission devices, input / output devices, and display devices for communication functions. Those skilled in the art will understand that the above structural description is merely illustrative and does not limit the structure of the computer terminal. For example, the computer terminal may include more or fewer components than described above, or have a different configuration than described above.

[0028] The memory can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the autonomous vehicle path planning method in this embodiment. The processor executes various functional applications and data processing by running the computer program stored in the memory, thereby implementing the aforementioned autonomous vehicle path planning method. The memory may include high-speed random access memory and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the mobile terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0029] The transmission device is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the mobile terminal's communication provider. In one example, the transmission device includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0030] Display devices can be, for example, touchscreen liquid crystal displays (LCDs) and touch displays (also referred to as "touchscreens" or "touch displays"). The LCD allows users to interact with the user interface of the mobile terminal. In some embodiments, the mobile terminal has a graphical user interface (GUI), which allows users to interact with the GUI through finger contact and / or gestures on a touch-sensitive surface. Optional human-computer interaction functions include: creating web pages, drawing, word processing, creating electronic documents, playing games, video conferencing, instant messaging, sending and receiving emails, call interfaces, playing digital video, playing digital music, and / or web browsing, etc. Executable instructions for performing the above human-computer interaction functions are configured / stored in one or more processor-executable computer program products or readable storage media.

[0031] This embodiment provides a path planning method for autonomous vehicles. Figure 1 This is a flowchart of an autonomous vehicle path planning method according to an embodiment of this application, such as... Figure 1 As shown, the process includes the following steps:

[0032] Step S11: Search in the state space and action space according to the current state of the vehicle to obtain a set of candidate action sequences. The set of candidate action sequences is used to represent the set of driving strategies of the vehicle in the current state. The state space is used to represent the state combination of vehicle state information and dynamic obstacle state information. The action space is used to represent the driving strategy adopted by the vehicle under the state combination. The vehicle state information is used to represent the motion state of the vehicle at any time. The dynamic obstacle state information is used to represent the motion state of the dynamic obstacles around the vehicle at any time.

[0033] In this embodiment, the current state of the vehicle includes dynamic information such as the vehicle's current position, heading, speed, and acceleration, as well as the position, speed, and expected behavior of surrounding dynamic obstacles, which are not limited here.

[0034] The state space contains all possible combinations of the vehicle's position, orientation, velocity, and acceleration at any given moment, as well as the possible combinations of the position and velocity of each dynamic obstacle at any given moment, which are not restricted here.

[0035] The action space is the set of operations that a vehicle can perform in any combination of states. It covers all vehicle control behaviors, such as acceleration, deceleration, lane changing, and turning, and is not restricted here.

[0036] The candidate action sequence set contains multiple sequences of consecutive actions. Each sequence corresponds to a possible path from the current state to the target state or sub-target state, which is not restricted here.

[0037] Searching in the state space and action space based on the vehicle's current state to obtain a set of candidate action sequences can be understood as follows: Based on the vehicle's current position, heading, speed, acceleration, and other dynamic information, as well as the position, speed, and expected movement path of surrounding dynamic obstacles such as other vehicles, pedestrians, or movable objects, a state space is constructed, and the action space is defined to include basic vehicle control behaviors such as acceleration, deceleration, lane changing, and turning. By alternately generating candidate states in the state space and selecting candidate actions in the action space, a set of candidate action sequences from the vehicle's current state to its future state is constructed.

[0038] It can be seen that through intelligent search in the state-action joint space, the system can generate multiple alternative driving paths and behavior strategies, providing a variety of candidate solutions for subsequent trajectory optimization and decision-making, thereby achieving safe, efficient and rule-compliant autonomous driving in complex and ever-changing traffic environments.

[0039] Step S12: Based on the candidate action sequence set, predict at least one driving trajectory of the vehicle in the current state;

[0040] In this embodiment, at least one driving trajectory refers to a series of possible paths predicted based on a set of candidate action sequences, from the current state of the vehicle to the target state or a sub-target state. Each driving trajectory is a spatiotemporal sequence of the vehicle's position, heading, speed, and acceleration changing over time when the vehicle executes a specific action sequence, reflecting the vehicle's driving route and dynamic behavior on the road, and is not limited here.

[0041] Predicting at least one driving trajectory for a vehicle in its current state based on a set of candidate action sequences can be understood as follows: given the current driving environment, the autonomous driving system uses a set of candidate action sequences as input and, through vehicle dynamics models and environmental constraints, predicts one or more driving trajectories based on each action sequence.

[0042] It can be seen that predicting at least one driving trajectory based on a set of candidate action sequences improves the ability of autonomous vehicles to react to environmental changes and the real-time nature of their decisions.

[0043] Step S13: Use novel atoms to optimize at least one driving trajectory to obtain an optimized initial driving trajectory set, wherein the novel atoms are used to determine the state features corresponding to the novel atoms in at least one driving trajectory.

[0044] In this embodiment, novelty atoms are used to help the algorithm prune the state space more efficiently, avoiding repeated searches of the same or similar states, thereby improving the efficiency of the trajectory optimization process. For example, novelty atoms may include, but are not limited to, the following state features: vehicle-at-lane(p): indicating the state of a vehicle in a specific lane p; obstacle_i-at-lane(p): indicating the state of a dynamic obstacle i located in a specific lane p; speed_bin(v): indicating the state of a vehicle speed falling within a certain speed range v; acceleration_range(a): indicating the state of a vehicle acceleration falling within a certain acceleration range a.

[0045] The initial driving trajectory set is used to represent the results obtained after preliminary screening by novelty atoms, which includes a variety of possible vehicle driving paths and behavioral strategies.

[0046] Using novel atoms to optimize at least one driving trajectory to obtain an optimized initial set of driving trajectories can be understood as using novel atoms to evaluate and optimize at least one predicted driving trajectory to eliminate duplicate, invalid or low-value trajectories, thereby obtaining an initial set of driving trajectories.

[0047] It can be seen that through the initial optimization of novel atoms, the system can generate more reasonable and effective driving paths within limited time and computing resources, providing strong support for the safe and efficient driving of autonomous vehicles in complex traffic environments.

[0048] Step S14: Filter the initial driving trajectory set using semantic sketch rules to obtain the target driving trajectory set. The semantic sketch rules are used to provide decision guidance for the vehicle's driving behavior.

[0049] In this embodiment, semantic sketch rules are decision-making guidance mechanisms used to guide vehicle driving behavior. Semantic sketch rules integrate knowledge from fields such as traffic rules, road types, lane information, and dynamic obstacle prediction to determine and select safe, compliant, and efficient driving trajectories from the initial set of driving trajectories. For example, semantic sketch rules are determined based on the following features: H: whether the vehicle is performing a lane change or merging operation, m: the number of blocked lanes, n: the number of dynamic obstacles blocking the lanes, and s: the total number of obstacles, including obstacles and traffic rule conflicts. Examples of semantic sketch rules: (1) When ¬H and m>0 and n=0, the vehicle can directly enter the target lane; (2) When ¬H and m>0 and n>0, the vehicle prioritizes adjusting its speed or changing lanes to reduce obstruction; (3) When H is true, the vehicle should complete the current action before evaluating the next trajectory, which is not restricted here.

[0050] The target driving trajectory set is the final candidate trajectory set obtained by filtering and refining the optimized initial driving trajectory set through the application of semantic sketch rules. For example, the trajectories in the target driving trajectory set, in addition to satisfying physical feasibility and novelty, can also adapt to the current traffic environment, follow traffic rules, and have high safety, compliance and driving efficiency; however, no restrictions are imposed here.

[0051] Filtering the initial set of driving trajectories using semantic sketch rules to obtain the target set of driving trajectories can be understood as using pre-defined semantic sketch rules to perform in-depth filtering and evaluation on the initially optimized set of driving trajectories, in order to identify and select trajectories that comply with traffic rules, have high traffic efficiency and low risk, thereby obtaining the target set of driving trajectories.

[0052] It can be seen that by filtering the initial driving trajectory set through semantic sketch rules to obtain the target driving trajectory set, it is possible to accurately identify and filter driving trajectories suitable for the current driving situation with less computational cost, thereby ensuring that autonomous vehicles can make efficient and safe driving decisions in complex and ever-changing real traffic environments.

[0053] Step S15: Plan the vehicle's driving path based on the target driving trajectory set.

[0054] In this embodiment of the application, planning the vehicle's driving path based on the target driving trajectory set can be understood as selecting the most suitable driving path from the target driving trajectory set after multiple optimizations and screenings, which will be the route that the vehicle will actually follow in the following journey.

[0055] It can be seen that by using the target driving trajectory set to plan the vehicle's driving path, the vehicle can achieve safe and efficient autonomous navigation under complex and ever-changing road conditions.

[0056] Through the above steps, firstly, a search is performed in the state space and action space based on the vehicle's current state to obtain a set of candidate action sequences. Then, at least one driving trajectory for the vehicle in its current state is predicted based on this set of candidate action sequences. Next, novelty atoms are used to optimize this at least one driving trajectory, resulting in an optimized initial driving trajectory set. Then, semantic sketch rules are used to filter this initial driving trajectory set, resulting in a target driving trajectory set. The semantic sketch rules provide decision-making guidance for the vehicle's driving behavior. Finally, the vehicle's driving path is planned based on the target driving trajectory set. By identifying and eliminating trajectories highly similar to known states through the novelty atom mechanism, and filtering the initial driving trajectory set using semantic sketch rules, the aim is to reduce collision risks and complete path planning. This achieves the technical effect of improving the real-time performance and efficiency of path planning, thereby solving the technical problem of low path planning efficiency for autonomous vehicles in complex traffic environments in existing technologies.

[0057] Optionally, the novel atoms include: the vehicle's lane, the lane containing the dynamic obstacle, and the vehicle speed range. The autonomous vehicle path planning method also includes the following steps:

[0058] A novelty table is constructed based on the vehicle's lane, the lane of the dynamic obstacle, and the vehicle speed range. The novelty table is used to store novelty atom features, which represent the state features that appear for the first time in at least one driving trajectory and correspond to the novelty atom.

[0059] In this embodiment, the novelty table is a data structure used to record and track various novel atomic features that appear for the first time in the state space. Specifically, the novelty table records the state features of the autonomous vehicle during the path planning process, and highlights the state features encountered for the first time or with significant novelty, so that the system can make efficient and intelligent decisions during the search process.

[0060] Constructing a novelty table based on the vehicle's lane, the lanes of dynamic obstacles, and the vehicle's speed range can be understood as using the vehicle's current lane information, the location and lane of surrounding dynamic obstacles, and the vehicle's own speed range to create a novelty table to store and manage novelty atoms related to the aforementioned state features, i.e., the first occurrence of the aforementioned state features in the state-action space.

[0061] As can be seen, the novelty table helps the system identify and record novel state features in the state space, thereby optimizing the search process and improving the efficiency and quality of path planning by avoiding redundant calculations and explorations.

[0062] Optionally, in step S13, at least one driving trajectory is optimized using novel atoms to obtain an optimized initial driving trajectory set, including the following steps:

[0063] Step S131: Obtain the state features corresponding to the novel atom in any driving trajectory;

[0064] Step S132: In response to the repetition of state features and novel atomic features, the repetitive state features are pruned.

[0065] Step S133: In response to the fact that the state features and novel atomic features do not overlap, the non-overlapping state features are stored in the novelty table.

[0066] Step S134: Determine the initial driving trajectory set based on the novelty atomic features in the novelty table.

[0067] In this embodiment of the application, obtaining the state features corresponding to the novel atom in any driving trajectory can be understood as identifying the state features associated with the novel atom from the currently searched driving trajectory. For example, the state features corresponding to the novel atom in any driving trajectory may include, but are not limited to: "the vehicle is in lane 1", "the vehicle speed is in the range of 50-60 km / h", and "the obstacle ahead is in lane 2".

[0068] In response to the duplication of state features and novelty atomic features, the pruning of duplicate state features can be understood as determining whether the acquired state features have appeared in previous searches. If the acquired state features are duplicated with novelty atomic features in the novelty table, the duplicate state features are deleted.

[0069] In response to the fact that the state feature and the novelty atom feature do not overlap, storing the non-overlapping state feature in the novelty table can be understood as follows: when the acquired state feature does not overlap with the novelty atom feature in the novelty table, it means that the state feature is appearing for the first time, and the first-appearing state feature is stored in the novelty table.

[0070] Determining the initial driving trajectory set based on the novelty atomic features in the novelty table can be understood as determining the vehicle's lane, the lane of the dynamic obstacle, and the speed range based on the novelty atomic features, and then determining the vehicle's initial driving trajectory set based on the vehicle's lane, the lane of the dynamic obstacle, and the speed range.

[0071] As can be seen, through the above steps, the embodiments of this application achieve optimized screening of state features through the novelty atom and novelty table mechanism, ensuring the diversity and quality of driving path planning results.

[0072] Optionally, the autonomous vehicle path planning method also includes the following steps:

[0073] Acquire information on vehicle actions, the number of blocked lanes, the number of dynamic obstacles obstructing vehicles, and traffic rules.

[0074] At least one semantic sketch rule is determined based on whether the vehicle performs a lane change or merging operation, the number of blocked lanes, the number of dynamic obstacles blocking the vehicle, and whether the vehicle violates traffic rules.

[0075] In this embodiment, the vehicle's actions refer to the current control commands of the autonomous vehicle, including but not limited to acceleration, deceleration, maintaining speed, lane changing, merging, obstacle avoidance, etc., which are not limited here.

[0076] The number of obstructed lanes indicates the number of lanes obstructed in the direction of vehicle travel due to static or dynamic obstacles. For example, on a multi-lane highway, the second lane is considered an obstructed lane if there is a slow-moving vehicle or obstacle ahead; this is not a limitation.

[0077] The number of dynamic obstacles blocking a vehicle is used to indicate the number of dynamic obstacles present in the vehicle's path, particularly in the current lane or the target lane.

[0078] Traffic rules refer to the various laws, regulations and road standards that vehicles must follow when driving on the road, including but not limited to traffic lights, speed limit signs, road markings, and rules for yielding to other vehicles at intersections, etc., which are not specified here.

[0079] Acquiring information such as vehicle actions, the number of blocked lanes, the number of dynamic obstacles blocking vehicles, and traffic rules can be understood as monitoring and collecting key information about the current vehicle status, including vehicle actions, the number of blocked lanes, the number of dynamic obstacles blocking vehicles, and traffic rules, thereby providing a basis for subsequent decision-making.

[0080] Determining at least one semantic sketch rule based on whether a vehicle performs a lane-changing or merging operation, the number of obstructed lanes, the number of dynamic obstacles blocking the vehicle, and whether the vehicle violates traffic rules can be understood as using the aforementioned real-time acquired vehicle state and environmental information to generate a set of high-level rules, i.e., semantic sketch rules, to guide the vehicle's next action. Semantic sketch rules combine specific road conditions and vehicle operating states to guide path planning algorithms in making reasonable choices in complex situations.

[0081] For example, when a vehicle is performing a lane-changing or merging maneuver, the system prioritizes the conditions for completion, such as whether there is sufficient space and time to complete the lane change, whether it will affect the normal driving of vehicles in other lanes, and whether it is necessary to adjust the speed in advance to match the traffic flow in the target lane. If there are many blocked lanes, it means that the vehicle's straight-ahead route is not smooth. In this case, the system may be inclined to look for lane-changing opportunities or adjust to a less congested lane to improve overall driving efficiency. When the number of dynamic obstacles blocking the vehicle increases, the system is more likely to adopt a strategy of slowing down or stopping to ensure a safe distance from the obstacles ahead and avoid potential collision risks. Finally, in all cases, the system will check whether the currently planned behavior complies with traffic rules. For example, it must stop when encountering a red light signal and adjust the speed when encountering a speed limit sign to ensure the legality of driving. There are no restrictions on this.

[0082] It can be seen that by establishing semantic sketch rules, vehicles can make intelligent decisions that meet both safety standards and traffic rules in various emergencies, thereby improving the driving experience.

[0083] Optionally, in step S14, the initial driving trajectory set is filtered using semantic sketch rules to obtain the target driving trajectory set, including the following steps:

[0084] Step S141: In response to any driving trajectory in the initial driving trajectory set not conforming to the semantic sketch rules, driving trajectories that do not conform to the semantic sketch rules are removed.

[0085] Step S142: In response to any driving trajectory in the initial driving trajectory set conforming to the semantic sketch rules, retain the driving trajectory that conforms to the semantic sketch rules;

[0086] Step S143: Determine the target driving trajectory set based on at least one driving trajectory that conforms to the semantic sketch rules.

[0087] In this embodiment, in response to any driving trajectory in the initial driving trajectory set failing to conform to the semantic sketch rules, removing driving trajectories that do not conform to the semantic sketch rules can be understood as filtering the driving trajectories in the initial driving trajectory set to ensure that all trajectories follow the pre-defined semantic sketch rules. If a driving trajectory violates the semantic rules during evaluation, such as unreasonable lane changing operations, ignoring traffic signals, or failing to effectively avoid obstacles, then that trajectory is removed; this is not restricted here.

[0088] In response to any driving trajectory in the initial driving trajectory set conforming to the semantic sketch rules, retaining driving trajectories that conform to the semantic sketch rules can be understood as follows: if a driving trajectory in the initial driving trajectory set conforms to all semantic sketch rules during evaluation, it indicates that the trajectory will be marked as compliant and retained in the candidate trajectory set.

[0089] Determining the target driving trajectory set based on at least one driving trajectory that conforms to the semantic sketch rule can be understood as unifying and integrating all driving trajectories that conform to the semantic sketch rule to obtain the target driving trajectory set.

[0090] As can be seen, by filtering any one of the initial driving trajectories in the semantic sketch rule, the target driving trajectory set is obtained, ensuring that the actual driving path of the vehicle is both reasonable and reliable, and can effectively cope with complex and ever-changing road conditions.

[0091] Optionally, in step S15, the vehicle's driving path is planned based on the target driving trajectory set, including the following steps:

[0092] Step S151: Evaluate multiple driving trajectories in the target driving trajectory set based on the preset trajectory cost function to obtain the driving trajectory to be driven. The preset trajectory cost function is obtained by weighted summation of driving time cost, collision risk cost, ride comfort cost and traffic rule violation cost.

[0093] Step S152: Decompose the driving trajectory into multiple control commands;

[0094] Step S153: Control the vehicle's movement according to multiple control commands to determine the vehicle's driving path.

[0095] In this embodiment, a preset trajectory cost function is used to quantify and compare the merits of each trajectory in the target trajectory set. The preset trajectory cost function calculates the costs in the following four aspects and sums them in a weighted manner to obtain the total cost of each trajectory: (1) Travel time cost: assesses the expected time required to complete the trajectory, with time efficiency as the indicator. (2) Collision risk cost: measures the probability of the trajectory colliding with dynamic or static obstacles during the journey. (3) Ride comfort cost: assesses the smoothness and comfort of the vehicle when traveling along the trajectory, usually involving factors such as acceleration changes and sharp turns. (4) Traffic rule violation cost: quantifies the degree of traffic rule violation of the trajectory, such as running red lights, speeding, etc., which are not limited here.

[0096] The driving trajectory to be driven is the driving trajectory with the minimum cost obtained through the above cost function evaluation. This driving path is not only physically feasible, but also minimizes driving time and collision risk, improves ride comfort, and complies with traffic rules. It is the path that the autonomous vehicle will execute.

[0097] The trajectory to be driven is determined by evaluating multiple driving trajectories in the target driving trajectory set based on a preset trajectory cost function. This can be understood as using the preset trajectory cost function to evaluate multiple driving trajectories in the target driving trajectory set from multiple dimensions, including driving time cost, collision risk cost, ride comfort cost, and traffic rule violation cost, and determining the trajectory with the minimum cost.

[0098] Decomposing the driving trajectory into multiple control commands can be understood as converting the driving trajectory into a series of control commands that can be directly executed by the vehicle. These commands typically include, but are not limited to: acceleration commands, instructing the vehicle to increase speed; deceleration commands, instructing the vehicle to slow down; steering commands, instructing the vehicle to adjust its direction to the left or right; lane change commands, guiding the vehicle to change lanes at appropriate times; and braking commands, requiring the vehicle to stop immediately in an emergency.

[0099] Controlling a vehicle to determine its path by issuing multiple control commands can be understood as follows: based on the generated sequence of control commands, the autonomous driving system will direct the various control systems of the vehicle (such as the power system, steering system, braking system, etc.) to perform corresponding operations so that the vehicle can travel along the trajectory to be traveled.

[0100] It can be seen that by evaluating the trajectory with the minimum cost through the preset trajectory cost function, and controlling the vehicle's driving based on the trajectory, the autonomous vehicle can achieve the predetermined goal while maintaining a high level of driving quality in complex and ever-changing traffic environments.

[0101] Optionally, in step S12, predicting at least one driving trajectory of the vehicle in its current state based on the candidate action sequence set includes the following steps:

[0102] Step S121: Simulate any action sequence in the candidate action sequence set using the vehicle's dynamic model to obtain the vehicle's first position state information at the next moment, wherein the first position state information is used to represent the vehicle's position, heading angle, and speed.

[0103] Step S122: Determine the second position state information of the dynamic obstacle at the next moment through the environmental prediction model, wherein the second position state information is used to represent the position, orientation and behavioral intention of the dynamic obstacle;

[0104] Step S123: Evaluate the feasibility of the first position status information based on the second position status information to obtain at least one driving trajectory.

[0105] In this embodiment of the application, the dynamic model is used to predict the future state changes of the vehicle after the candidate action sequence is applied, including key parameters such as position, heading angle, and speed, which are not limited here.

[0106] Environmental prediction models are used to predict the possible location, direction, and behavioral intentions of dynamic obstacles on the road (such as other vehicles, pedestrians, bicycles, etc.) at a future moment, without any restrictions.

[0107] Simulating any action sequence from the candidate action sequence set using the vehicle's dynamics model to obtain the vehicle's first position state information at the next moment can be understood as using the vehicle's dynamics model to simulate each candidate action sequence proposed by the autonomous driving system, and predicting the vehicle's position, heading angle, speed, and other state information at the next moment.

[0108] The second position state information of a dynamic obstacle at the next moment is determined by an environmental prediction model. The second position state information is used to represent the position, direction and behavioral intention of the dynamic obstacle. This can be understood as using the environmental prediction model to predict the dynamic obstacle (such as other vehicles, pedestrians, etc.) to obtain the position, direction and behavioral intention of the dynamic obstacle at the next moment.

[0109] Assessing the feasibility of the first positional state information based on the second positional state information to obtain at least one driving trajectory can be understood as combining the first positional state information predicted by the vehicle dynamics model with the second positional state information predicted by the environmental prediction model to evaluate the feasibility of the vehicle's future path. For example, the evaluation process typically involves checking whether the predicted vehicle state conflicts with dynamic obstacles on the road, violates traffic rules, and meets safety and comfort requirements, thus selecting feasible driving trajectories from candidate action sequences.

[0110] As can be seen, through the above steps, the autonomous driving system can effectively generate a driving trajectory that adapts to the current traffic environment, which not only improves driving safety but also enhances the vehicle's decision-making ability when facing dynamic obstacles and uncertain traffic conditions.

[0111] Figure 2 This is a flowchart of an optional vehicle path and behavior joint planning method according to an embodiment of this application, such as... Figure 2As shown, firstly, vehicle information (current vehicle position, speed, and motion state), target information (desired destination and sub-target constraints), environmental information (position, speed, and behavior prediction results of dynamic obstacles), and traffic rules and road constraints (such as lane boundaries, traffic lights, and traffic signs) are acquired through onboard sensors and high-precision maps. Then, based on the aforementioned vehicle information, target information, environmental information, and traffic rules and road constraints, a joint search problem involving state space and action space is constructed.

[0112] The state space is constructed as follows, and the vehicle's geometric pose is represented using a special Euclidean group in 2 dimensions (SE(2)):

[0113]

[0114] in, The position of the vehicle in the plane. This is the heading angle.

[0115] Based on this, the vehicle state space is expanded to include speed. and acceleration The vehicle's dynamic status is obtained:

[0116]

[0117] in, This represents a set of dynamic parameters, including velocity and acceleration.

[0118] For N dynamic obstacles in a traffic environment, the obstacle state is represented as:

[0119]

[0120] The joint state space consists of vehicle state, obstacle state, and vehicle dynamics state, and its representation is as follows:

[0121]

[0122] The state space includes the vehicle's position in a two-dimensional space and , N The two-dimensional position and orientation combination of a dynamic obstacle and the dynamic state of the vehicle. Parameters including velocity, acceleration, etc. N This represents the number of dynamic obstacles.

[0123] The action space is constructed as follows, including discrete control actions (lane change, speed control, turning or straight driving, merging), and represented by vectors:

[0124] a = [Δv, Δθ]

[0125] Here, Δv represents the speed increment (acceleration / deceleration), and Δθ represents the heading angle increment (corresponding to the change in steering wheel angle). The feasibility of the action depends on vehicle dynamics constraints and collision detection results, and an inertial geometry verification strategy is used for judgment.

[0126] In addition, the vehicle's state and actions must meet the following dynamic constraints to ensure that the trajectory is continuous and physically feasible:

[0127]

[0128] in, Current vehicle status For the current action, This represents the vehicle dynamics model.

[0129] The system generates a finite set of states through an adaptive sampling module. Vehicle state sampling is performed near key lanes, intersections, or potential obstacles to ensure state connectivity; obstacle state sampling considers road restrictions and traffic rule constraints; action candidate sampling forms a finite set of candidate actions by sampling action parameters (such as target lane for lane changing and speed range). The sampling density can be dynamically adjusted according to the complexity of the traffic environment, reducing computational overhead while ensuring search completeness.

[0130] To improve search efficiency, novel atoms are introduced during the search process to characterize state features as follows:

[0131] (1) vehicle-at-lane(p): The vehicle is located in lane p;

[0132] (2) obstacle_i-at-lane(p): obstacle i is located in lane p;

[0133] (3) speed_bin(v): The vehicle speed falls within the interval v.

[0134] In Lazy Iterative Width-based Search (Lazy-IW(k)) or Lazy Symbolic-Interaction Width-based Search (Lazy-SIWR), the multiple parent node mechanism ensures that potential solutions are not lost. Novelty atoms are used to prune repetitive or low-novelty states, thereby avoiding redundant computation and improving search efficiency.

[0135] Furthermore, semantic sketch rules guide high-level behavior searches. These rules are based on features such as whether the vehicle is currently performing a lane-changing or merging operation (H), the number of blocked lanes (m), the number of dynamic obstacles blocking the lanes (n), and the total number of obstructions (s, including obstacles and traffic rule conflicts). Based on these features, the following rule examples can be derived: when ¬H and m>0 and n=0, the vehicle can directly enter the target lane; when ¬H and m>0 and n>0, the vehicle prioritizes adjusting its speed or changing lanes to reduce obstruction; when H is true, the vehicle should complete its current action before evaluating its next trajectory. The semantic sketch rules, combined with the Lazy-SIWR sub-objective generation mechanism, improve search efficiency and behavioral rationality by prioritizing states that reduce obstruction or complete sub-objectives.

[0136] Based on the current vehicle state, candidate action sequences are sampled from the action space to generate and corresponding state trajectories are predicted. The state trajectories are then filtered using novelty atoms and semantic sketch rules. Specifically, the novelty atom mechanism is used to eliminate low-novelty trajectories, reducing the search space. Candidate trajectories are searched alternately in the state space, and lazy geometry, semantic sketch rules, and dynamic verification are combined to compute only necessary nodes, improving search efficiency.

[0137] To optimize the quality of the generated trajectory, the cost function is defined as follows:

[0138]

[0139] in, This represents the cost of travel time, used to measure the time required for a vehicle to complete a planned task. This represents the cost of collision risk, used to measure the risk of a vehicle colliding with dynamic or static obstacles during path execution. This represents the trade-off between trajectory smoothness and comfort, used to measure the impact of vehicle acceleration, steering changes, and other factors on the riding experience. This represents the cost of violating traffic rules and is used to balance the importance of various indicators in the overall cost function. These are weighting coefficients used to balance the importance of various indicators.

[0140] Finally, the generated trajectory is evaluated based on the above cost function, and the trajectory with the minimum cost is selected as the vehicle's driving path, providing strong support for the safe driving of vehicles in complex traffic environments.

[0141] Figure 3 This is a schematic diagram of an optional multi-parent node and orphan chain processing mechanism according to an embodiment of this application, as shown below. Figure 3As shown, a common problem in state-space search during autonomous driving path planning is that a current node may be abandoned during exploration because the path or behavior corresponding to its parent node does not meet safety or regulatory requirements, thus forming an orphan chain, even if the current node itself is a perfectly feasible state. To overcome the orphan chain problem, this application links each child node to multiple potential parent nodes (alternative paths) when constructing the search tree. A parent node can potentially reach the state of the child node by executing a different set of action sequences. In this way, even if a certain action sequence fails to be validated, the system can still select a suitable path from other valid action sequences to proceed.

[0142] Figure 4 This is a schematic diagram of an optional novelty table repair mechanism according to an embodiment of this application, such as... Figure 4 As shown, the novelty table repair mechanism aims to maintain the accuracy and timeliness of the novelty table during the dynamic search process. When the system detects a change in the attributes or environmental conditions of a node, causing it to no longer meet the original novelty criteria, the system automatically updates the records in the novelty table to ensure that the information in the table reflects the latest state space structure. Specifically, it first monitors changes in the attributes of state nodes, such as position, velocity, and obstacle positions, as well as changes in environmental rules (such as traffic light status and road construction information). Next, it re-evaluates the novelty of the affected nodes, including their uniqueness in the state space and their differences from explored nodes. Then, it modifies the records of relevant nodes in the novelty table, which may include deleting outdated records (e.g., deleting the novelty contributed by node X when it loses its parent node and becomes an orphan node), adding new records, or updating the novelty state of existing records. Finally, based on the updated novelty table, the search strategy is adjusted, pruning nodes that no longer possess sufficient novelty or re-evaluating their value as parent nodes to avoid losing feasible solutions. Through the above mechanism, the system can not only promptly correct information discrepancies in the novelty table, but also dynamically adjust the search space to ensure the efficiency of the search process and the reliability of the search results.

[0143] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0144] According to an embodiment of this application, an embodiment of an autonomous vehicle path planning device is provided. It should be noted that the device can be used to execute the above-described autonomous vehicle path planning method.

[0145] Figure 5 This is a schematic diagram of an autonomous vehicle path planning device according to an embodiment of this application, such as... Figure 5 As shown, the autonomous vehicle path planning device 500 includes: a search module 501, used to search in the state space and action space according to the current state of the vehicle to obtain a set of candidate action sequences, wherein the set of candidate action sequences represents the set of driving strategies of the vehicle in the current state, the state space represents the state combination of vehicle state information and dynamic obstacle state information, the action space represents the driving strategy adopted by the vehicle under the state combination, the vehicle state information represents the motion state of the vehicle at any time, and the dynamic obstacle state information represents the motion state of the dynamic obstacles around the vehicle at any time; and a prediction module 502. The system is configured to predict at least one driving trajectory of the vehicle in its current state based on a set of candidate action sequences; the optimization module 503 is configured to optimize at least one driving trajectory using novelty atoms to obtain an optimized initial driving trajectory set, wherein the novelty atoms are used to determine the state features corresponding to the novelty atoms in at least one driving trajectory; the filtering module 504 is configured to filter the initial driving trajectory set using semantic sketch rules to obtain a target driving trajectory set, wherein the semantic sketch rules are used to provide decision guidance for the vehicle's driving behavior; and the planning module 505 is configured to plan the vehicle's driving path based on the target driving trajectory set.

[0146] It should be noted that the above modules can be implemented by software or hardware. For the latter, they can be implemented in the following ways, but are not limited to: all the above modules are located in the same processor; or, the above modules are located in different processors in any combination.

[0147] Embodiments of this application also provide a vehicle, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods described in various embodiments of this application when it runs.

[0148] Optionally, in this embodiment, the processor can be configured to perform the following steps via a computer program:

[0149] Step S11: Search in the state space and action space according to the current state of the vehicle to obtain a set of candidate action sequences. The set of candidate action sequences is used to represent the set of driving strategies of the vehicle in the current state. The state space is used to represent the state combination of vehicle state information and dynamic obstacle state information. The action space is used to represent the driving strategy adopted by the vehicle under the state combination. The vehicle state information is used to represent the motion state of the vehicle at any time. The dynamic obstacle state information is used to represent the motion state of the dynamic obstacles around the vehicle at any time.

[0150] Step S12: Based on the candidate action sequence set, predict at least one driving trajectory of the vehicle in the current state;

[0151] Step S13: Use novel atoms to optimize at least one driving trajectory to obtain an optimized initial driving trajectory set, wherein the novel atoms are used to determine the state features corresponding to the novel atoms in at least one driving trajectory.

[0152] Step S14: Filter the initial driving trajectory set using semantic sketch rules to obtain the target driving trajectory set. The semantic sketch rules are used to provide decision guidance for the vehicle's driving behavior.

[0153] Step S15: Plan the vehicle's driving path based on the target driving trajectory set.

[0154] Embodiments of this application also provide a computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0155] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0156] Step S11: Search in the state space and action space according to the current state of the vehicle to obtain a set of candidate action sequences. The set of candidate action sequences is used to represent the set of driving strategies of the vehicle in the current state. The state space is used to represent the state combination of vehicle state information and dynamic obstacle state information. The action space is used to represent the driving strategy adopted by the vehicle under the state combination. The vehicle state information is used to represent the motion state of the vehicle at any time. The dynamic obstacle state information is used to represent the motion state of the dynamic obstacles around the vehicle at any time.

[0157] Step S12: Based on the candidate action sequence set, predict at least one driving trajectory of the vehicle in the current state;

[0158] Step S13: Use novel atoms to optimize at least one driving trajectory to obtain an optimized initial driving trajectory set, wherein the novel atoms are used to determine the state features corresponding to the novel atoms in at least one driving trajectory.

[0159] Step S14: Filter the initial driving trajectory set using semantic sketch rules to obtain the target driving trajectory set. The semantic sketch rules are used to provide decision guidance for the vehicle's driving behavior.

[0160] Step S15: Plan the vehicle's driving path based on the target driving trajectory set.

[0161] Embodiments of this application also provide an electronic device including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the methods in various embodiments of this application.

[0162] Optionally, in this embodiment, the processor can be configured to perform the following steps via a computer program:

[0163] Step S11: Search in the state space and action space according to the current state of the vehicle to obtain a set of candidate action sequences. The set of candidate action sequences is used to represent the set of driving strategies of the vehicle in the current state. The state space is used to represent the state combination of vehicle state information and dynamic obstacle state information. The action space is used to represent the driving strategy adopted by the vehicle under the state combination. The vehicle state information is used to represent the motion state of the vehicle at any time. The dynamic obstacle state information is used to represent the motion state of the dynamic obstacles around the vehicle at any time.

[0164] Step S12: Based on the candidate action sequence set, predict at least one driving trajectory of the vehicle in the current state;

[0165] Step S13: Use novel atoms to optimize at least one driving trajectory to obtain an optimized initial driving trajectory set, wherein the novel atoms are used to determine the state features corresponding to the novel atoms in at least one driving trajectory.

[0166] Step S14: Filter the initial driving trajectory set using semantic sketch rules to obtain the target driving trajectory set. The semantic sketch rules are used to provide decision guidance for the vehicle's driving behavior.

[0167] Step S15: Plan the vehicle's driving path based on the target driving trajectory set.

[0168] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0169] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0170] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0171] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0172] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0173] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0174] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A path planning method for autonomous vehicles, characterized in that, The method includes: Based on the current state of the vehicle, a search is performed in the state space and action space to obtain a set of candidate action sequences. The set of candidate action sequences represents the set of driving strategies of the vehicle in the current state. The state space represents the state combination of vehicle state information and dynamic obstacle state information. The action space represents the driving strategy adopted by the vehicle under the state combination. The vehicle state information represents the motion state of the vehicle at any time. The dynamic obstacle state information represents the motion state of the dynamic obstacles around the vehicle at any time. Based on the set of candidate action sequences, at least one driving trajectory of the vehicle in the current state is predicted; The novel atom is used to optimize the at least one driving trajectory to obtain an optimized initial driving trajectory set, wherein the novel atom is used to determine the state features corresponding to the novel atom in the at least one driving trajectory; The initial driving trajectory set is filtered by semantic sketch rules to obtain the target driving trajectory set, wherein the semantic sketch rules are used to provide decision guidance for the driving behavior of the vehicle; The vehicle's driving path is planned based on the target driving trajectory set.

2. The method according to claim 1, characterized in that, The novel atoms include: the vehicle's lane, the lane containing the dynamic obstacle, and the vehicle speed range; the method further includes: A novelty table is constructed based on the vehicle's lane, the lane of the dynamic obstacle, and the vehicle speed range. The novelty table is used to store novelty atom features, which represent the state features that appear for the first time in the at least one driving trajectory and correspond to the novelty atom.

3. The method according to claim 2, characterized in that, The process of optimizing the at least one driving trajectory using novel atoms to obtain an optimized initial driving trajectory set includes: Obtain the state features corresponding to the novel atom in any driving trajectory; In response to the duplication of the state feature and the novel atomic feature, the duplicated state feature is pruned. In response to the absence of duplication between the state feature and the novelty atom feature, the state feature that does not appear repeatedly is stored in the novelty table; The initial set of driving trajectories is determined based on the novel atomic features in the novelty table.

4. The method according to claim 1, characterized in that, The method further includes: The system acquires the vehicle's actions, the number of blocked lanes, the number of dynamic obstacles blocking the vehicle, and traffic rules. At least one semantic sketch rule is determined based on whether the vehicle performs a lane change or merging operation, the number of obstructed lanes, the number of dynamic obstacles blocking the vehicle, and whether the vehicle violates the traffic rules.

5. The method according to claim 4, characterized in that, The step of filtering the initial driving trajectory set using semantic sketch rules to obtain the target driving trajectory set includes: In response to any driving trajectory in the initial driving trajectory set not conforming to the semantic sketch rules, the driving trajectory that does not conform to the semantic sketch rules is removed; In response to any driving trajectory in the initial driving trajectory set conforming to the semantic sketch rule, the driving trajectory that conforms to the semantic sketch rule is retained; The target driving trajectory set is determined based on at least one driving trajectory that conforms to the semantic sketch rules.

6. The method according to claim 1, characterized in that, The step of planning the vehicle's driving path based on the target driving trajectory set includes: The target driving trajectory set is evaluated based on a preset trajectory cost function to obtain the driving trajectory to be driven. The preset trajectory cost function is obtained by weighted summation of driving time cost, collision risk cost, ride comfort cost and traffic rule violation cost. The driving trajectory is decomposed into multiple control commands; The vehicle is controlled to travel according to the multiple control commands in order to determine the vehicle's travel path.

7. The method according to claim 1, characterized in that, The step of predicting at least one driving trajectory of the vehicle in the current state based on the candidate action sequence set includes: The vehicle's dynamic model is used to simulate any action sequence in the candidate action sequence set to obtain the vehicle's first position state information at the next moment, wherein the first position state information is used to represent the vehicle's position, heading angle, and speed. The second position state information of the dynamic obstacle at the next moment is determined by an environmental prediction model, wherein the second position state information is used to represent the position, orientation and behavioral intention of the dynamic obstacle; The feasibility of the first location status information is evaluated based on the second location status information to obtain the at least one driving trajectory.

8. A vehicle, characterized in that, include: Memory, which stores executable programs; A processor for running the executable program, wherein the executable program, when running on the processor, performs the autonomous vehicle path planning method as described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program is configured to execute the autonomous vehicle path planning method according to any one of claims 1 to 7 when run on a computer or processor.

10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to perform the autonomous vehicle path planning method as described in any one of claims 1 to 7.

Images