Local trajectory planning method and system based on random model

Abstract

The invention belongs to the technical field of automatic driving, and particularly relates to a local trajectory planning method and system based on a random model, and the method comprises the steps: obtaining the trajectory data of traffic participants based on a constructed traffic database, and taking the trajectory data as model training data; training and evaluating a long short-term memory network model based on the model training data to obtain an optimal long short-term memory network model; acquiring trajectory data of the dynamic obstacle in a fixed time interval, and predicting a motion trajectory of the dynamic obstacle in a future time interval based on the optimal long-short-term memory network model; determining potential collision points according to the motion trail of the dynamic obstacle and the trail of the vehicle, and constructing a regional sampling probability set by combining the sampling probabilities of different regions in the sampling region; carrying out local path search by adopting a fast expansion random tree algorithm improved based on a random model to generate a local path; and performing speed planning on the generated local path by adopting a dynamic planning method, and solving to complete a vehicle local trajectory planning task.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving, and particularly relates to a local trajectory planning method and system based on a stochastic model. Background Art

[0002] As a core technology in the field of autonomous driving, path planning methods are developing rapidly, aiming to calculate and optimize the safest and most efficient driving path in real time through the fusion of advanced algorithms and sensors in a complex and changing road environment.

[0003] Existing path planning methods are difficult to quickly and accurately generate safe local paths in the face of complex road environments, especially in the presence of multiple dynamic obstacles. Most existing methods are based on offline calculations or simple heuristic algorithms and are difficult to apply to high-dynamic environments. Especially when there are many obstacles and trajectories are intertwined in the environment, the path solution becomes more complex and it is impossible to achieve a quick solution for local paths. The widely used A* * algorithm, Dijkstra algorithm, and Rapidly-Exploring Random Trees (RRT) algorithm in existing methods can be applied to static environments, but they are adapted to complex planning tasks for dynamic environments. Among them, graph search-based algorithms (such as A* * algorithm and Dijkstra algorithm) can find the optimal path, but their computational complexity is high and it is difficult to meet the real-time requirements. In addition, the Long Short-Term Memory network (LSTM), as a special Recurrent Neural Network (RNN), can perform well when processing time series data and is thus widely applicable to fields such as trajectory prediction and motion planning. However, relying solely on the Long Short-Term Memory network (LSTM) for path planning is difficult to achieve a quick and effective path solution in complex dynamic environments. Summary of the Invention

[0004] The present invention aims to provide a local trajectory planning method and system based on a stochastic model. By combining the Long Short-Term Memory network model and the Rapidly-Exploring Random Trees algorithm improved based on the stochastic model, and combining the planning strategy of spatio-temporal separation of path and speed, the complexity of the path planning problem in complex dynamic scenarios is reduced, the solution speed of path planning is improved, and the real-time performance of the algorithm is enhanced. At the same time, it can not only accurately predict the trajectories of dynamic obstacles but also achieve fast local path planning in complex road environments.

[0005] A local trajectory planning method based on a stochastic model includes the following steps: Based on the constructed traffic database, obtain the trajectory data of traffic participants as model training data; Based on the model training data, use k the k-fold cross-validation mechanism to train and evaluate the Long Short-Term Memory network model to obtain the optimal Long Short-Term Memory network model; Obtain the trajectory data of dynamic obstacles within a fixed time interval, and predict the motion trajectory of dynamic obstacles in the future time interval based on the optimal long short-term memory network model; Determine potential collision points according to the motion trajectories of dynamic obstacles and the vehicle's own trajectory, and construct a regional sampling probability set by combining the sampling probabilities of different regions within the sampling area; Based on the regional sampling probability set, use the rapidly-exploring random tree algorithm improved by a randomness model to perform local path search and generate a local path; Use the dynamic programming method to perform speed planning on the generated local path, and solve to complete the vehicle local trajectory planning task.

[0006] By combining the long short-term memory network model and the rapidly-exploring random tree algorithm improved by a randomness model, and combining the planning strategy of spatio-temporal separation of path and speed, the complexity of the path planning problem in complex dynamic scenarios is reduced, the solution speed of path planning is improved, and the real-time performance of the algorithm is enhanced; at the same time, it can not only accurately predict the trajectory of dynamic obstacles, but also achieve fast local path planning in complex road environments.

[0007] Furthermore, the process of obtaining the trajectory data of traffic participants as model training data based on the constructed traffic database specifically includes the following steps: Extract the trajectory data of traffic participants based on the constructed traffic database; Preprocess the trajectory data of traffic participants; Select features from the preprocessed trajectory data of traffic participants and use it as model training data.

[0008] Furthermore, based on the model training data, using k The process of training and evaluating the long short-term memory network model using the k-fold cross-validation mechanism to obtain the optimal long short-term memory network model specifically includes the following steps: Based on the model training data, divide it into multiple groups of matching training sets and validation sets; Sequentially input multiple training sets into the long short-term memory network model, and train to obtain the root mean square error of the corresponding validation set; Combine all the obtained root mean square errors to construct k The k-fold cross-validation mechanism to evaluate the performance of the long short-term memory network model and obtain k The k-fold cross-validation value; Repeat training multiple times, and select the long short-term memory network model corresponding to the smallest k k-fold cross-validation value as the optimal long short-term memory network model.

[0009] Further, the process of obtaining the trajectory data of dynamic obstacles within a fixed time interval and predicting the motion trajectory of dynamic obstacles within a future time interval based on the optimal long short-term memory network model specifically includes the following steps: Obtain the trajectory data of multiple dynamic obstacles on the road within a fixed time interval; Select features from the trajectory data of multiple preprocessed dynamic obstacles within a fixed time interval and use it as the trajectory data; Input the trajectory data of multiple dynamic obstacles within a fixed time interval after feature selection into the optimal long short-term memory network model to train and output the motion trajectory of dynamic obstacles within a future time interval.

[0010] Further, the process of determining potential collision points based on the motion trajectories of dynamic obstacles and the vehicle's own trajectory and constructing a regional sampling probability set in combination with the sampling probabilities of different regions within the sampling area specifically includes the following steps: Based on the motion trajectories of multiple dynamic obstacles and the vehicle's own trajectory, predict and calculate the distance between the vehicle and each dynamic obstacle, and solve to obtain the minimum collision time point; Based on the minimum collision time point, determine the specific location where the collision occurs and mark it as the potential collision point of the vehicle; Based on the potential collision point of the vehicle, determine the sampling area and obtain the initial sampling probability within the sampling area; Update the sampling probability within the single sampling area formed by the trajectory of a single dynamic obstacle; Update the sampling probability within the intersection area formed by the trajectories of multiple dynamic obstacles; Update the sampling probability within the occupied area based on the area not occupied by the trajectory of the dynamic obstacle; Combine the sampling probabilities of the single sampling area, the intersection area, and the unoccupied area to construct a regional sampling probability set.

[0011] Further, based on the regional sampling probability set, the process of performing local path search using an improved rapidly-exploring random tree algorithm to generate a local path specifically includes the following steps: Based on the sampling area, define a starting point and a target point, and use the starting point as the root node of the initialized random tree; Generate a random point in the configuration space based on the regional sampling probability set; Based on the random tree, find the node closest to the random point as the neighboring point; Calculate the distance from the neighboring point to the random point and expand a step length in this direction to generate a new node, such that the new node satisfies the maximum step length limit; Perform collision detection on the path from the neighboring point to the new node to ensure that the path meets the collision-free requirement; Perform distance detection on the distance from the target point to the new node, such that the distance meets a preset distance threshold; Repeat the above steps until the maximum number of iterations is reached, then backtrack from the target point to the starting point to generate a complete local path.

[0012] Furthermore, use the dynamic programming method to perform speed planning on the generated local path. The process of solving and completing the vehicle local trajectory planning task specifically includes the following steps: Smooth the generated local path; Construct an ST graph and mark the areas occupied by the trajectories of dynamic obstacles; Based on the areas occupied by the trajectories of dynamic obstacles, use the dynamic programming method to perform speed planning on the local path, solve to obtain the ST path curve with the minimum cost, and complete the vehicle local trajectory planning task.

[0013] A system for a local trajectory planning method based on a stochastic model, including: A data acquisition module, which is used to obtain the trajectory data of traffic participants as model training data based on the constructed traffic database; A model construction module, which is used to train and evaluate a long short-term memory network model using k the k-fold cross-validation mechanism based on the model training data to obtain the best long short-term memory network model; A trajectory prediction module, which is used to obtain the trajectory data of dynamic obstacles within a fixed time interval and predict the motion trajectory of dynamic obstacles within a future time interval based on the best long short-term memory network model; A probability calculation module, which is used to determine potential collision points based on the motion trajectories of dynamic obstacles and the vehicle's own trajectory, and construct a regional sampling probability set in combination with the sampling probabilities of different regions within the sampling area; A local path generation module, which is used to perform local path search using an improved rapidly-exploring random tree algorithm based on the regional sampling probability set to generate a local path; A local trajectory planning module, which is used to perform speed planning on the generated local path using the dynamic programming method to solve and complete the vehicle local trajectory planning task.

[0014] An electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the above-mentioned method is implemented.

[0015] A computer-readable storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method is implemented.

[0016] The beneficial effects of the present invention are: The present invention combines a long short-term memory network model and a rapidly-exploring random tree algorithm improved based on a stochastic model, and combines a path planning strategy with spatio-temporal separation of path and speed, reducing the complexity of path planning problems in complex dynamic scenarios, improving the solution speed of path planning, and enhancing the real-time performance of the algorithm. At the same time, it can not only accurately predict the trajectories of dynamic obstacles but also achieve fast local path planning in complex road environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flowchart of the present invention; Figure 2 is a method flowchart of the present invention; Figure 3 is a schematic structural diagram of the system of the present invention; Figure 4 is a schematic structural diagram of a computer device in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0019] It should be noted that the following describes various aspects of embodiments within the scope of the appended claims. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on this disclosure, those skilled in the art should understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement a device and / or practice a method. In addition, this device can be implemented and this method can be practiced using other structures and / or functions in addition to one or more of the aspects described herein.

[0020] In addition, in the following description, specific details are provided to facilitate a thorough understanding of the examples. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0021] Embodiment 1 Figure 1Shown is a local trajectory planning method based on a stochastic model. By combining a long short-term memory network model and a rapidly-exploring random tree algorithm improved based on the stochastic model, and integrating a planning strategy of spatio-temporal separation of path and speed, the complexity of the path planning problem in complex dynamic scenarios is reduced, the solution speed of path planning is improved, and the real-time performance of the algorithm is enhanced. At the same time, it can not only accurately predict the trajectories of dynamic obstacles but also achieve fast local path planning in complex road environments. The specific steps are as follows: S1: Based on the constructed traffic database, obtain the trajectory data of traffic participants as model training data; S11: Extract the trajectory data of traffic participants based on the constructed traffic database; S12: Preprocess the trajectory data of traffic participants; In this embodiment, preprocessing the trajectory data of traffic participants includes missing value processing, outlier processing, and normalization processing to ensure data quality.

[0022] S13: Select features from the preprocessed trajectory data of traffic participants and use it as model training data.

[0023] Among them, the expression for feature selection is: ; In the formula, represents the data feature of the th dynamic obstacle at a certain moment; represents the th road information feature; represents the th category feature value of the dynamic obstacle; represents the th coordinate of the dynamic obstacle in the X-axis direction; represents the th coordinate of the dynamic obstacle in the Y-axis direction; represents the time; represents the th speed of the dynamic obstacle; represents the th acceleration of the dynamic obstacle; represents the th jerk of the dynamic obstacle.

[0024] S2: Based on the model training data, use the k k-fold cross-validation mechanism to train and evaluate the long short-term memory network model to obtain the optimal long short-term memory network model; S21: Based on the model training data, divide it into multiple groups of matching training sets and validation sets; In this embodiment, the model training data is divided into k subsets, and one subset is sequentially selected as the validation set, and the remaining k -1 subsets are used as the corresponding training sets to form multiple sets of matching training sets and validation sets.

[0025] S22: Input the multiple training sets into the long short-term memory network model in sequence, and train to obtain the root mean square error of the corresponding validation set ; S23: Combine all the obtained root mean square errors to construct k k-fold cross-validation mechanism, evaluate the performance of the long short-term memory network model, and obtain k k-fold cross-validation value ; Among them, k the expression of the k-fold cross-validation mechanism is: ; In the formula, represents the performance evaluation index of the long short-term memory network model; represents the root mean square error corresponding to the th data; , represents the total number of data groups.

[0026] S24: Repeat training multiple times, and select the long short-term memory network model corresponding to the smallest k k-fold cross-validation value as the best long short-term memory network model.

[0027] S3: Obtain the trajectory data of dynamic obstacles within a fixed time interval, and predict the movement trajectory of dynamic obstacles within a future time interval based on the best long short-term memory network model; S31: Obtain the trajectory data of multiple dynamic obstacles on the road within a fixed time interval; S32: Select features from the trajectory data of multiple dynamic obstacles within a fixed time interval after preprocessing, and use it as the trajectory data; It should be noted that the preprocessing and feature selection methods are the same as those for processing the model training data.

[0028] S33: Input the trajectory data of multiple dynamic obstacles within a fixed time interval after feature selection into the best long short-term memory network model, and train to output the movement trajectory of dynamic obstacles within a future time interval.

[0029] Among them, within the future time interval, the movement trajectory of any group of dynamic obstacles is expressed as: ; ; In the formula, represents the th coordinate vector of the dynamic obstacle with respect to time ; represents the th trajectory fitting function of the dynamic obstacle with respect to time ; represents the area occupied by the trajectory of the th dynamic obstacle; wherein, in the case of not being affected by environmental factors, the trajectory of the dynamic obstacle within a future time interval is represented as: ; In the formula, represents the coordinate vector of the dynamic obstacle with respect to time without being affected by environmental factors; represents the trajectory fitting function of the dynamic obstacle with respect to time without being affected by environmental factors.

[0030] S4: Determine potential collision points according to the motion trajectories of the dynamic obstacles and the vehicle's own trajectory, and construct a regional sampling probability set in combination with the sampling probabilities of different regions within the sampling area; S41: Based on the motion trajectories of multiple dynamic obstacles and the vehicle's own trajectory, predict and calculate the distance between the vehicle and each dynamic obstacle, and solve to obtain the minimum collision time point; wherein, the calculation expression for the distance between the vehicle and the dynamic obstacle is: ; In the formula, represents the distance between the vehicle and the dynamic obstacle with respect to time ; represents the collision threshold; S42: Based on the minimum collision time point, determine the specific position where the dynamic obstacle collision occurs, and mark it as the potential collision point of the vehicle; S43: Based on the potential collision point of the vehicle, determine the sampling area and obtain the initial sampling probability within the sampling area; wherein, the calculation expression for the initial sampling probability is: ; In the formula, represents the initial sampling probability within the sampling area ; represents the sampling area; S44: Update the sampling probability within the single sampling area based on the single sampling area formed by the trajectory of a single dynamic obstacle; In this embodiment, each dynamic obstacle in the sampling area has a corresponding trajectory area and the maximum potential collision point ; in the trajectory area , the sampling probability model shows the characteristic of radiating from the maximum potential collision point to the surrounding, that is, the probability of being sampled at the position closer to the maximum probability collision point is smaller. That is, the sampling probability in a single sampling area is updated using the reverse Gaussian distribution function, and its update expression is: ; In the formula, represents the sampling probability of the th dynamic obstacle in the updated single sampling area with respect to the collision point; represents the normalization coefficient corresponding to the th dynamic obstacle; represents the standard deviation; represents the coordinate of the potential collision point in the X-axis direction; represents the coordinate of the th dynamic obstacle in the Y-axis direction; S45: Update the sampling probability in the intersection area based on the intersection area formed by the trajectories of multiple dynamic obstacles; In this embodiment, each dynamic obstacle moves independently, without interfering with each other in the space-time dimension, and there is an intersection between the trajectory areas in the space dimension, forming an intersection area. When an intersection area is formed by the trajectory intersections of multiple dynamic obstacles, the sampling probability in the intersection area is updated using the probability weighting of the trajectory areas of each dynamic obstacle, and its update expression is: + + ; In the formula, represents the sampling probability in the intersection area of the trajectory area of the th dynamic obstacle, the trajectory area of the th dynamic obstacle, and the trajectory area of the th dynamic obstacle; , , respectively represent the weight coefficients sampled in the trajectory area , the trajectory area , and the trajectory area ; , , respectively represent the th dynamic obstacle, the The sampling probability of the th dynamic obstacle; S46: Update the sampling probability in the occupied area based on the area unoccupied by the trajectory of the dynamic obstacle; Among them, the update expression of the sampling probability in the unoccupied area is: ; In the formula, represents the sampling probability in the area unoccupied by the dynamic obstacle; represents the area of the area unoccupied by the dynamic obstacle; represents the sum of the sampling probabilities of the areas occupied by the trajectory of the dynamic obstacle.

[0031] S47: Combine the sampling probabilities of the single sampling area, the intersection area, and the unoccupied area to construct a set of area sampling probabilities .

[0032] S5: Based on the set of area sampling probabilities, use the rapidly-exploring random tree algorithm improved based on the randomness model to perform local path search and generate a local path; S51: Define the starting point and the target point based on the sampling area, and use the starting point as the root node of the initialized random tree; S52: Generate a random point in the configuration space based on the set of area sampling probabilities; S53: Find the node closest to the random point as the neighboring point based on the random tree; S54: Calculate the distance from the neighboring point to the random point , and expand a step length in this direction to generate a new node , so that the new node meets the maximum step length limit; In this embodiment, when the maximum step length limit is met, the random point is used as the new node ; when the maximum step length limit is not met, the neighboring point moves a step length along the direction towards the random point to generate a new node ; S55: Perform collision detection on the path from the neighboring point to the new node so that the path meets the collision-free requirement; In this embodiment, when the collision-free requirement is not met, a new random point is regenerated 。

[0033] S56: Detect the distance from the target point to the new node such that the distance meets a preset distance threshold; In this embodiment, when the distance does not meet the preset distance threshold, a random point is regenerated 。

[0034] S57: Repeat the above steps until the maximum number of iterations is reached, and backtrack from the target point to the starting point to generate a complete local path.

[0035] S6: Use the dynamic programming method to perform speed planning on the generated local path to complete the vehicle local trajectory planning task.

[0036] S61: Smooth the generated local path; In this embodiment, a Bezier curve is used to smooth the generated local path, and its expression is: ; In the formula, represents the local path with respect to time after being smoothed by the Bezier curve; represents time, ; represents the serial number of the path point of the generated local path, , represents the total number of path points of the generated local path; represents the path point of the generated local path.

[0037] S62: Construct an ST graph and mark the area occupied by the dynamic obstacle trajectory; S63: Based on the area occupied by the dynamic obstacle trajectory, use the dynamic programming method to perform speed planning on the local path to solve and obtain the ST path curve with the minimum cost, and complete the vehicle local trajectory planning task; Among them, the cost function expression of dynamic programming is: ; In the formula, represents the value of the cost function of dynamic programming; represents the driving time between adjacent path points; represents the path error; represents the th path point speed; represents the th path point speed; represents the The acceleration of a path point; Indicates the desired speed; , , , , Indicate the weight coefficients of the terms in the cost function.

[0038] In this embodiment, when the solution fails, it means that the generated local vehicle trajectory path is not feasible, and it is necessary to regenerate the local path and perform speed planning.

[0039] Embodiment 2 As Figure 2 shown, this embodiment provides a local trajectory planning method based on a stochastic model, which specifically includes the following steps: T1: Based on the constructed traffic database, obtain the trajectory data of traffic participants as model training data; T11: Extract the trajectory data of traffic participants based on the constructed traffic database; T12: Preprocess the trajectory data of traffic participants; T13: Select features from the preprocessed trajectory data of traffic participants and use it as model training data.

[0040] T2: Based on the model training data, use k k-fold cross-validation mechanism to train and evaluate the long short-term memory network model to obtain the best long short-term memory network model; T21: Based on the model training data, divide it into multiple sets of matching training sets and validation sets; T22: Sequentially input multiple training sets into the long short-term memory network LSTM model, and train to obtain the root mean square error of the corresponding validation set; T23: Combine all the obtained root mean square errors to construct k k-fold cross-validation mechanism, evaluate the performance of the long short-term memory network LSTM model, and obtain k k-fold cross-validation value; T24: Repeat training multiple times and select the long short-term memory network LSTM model corresponding to the smallest k k-fold cross-validation value as the best long short-term memory network LSTM model.

[0041] T3: Obtain the trajectory data of dynamic obstacles within a fixed time interval, and predict the motion trajectory of dynamic obstacles within a future time interval based on the best long short-term memory network model; T31: Real-time obtain the trajectory data of multiple dynamic obstacles on the road within a fixed time interval; T32: Select features from the trajectory data of multiple preprocessed dynamic obstacles within a fixed time interval, and use it as the trajectory data; T33: Input the trajectory data of multiple dynamic obstacles with selected features within a fixed time interval into the optimal long short-term memory network model to train and output the motion trajectories of dynamic obstacles in the future time interval.

[0042] T4: Determine potential collision points based on the motion trajectories of dynamic obstacles and the vehicle's own trajectory, and construct a regional sampling probability set in combination with the sampling probabilities of different regions within the sampling area; T41: Based on the motion trajectories of multiple dynamic obstacles and the vehicle's own trajectory, predict and calculate the distance between the vehicle and each dynamic obstacle, and solve to obtain the minimum collision time point; T42: Based on the minimum collision time point, determine the specific position where the dynamic obstacle collision occurs and mark it as the vehicle's potential collision point; T43: Based on the vehicle's potential collision point, determine the sampling area and obtain the initial sampling probability within the sampling area; T44: Update the sampling probability within the single sampling area based on the single sampling area formed by the trajectory of a single dynamic obstacle; T45: Update the sampling probability within the intersection area based on the intersection area formed by the trajectories of multiple dynamic obstacles; T46: Update the sampling probability within the occupied area based on the area not occupied by the trajectory of the dynamic obstacle; T47: Combine the sampling probabilities of the single sampling area, the intersection area, and the unoccupied area to construct a regional sampling probability set 。

[0043] T5: Based on the regional sampling probability set, use the rapidly-exploring random tree RRT algorithm improved based on the stochastic model to perform local path search and generate a local path; T51: Define the starting point and the target point based on the sampling area, and use the starting point as the root node for initializing the random tree; T52: Generate a random point in the configuration space based on the regional sampling probability set; T53: Based on the random tree, find the node closest to the random point as the neighboring point ; T54: Calculate the distance from the neighboring point to the random point , and expand a step length in this direction to generate a new node , determine whether the new node meets the maximum step - length limit; T541: When the new node meets the maximum step - length limit, make the random point serve as the new node , and execute T55; T542: When it does not meet the maximum step - length limit, then make the neighboring point move one step along the direction towards the random point to generate a new node , and execute T55; T55: Perform collision detection on the path from the neighboring point to the new node , and determine whether the path meets the collision - free requirement; T551: When the path meets the collision - free requirement, that is, there is no collision on the path, execute T56; T552: When the path does not meet the collision - free requirement, that is, there is a collision on the path, return to T52, that is, regenerate the random point .

[0044] T56: Perform distance detection on the distance from the target point to the new node , and determine whether the distance meets the preset distance threshold; T561: When the distance meets the preset distance threshold, complete the path planning for this point and return the path; T562: When the distance does not meet the preset distance threshold, then return to T52, that is, regenerate the random point .

[0045] T57: Repeat the above steps until the maximum number of iterations is reached or the target point is successfully found , backtrack from the target point to the starting point , and generate a complete local path.

[0046] T6: Use the dynamic programming method to perform speed planning on the generated local path to solve and complete the vehicle local trajectory planning task.

[0047] T61: Based on the generated local path, construct an ST graph and mark the area occupied by the dynamic obstacle trajectory; T62: Determine whether the generated local path occupies the dynamic obstacle trajectory; T621: When it occupies the dynamic obstacle trajectory, use the dynamic programming method to perform speed planning on the local path, solve and determine whether the speed solution is successful; T6211: When the speed solution is successful, complete the vehicle local trajectory planning task; T6212: When the speed solution fails, return T5; T622: When not occupying the dynamic obstacle trajectory, smooth the generated local path and endow it with speed characteristics to complete the vehicle local trajectory planning task; Embodiment 3 Based on the same design concept, as Figure 3 shown, in this embodiment, a local trajectory planning system based on a stochastic model is provided, including a data acquisition module, a model construction module, a trajectory prediction module, a probability calculation module, a local path generation module, and a local trajectory planning module.

[0048] Specifically, the data acquisition module is used to obtain the trajectory data of traffic participants as model training data based on the constructed traffic database; Specifically, the model construction module is used to train and evaluate a long short-term memory network model using k a k-fold cross-validation mechanism based on the model training data to obtain the best long short-term memory network model; Specifically, the trajectory prediction module is used to obtain the trajectory data of dynamic obstacles within a fixed time interval and predict the movement trajectory of dynamic obstacles within a future time interval based on the best long short-term memory network model; Specifically, the probability calculation module is used to determine potential collision points according to the movement trajectory of dynamic obstacles and the vehicle's own trajectory, and construct a regional sampling probability set in combination with the sampling probabilities of different regions within the sampling area; Specifically, the local path generation module is used to perform local path search using an improved rapidly-exploring random tree algorithm based on the regional sampling probability set to generate a local path; Specifically, the local trajectory planning module is used to perform speed planning on the generated local path using dynamic programming to solve and complete the vehicle local trajectory planning task.

[0049] Embodiment 4 Based on the same technical concept, an embodiment of the present application further provides a computer device, including a memory 1 and a processor 2, as Figure 4 shown, the memory 1 stores a computer program, and when the processor 2 executes the computer program, it implements the method described in any one of the above.

[0050] Among them, the memory 1 at least includes one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), magnetic memory, magnetic disk, optical disc, etc. The memory 1 can be an internal storage unit of the local trajectory planning system based on the randomness model in some embodiments, such as a hard disk. The memory 1 can also be an external storage device of the local trajectory planning system based on the randomness model in other embodiments, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. Further, the memory 1 can also include both the internal storage unit and the external storage device of the local trajectory planning system based on the randomness model. The memory 1 can be used not only to store the application software and various types of data installed in the local trajectory planning system based on the randomness model, such as the code of the local trajectory planning system program, etc., but also to temporarily store the data that has been output or will be output.

[0051] The processor 2 can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor or other data processing chips in some embodiments, and is used to run the program code stored in the memory 1 or process data, such as executing the local trajectory planning system program based on the randomness model, etc.

[0052] The disclosed embodiments of the present invention also provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is run by a processor, it executes the steps of the method described in the foregoing method embodiments. Among them, the storage medium can be a volatile or non-volatile computer-readable storage medium.

[0053] The computer program product of the application page content refreshing method provided by the disclosed embodiments of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the steps of the method described in the foregoing method embodiments. For details, please refer to the foregoing method embodiments and will not be elaborated here.

[0054] The disclosed embodiments of the present invention also provide a computer program, which implements any one of the methods in the foregoing embodiments when executed by a processor. The computer program product can be specifically implemented in a manner of hardware, software or a combination thereof. In an optional embodiment, the computer program product is specifically embodied as a computer storage medium. In another optional embodiment, the computer program product is specifically embodied as a software product, such as a Software Development Kit (SDK), etc.

[0055] It is understandable that the same or similar parts in the above embodiments can be referred to each other, and for the content not detailed in some embodiments, reference can be made to the same or similar content in other embodiments.

[0056] It should be noted that in the description of the present invention, terms such as "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "a plurality of" refers to at least two.

[0057] Any process or method description in the flowchart or described in other ways herein can be understood to represent a module, segment, or part of code including one or more executable instructions for implementing a specific logical function or process. And the scope of the preferred embodiments of the present invention includes additional implementations, where the functions can be executed in a substantially simultaneous manner or in a reverse order according to the involved functions, rather than in the order shown or discussed, which should be understood by those skilled in the technical field to which the embodiments of the present invention belong.

[0058] It should be understood that each part of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following well-known technologies in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0059] Those of ordinary skill in the technical field can understand that all or part of the steps carried by the methods in the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0060] In addition, each functional unit in various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically alone, or two or more units can be integrated into a module. The above integrated module can be implemented in the form of hardware or in the form of a software functional module. When the above integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0061] The above-mentioned storage medium can be a read-only memory, a magnetic disk, an optical disk, etc.

[0062] In the description of this specification, the descriptions referring to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0063] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A local trajectory planning method based on a stochastic model, characterized in that: The steps include: Based on the constructed traffic database, the trajectory data of traffic participants are obtained as model training data; Based on the model training data, k The 1-fold cross-validation mechanism is used to train and evaluate the LSTM network model to obtain the best LSTM network model. Obtain the trajectory data of dynamic obstacles in a fixed time interval, and predict the movement trajectory of dynamic obstacles in the future time interval based on the optimal long short-term memory network model; Determine the potential collision point based on the motion trajectory of the dynamic obstacle and the vehicle's own trajectory, and build a regional sampling probability set based on the sampling probabilities of different regions within the sampling area; Based on the regional sampling probability set, a fast expanding random tree algorithm based on the improved randomness model is used to search for local paths and generate local paths; The dynamic programming method is used to plan the speed of the generated local path and solve the task of vehicle local trajectory planning.

2. A local trajectory planning method based on a stochastic model according to claim 1, characterized in that: Based on the constructed traffic database, the process of obtaining the trajectory data of traffic participants as model training data specifically includes the following steps: Based on the constructed traffic database, the trajectory data of traffic participants are extracted; Preprocessing the trajectory data of traffic participants; The preprocessed trajectory data of traffic participants are used for feature selection and used as model training data.

3. A local trajectory planning method based on a stochastic model according to claim 1, characterized in that: Based on the model training data, k The fold-cross validation mechanism is used to train and evaluate the LSTM network model. The process of obtaining the best LSTM network model includes the following steps: Based on the model training data, it is divided into multiple sets of matching training sets and validation sets; Multiple training sets are sequentially input into the long short-term memory network model, and the root mean square error of the corresponding validation set is obtained through training; Combine all the root mean square errors obtained to construct k The performance of the long short-term memory network model was evaluated by using the fold cross-validation mechanism. k Fold cross validation value; Repeat the training multiple times and select the smallest k The LSTM network model corresponding to the fold cross-validation value is taken as the best LSTM network model.

4. A local trajectory planning method based on a stochastic model according to claim 1, characterized in that: The process of obtaining the trajectory data of a dynamic obstacle in a fixed time interval and predicting the trajectory of the dynamic obstacle in a future time interval based on the optimal long short-term memory network model specifically includes the following steps: Obtain trajectory data of multiple dynamic obstacles on the road within a fixed time interval; Feature selection is performed on the pre-processed trajectory data of multiple dynamic obstacles within a fixed time interval, and the feature selection is used as trajectory data; The trajectory data of multiple dynamic obstacles selected by features within a fixed time interval are input into the optimal long short-term memory network model, and the motion trajectory of the dynamic obstacles in the future time interval is trained and output.

5. The local trajectory planning method based on a stochastic model according to claim 1, characterized in that: The process of determining the potential collision point based on the motion trajectory of the dynamic obstacle and the vehicle's own trajectory and constructing the regional sampling probability set based on the sampling probabilities of different regions in the sampling area includes the following steps: Based on the motion trajectories of multiple dynamic obstacles and the vehicle's own trajectory, the distance between the vehicle and each dynamic obstacle is predicted and calculated, and the minimum collision time point is obtained; Based on the minimum collision time point, determine the specific location where the collision occurred and mark it as a potential collision point of the vehicle; Based on the potential collision points of the vehicles, a sampling area is determined, and an initial sampling probability within the sampling area is obtained; Based on a single sampling area formed by the trajectory of a single dynamic obstacle, the sampling probability within the single sampling area is updated; Based on the intersection area formed by the trajectories of multiple dynamic obstacles, the sampling probability in the intersection area is updated; Based on the area not occupied by the trajectory of the dynamic obstacle, the sampling probability in the occupied area is updated; The sampling probabilities of single sampling areas, intersection areas, and unoccupied areas are combined to construct a set of regional sampling probabilities.

6. A local trajectory planning method based on a stochastic model according to claim 1, characterized in that: Based on the regional sampling probability set, the improved fast expanding random tree algorithm is used to search for local paths. The process of generating local paths specifically includes the following steps: Based on the sampling area, define the starting point and target point, and use the starting point as the root node to initialize the random tree; Generate a random point in the configuration space based on the set of region sampling probabilities; Based on the random tree, find the node closest to the random point as the neighboring point; Calculate the distance from the neighboring point to the random point, and extend a step in this direction to generate a new node so that the new node meets the maximum step limit; Perform collision detection on the path from the adjacent point to the new node to ensure that the path meets the collision-free requirement; Perform distance detection on the distance from the target point to the new node so that the distance meets the preset distance threshold; Repeat the above steps until the maximum number of iterations is reached, backtrack from the target point to the starting point, and generate a complete local path.

7. The local trajectory planning method based on a stochastic model according to claim 1, characterized in that: The dynamic programming method is used to plan the speed of the generated local path. The process of solving the task of vehicle local trajectory planning includes the following steps: Smoothing the generated local path; Construct the ST graph and mark the area occupied by the dynamic obstacle trajectory; Based on the area occupied by the dynamic obstacle trajectory, the dynamic programming method is used to plan the speed of the local path, and the ST path curve with the minimum cost is solved to complete the vehicle local trajectory planning task.

8. A system using the local trajectory planning method based on a stochastic model as claimed in claim 1, characterized in that: include: A data acquisition module, which is used to acquire the trajectory data of traffic participants as model training data based on the constructed traffic database; Model building module, which is used to train the model based on data, using k The 1-fold cross-validation mechanism is used to train and evaluate the LSTM network model to obtain the best LSTM network model. The trajectory prediction module is used to obtain the trajectory data of dynamic obstacles in a fixed time interval and predict the movement trajectory of dynamic obstacles in the future time interval based on the optimal long short-term memory network model; The probability calculation module is used to determine the potential collision point according to the motion trajectory of the dynamic obstacle and the vehicle's own trajectory, and to construct a regional sampling probability set by combining the sampling probabilities of different regions within the sampling area; A local path generation module is used to search for local paths based on a regional sampling probability set and to generate local paths using an improved fast-expanding random tree algorithm; The local trajectory planning module is used to use a dynamic planning method to perform speed planning on the generated local path and solve the vehicle local trajectory planning task.

9. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images