CN118013868A - Vehicle state prediction method and device - Google Patents
Vehicle state prediction method and device Download PDFInfo
- Publication number
- CN118013868A CN118013868A CN202410425010.3A CN202410425010A CN118013868A CN 118013868 A CN118013868 A CN 118013868A CN 202410425010 A CN202410425010 A CN 202410425010A CN 118013868 A CN118013868 A CN 118013868A
- Authority
- CN
- China
- Prior art keywords
- information
- vehicle
- lane
- prediction
- historical
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 82
- 230000007246 mechanism Effects 0.000 claims abstract description 35
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims description 66
- 238000003860 storage Methods 0.000 claims description 34
- 230000001133 acceleration Effects 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000004088 simulation Methods 0.000 claims description 7
- 239000013598 vector Substances 0.000 description 28
- 230000006870 function Effects 0.000 description 24
- 230000008569 process Effects 0.000 description 23
- 230000006399 behavior Effects 0.000 description 20
- 230000003993 interaction Effects 0.000 description 16
- 238000004590 computer program Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 10
- 230000014509 gene expression Effects 0.000 description 9
- 238000012545 processing Methods 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000002776 aggregation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000003313 weakening effect Effects 0.000 description 1
Landscapes
- Traffic Control Systems (AREA)
Abstract
The application discloses a vehicle state prediction method and device, and relates to the technical field of software. The method of the application comprises the following steps: acquiring running state information, vehicle road relation information and historical decision information of a vehicle, wherein the vehicle road relation information is used for representing the position relation between the vehicle and a road, and the vehicle road relation information is calculated based on the running state information; the historical decision information is used for representing vehicle road relation information of the vehicle in a historical target period; and based on the driving state information, the vehicle road relation information and the historical decision information, performing prediction operation by using a vehicle state prediction model to obtain a prediction result, wherein the vehicle state prediction model is trained by a attention mechanism algorithm through the historical behavior state information, the historical vehicle road relation information and the historical decision information. The application is used for realizing the prediction function of the vehicle state.
Description
Technical Field
The present application relates to the field of software technologies, and in particular, to a method and an apparatus for predicting a vehicle state.
Background
Along with the development of technology, the prediction of vehicle behavior is also becoming a main development direction in intelligent transportation and simulation scenes.
At present, in the existing prediction process of the vehicle state, the prediction is generally performed by combining a preset logic expression on the basis of combining the speed and the acceleration of the vehicle, that is, after the speed and the acceleration of the vehicle are determined, the running state of the vehicle is determined by calculating and analyzing the logic expression. However, in practical applications, the parameters set in these logic expressions are generally set in combination with experience, which results in that the prediction result of the vehicle state is affected once the experience of the expert who sets the expression is insufficient, so that the accuracy of the prediction result of the vehicle state is low.
Disclosure of Invention
The embodiment of the application provides a vehicle state prediction method and device, and mainly aims to realize the vehicle state prediction method so as to solve the problem that the accuracy of the existing vehicle state prediction result is low.
In order to solve the technical problems, the embodiment of the application provides the following technical scheme:
In a first aspect, the present application provides a method for predicting a vehicle state, the method comprising:
Acquiring running state information, vehicle road relation information and historical decision information of a vehicle, wherein the vehicle road relation information is used for representing the position relation between the vehicle and a road, and the vehicle road relation information is calculated based on the running state information; the historical decision information is used for representing vehicle road relation information of the vehicle in a historical target period;
And based on the driving state information, the vehicle road relation information and the historical decision information, performing prediction operation by using a vehicle state prediction model to obtain a prediction result, wherein the vehicle state prediction model is trained by a attention mechanism algorithm through the historical behavior state information, the historical vehicle road relation information and the historical decision information.
In a second aspect, the present application also provides a device for predicting a vehicle state, the device comprising:
A setting unit configured to set a model frame of the vehicle state prediction model; the model framework comprises a first input module, a second input module, a third input module and a prediction execution module; the first input module is used for inputting vehicle road relation information, the second input module is used for inputting driving state information, and the third input module is used for inputting historical decision information; the prediction execution module is used for inputting the input results of the first input module, the second input module and the third input module and executing prediction;
the training set acquisition unit is used for acquiring a target training set, wherein a training sample in the target training set comprises historical behavior state information, historical vehicle road information, historical decision information and corresponding actual running state information;
and the training unit is used for training the target training set and the model frame based on the attention mechanism algorithm to generate the vehicle state prediction model.
In a third aspect, the present application further provides a storage medium, where the storage medium includes a stored program, where the program, when executed, controls a device where the storage medium is located to execute the method for predicting a vehicle state according to the first aspect.
In a fourth aspect, the present application also provides a prediction apparatus of a vehicle state, the apparatus including a storage medium; and one or more processors coupled to the storage medium, the processors configured to execute the program instructions stored in the storage medium; the program instructions, when executed, perform the method of predicting a vehicle state of any one of the first aspects.
By means of the technical scheme, the technical scheme provided by the application has at least the following advantages:
The application provides a prediction method and a device for vehicle state, which can firstly acquire running state information of a vehicle, vehicle road relation information and historical decision information, wherein the vehicle road relation information is used for representing the position relation between the vehicle and a road, and the vehicle road relation information is calculated based on the running state information; the historical decision information is used for representing vehicle road relation information of the vehicle in a historical target period; and then, based on the driving state information, the vehicle road relation information and the historical decision information, a prediction operation is performed by using a vehicle state prediction model to obtain a prediction result, wherein the vehicle state prediction model is trained by using the historical behavior state information, the historical vehicle road relation information and the historical decision information through an attention mechanism algorithm, so that a prediction function of the vehicle state is realized. Compared with the prior art, in the prediction process of the vehicle state, the prediction is performed based on the vehicle state prediction model which is trained based on the historical behavior state information, the historical vehicle road relation information and the historical decision information, that is, the vehicle state prediction model is actually trained on the basis of the condition of vehicle historical driving as a sample, unlike the prior art which is based on manually set expression parameters, the model does not need to manually set parameters, so that the threshold requirement on setting experience in the manual setting process is avoided, the problem that the accuracy of a follow-up prediction result is influenced due to the fact that the set parameters are inaccurate due to insufficient experience is avoided, and the accuracy of the prediction result can be improved. Meanwhile, in the application, the method for predicting the vehicle state can be used as a prediction evidence from three aspects of vehicle running state information, vehicle road relation and historical decision information, and compared with the prior art for predicting the vehicle running state only through acceleration, speed and the like of the vehicle, the method increases the consideration of the two aspects of the interrelationship between the vehicle and the road and the vehicle historical decision, thereby comprehensively analyzing and predicting the finally predicted vehicle state from more aspects of angles, and further improving the accuracy of a prediction result.
The foregoing description is only an overview of the present application, and is intended to be implemented in accordance with the teachings of the present application in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present application more readily apparent.
Drawings
The above, as well as additional purposes, features, and advantages of exemplary embodiments of the present application will become readily apparent from the following detailed description when read in conjunction with the accompanying drawings. In the drawings, wherein like or corresponding reference numerals indicate like or corresponding parts, there are shown by way of illustration, and not limitation, several embodiments of the application, in which:
FIG. 1 shows a flow chart of a method for predicting a vehicle state according to an embodiment of the present application;
FIG. 2 is a flowchart of another method for predicting a vehicle state according to an embodiment of the present application;
Fig. 3 is a schematic structural diagram of a model framework when a prediction method of a vehicle state is executed according to an embodiment of the present application;
FIG. 4 is a block diagram showing the constitution of a prediction apparatus for vehicle state according to an embodiment of the present application;
fig. 5 shows a block diagram of another vehicle state prediction apparatus according to an embodiment of the present application.
Detailed Description
Exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the application to those skilled in the art.
It is noted that unless otherwise indicated, technical or scientific terms used herein should be given the ordinary meaning as understood by one of ordinary skill in the art to which this application belongs.
The embodiment of the application provides a flow chart of a vehicle state prediction method, as shown in fig. 1, comprising the following steps:
101. And acquiring running state information, vehicle road relation information and historical decision information of the vehicle.
The vehicle road relation information is used for representing the position relation between a vehicle and a road, and is calculated based on the running state information; the historical decision information is used for representing vehicle road relation information of the vehicle in a historical target period.
In this embodiment, the driving status information may be understood as information related to driving related to the driving of the vehicle, such as speed, acceleration, destination of the driving of the vehicle, origin of the driving of the vehicle, etc., and of course, in practical application, the user may choose from these and other related information based on the need, which is not limited herein. In addition, the vehicle road relation information in the present embodiment may be understood as information capable of expressing a relation between a vehicle and a road, such as a road section on which the vehicle is traveling, a specific position of the vehicle in the road, and the like, which are calculated based on the traveling state information of the vehicle, and which are calculated based on the traveling state information of the vehicle in combination with the road-related information. In addition, the history decision information in the present embodiment may be understood as a relationship between the road and the vehicle during the previous travel of the vehicle on the same road, or may be understood as a relationship between the vehicle and the road at a previous time.
It should be noted that, the driving state information, the vehicle road relation information and the historical decision information may be acquired and calculated through a vehicle terminal of the vehicle, or may be acquired through interaction between the vehicle terminal and the remote cloud, where the method for acquiring the data includes, but is not limited to, any one of the above methods, and the user may select the method by himself or herself based on the need.
102. And based on the driving state information, the vehicle road relation information and the historical decision information, performing prediction operation by using a vehicle state prediction model to obtain a prediction result.
The vehicle state prediction model is trained through an attention mechanism algorithm through historical behavior state information, historical vehicle road relation information and historical decision information.
After the driving state information, the vehicle road relation information and the historical decision information are obtained in the foregoing steps, in this embodiment, the prediction operation may be performed based on the three items of data through a pre-trained prediction model, where the prediction model in this embodiment, that is, the vehicle state prediction model is obtained by training through an attention mechanism algorithm based on the historical behavior state information, the historical vehicle road relation information and the historical decision information, and because a sample in the training process is the historical data of the three items of data obtained in the foregoing steps, the prediction based on the model can accurately analyze the current driving situation of the vehicle and obtain a corresponding prediction result, and compared with the prediction based on the conventional expression and the related parameters, the prediction model can analyze the characteristics of the driving of the vehicle based on the sample, and based on the prediction based on the current collected data (that is, the driving state information, the vehicle road relation information and the historical decision information), so that the prediction result is more scientific and accurate, and the problem that the accuracy of the prediction result is affected due to the lower accuracy of the manually set expression parameters is avoided.
In addition, in the present embodiment, the prediction model is trained based on an attention mechanism algorithm, which can be understood as a working mode simulating human attention, and the attention to important information and the omission of non-important information are realized through the weighting processing of input data. The algorithm generally includes three main steps: 1. calculating attention weight: from the features of the input data, the attention weight corresponding to each feature is calculated to determine its importance. This is typically done by using neural networks to learn and predict the attention weights; 2. and (3) weighting: multiplying the input data with the attention weight to obtain a weighted data representation, emphasizing important features and weakening non-important features; 3. attention aggregation: and carrying out aggregation processing on the weighted data representation to obtain the final attention output. This typically involves summing or averaging the weighted data to arrive at a final attention representation; through the steps, the attention mechanism algorithm can realize attention and neglect on input data, so that the perception capability of the model on important information is improved. That is, by calculating several important input item data in the above manner, a "score" after the last focused "attention" is focused can be obtained, and the "score" can represent the calculation result in the case of more "attention" to some data. Therefore, by performing model training through the attention mechanism algorithm, when more data items are input, the relatively more accurate prediction result can be obtained by combining the attention characteristic, so that the interference of some less important data items on the whole prediction result can be avoided, the accuracy of the prediction result can be improved, the more important selection of various data items can be ensured to perform attention under the intelligent traffic scene based on the method of the embodiment, the running condition of the vehicle is predicted, the influence of other less important parameters on the prediction result is avoided, and the more accurate prediction on the running of the vehicle is realized.
Based on this, the present embodiment provides a prediction method of a vehicle state. Compared with the prior art, in the prediction process of the vehicle state, the prediction is performed based on the vehicle state prediction model which is trained based on the historical behavior state information, the historical vehicle road relation information and the historical decision information, that is, the vehicle state prediction model is actually trained on the basis of the condition of vehicle historical driving as a sample, unlike the prior art which is based on manually set expression parameters, the model does not need to manually set parameters, so that the threshold requirement on setting experience in the manual setting process is avoided, the problem that the accuracy of a follow-up prediction result is influenced due to the fact that the set parameters are inaccurate due to insufficient experience is avoided, and the accuracy of the prediction result can be improved. Meanwhile, in the application, the method for predicting the vehicle state can be used as a prediction evidence from three aspects of vehicle running state information, vehicle road relation and historical decision information, and compared with the prior art for predicting the vehicle running state only through acceleration, speed and the like of the vehicle, the method increases the consideration of the two aspects of the interrelationship between the vehicle and the road and the vehicle historical decision, thereby comprehensively analyzing and predicting the finally predicted vehicle state from more aspects of angles, and further improving the accuracy of a prediction result.
Further, as a further description and refinement of the foregoing embodiments, the present application also provides a prediction of a vehicle state, as specifically shown in fig. 2:
201. setting a model framework of the vehicle state prediction model.
The model framework comprises a first input module, a second input module, a third input module and a prediction execution module.
The first input module is used for inputting vehicle road relation information, the second input module is used for inputting driving state information, and the third input module is used for inputting historical decision information; the prediction execution module is used for inputting the input results of the first input module, the second input module and the third input module and executing prediction.
In this embodiment, the model frame of the vehicle state prediction model is specifically divided into two parts, wherein one part is a module specifically used for inputting each information, and the other part is a module specifically used for performing prediction, that is, the first input module, the second input module and the third input module are all the former part of the model frame, and the prediction execution module is the latter part. Of course, in practical application, the input data may need to be encoded to a certain extent, that is, different input data needs to be converted into encoded information that can be conveniently identified in the subsequent prediction, so after the first input module, the second input module and the third input module, a corresponding encoding module needs to be added on the basis of each input module, so as to convert the received vehicle road relation information, driving state information and historical decision information into corresponding encoded information respectively.
In addition, in practical application, the schematic diagram of the structure of the model frame may specifically refer to fig. 3, where the vehicle road relationship information in this embodiment may not be data obtained by preprocessing alone, but may specifically be data determined by combining with driving status information of a vehicle based on road information, and in this case, the first input module in the model frame of this embodiment may specifically further include a lane information module and a lane interaction module in fig. 3, where the lane information module is configured to receive information related to a lane, and the vehicle road interaction module is configured to analyze and calculate, based on the driving status information transmitted from the second input module, the relationship between the vehicle and the road, that is, the vehicle road relationship information. Of course, in the present embodiment, the vehicle running state information may actually include the relationship between vehicles, and then, in the model frame in the present embodiment, the information may be analyzed and calculated by the running information module and the vehicle interaction module included in the second input module. The driving information module is configured to receive driving status information of the vehicle, and in fig. 3, the third input module is actually configured of a history decision module and is configured to receive history decision information. Of course, as can be seen from the foregoing description, since prediction is not directly performed based on the vehicle road relation information, the driving state information, and the history decision information in the process of actually performing prediction, but it is necessary to convert these information into corresponding encoded information, a corresponding encoding module is also provided in fig. 3 so as to convert information received from the outside into corresponding encoded information.
202. And obtaining a target training set.
The training samples in the target training set comprise historical behavior state information, historical vehicle road information, historical decision information and corresponding actual running state information.
In this embodiment, since the training process is actually a process of tuning specific parameters based on the model frame based on the sample data, training samples required for training, that is, the target training set, need to be acquired first in the process. Of course, the information contained in the target training set is the data of the known specific driving situation of the vehicle, that is, the recorded historical data, which contains the historical behavior state information, the historical vehicle road information and the actual running state information, and the actual running state information is the information representing the specific driving situation of the vehicle, and the prediction result after the prediction model of the embodiment is executed corresponds to the prediction result, that is, by taking the data of the known result as a sample for training, the model can be ensured to determine the specific characteristics of the vehicle driving based on the sample, so that when the follow-up prediction is performed again, the follow-up driving situation and track of the vehicle can be analyzed based on the real-time behavior state information, the vehicle road information and the historical decision information in the driving process of the current vehicle, thereby ensuring that the prediction result is more similar to the actual result, and ensuring that the prediction result is more accurate.
203. And training the target training set and the model framework based on the attention mechanism algorithm to generate the vehicle state prediction model.
Specifically, in the process of model training, the method can further specifically include the following steps:
Firstly, setting a loss function through an actual result and the predicted result, and determining a distortion proportion parameter of the loss function;
Then, respectively inputting the historical vehicle road information of the training sample into a first input module of the model framework, inputting the historical driving state information of the training sample into a second input module of the model framework, inputting the historical decision information of the training sample into a third input module, and controlling the prediction module to predict by combining an attention mechanism algorithm to obtain a simulation prediction result;
then, comparing the simulation prediction result with the corresponding actual running state information, and determining the prediction parameters of the model framework based on the comparison result and the loss function;
finally, the vehicle state prediction model is generated based on the model framework and the prediction parameters.
In this embodiment, the loss function may be understood as a function that converges and controls when the model is trained, and may ensure a degree of closeness between the predicted result and the actual result. Specifically, the loss function may be:
Wherein, And/>The average error and the true error of the predicted result and the true result are respectively. /(I)To lose the proportion parameter, and/>And/>Then the predicted parameter.
Then based on the control of the loss function during model training, the base can be ensuredAnd/>The prediction parameters which are particularly suitable for the model are reversely deduced, namely, the more suitable/>, are solvedAnd/>Thus, the prediction parameters matched with the model framework are obtained, and the function of parameter optimization of the vehicle state prediction model constructed based on the neural network is realized.
In addition, as can be seen from the foregoing description of the steps, in the present embodiment, the model frame in the vehicle state prediction model includes a vehicle interaction module for calculating the relationship between vehicles and a lane interaction module for calculating the relationship between vehicles and roads, and in this case, the actual calculation process of these two modules may be:
the vehicle interaction module is mainly used for calculating the influence of surrounding vehicles on the target vehicle. The function of the vehicle interaction module mainly depends on an attention mechanism algorithm, and specifically, the calculation mode of the attention mechanism algorithm is as follows:
Where q, k, v are inputs to the attention mechanism algorithm; ，/>，/> The matrices with different parameters are used for converting the input vector into a higher dimension for calculation; /(I) Is the dimension of vector K. In the vehicle interaction module, the codes of the target vehicles are taken as q vectors, the codes of surrounding vehicles are taken as k vectors and v vectors, and the attention mechanism is calculated to obtain a new target vehicle code/>. In addition, q, k, v are typically in the form of [ N, D ], N being the number of vehicles or roads and D being the dimension of the vehicle or road information. /(I)，/>，/>Respectively, a matrix for information characterization, namely, converting the original input information from D dimension to a space with higher dimension, wherein parameter values in the matrix are determined through training of deep learning,/>The dimension of the high-dimensional space is characterized by the W matrix.
In addition, the lane interaction module is used for calculating the influence of lane information on the target vehicle. The vehicle road interaction module is also composed of a mutual attention mechanism, in which the target vehicle code output by the vehicle interaction module is usedAs q vector, use the coding vector/>, of the lane segmentAs a k vector and a v vector. The vehicle road interaction module calculates a new target vehicle code/>As an output.
Furthermore, in the present embodiment, the encoding vector of the lane segmentThe determination process of (2) may be: according to Euclidean distance from the position of the vehicle at the moment t, screening out the vector of the sampling point and the lane segment thereof which are closer to the central line of the lane, and writingWherein/>To screen the number of post-lane segments. And encoding the information of each lane segment by a lane information encoding module. The lane information coding module is composed of a multi-layer perceptron, inputs a 4-dimensional vector formed by connecting sampling points and vectors, outputs a 64-dimensional high-dimensional coded vector, and is recorded as/>。
204. And acquiring running state information, vehicle road relation information and historical decision information of the vehicle.
The vehicle road relation information is used for representing the position relation between a vehicle and a road, and is calculated based on the running state information; the historical decision information is used for representing vehicle road relation information of the vehicle in a historical target period.
As can be seen from the construction and training of the prediction model in the foregoing steps, the vehicle road relationship in this embodiment may be performed based on the description in the foregoing step 203, and in some cases, the determination process of the vehicle road relationship in this embodiment may also be determined in another manner, which specifically includes:
obtaining lane information, wherein the lane information at least comprises a lane route and the lane width;
acquiring running state information of a vehicle, wherein the running state information at least comprises speed, acceleration, running time, starting position and distance track width;
calculating position coordinates of the vehicle based on the speed, the acceleration, the running time and the starting position, and determining a route position of the vehicle based on the position coordinates and the lane route;
Calculating the Euclidean distance from a vehicle to a lane center line based on the lane width and the distance lane edge width, and determining the lane position of the vehicle based on the Euclidean distance;
The course position and the lane position are determined as the vehicle road relation information.
In this way, it is ensured that in the course of acquiring the road information of the vehicle, the course position of the vehicle in the lane course can be determined on the basis of the acceleration, speed, travel time and starting position of the vehicle, and the euclidean distance of the vehicle to the center line of the lane is calculated on the basis of the lane width and the distance-to-lane-side width, and the lane position is determined in this way, which corresponds to the case where the x-axis and the y-axis of the vehicle in the course formed by this road are determined. Thus, the accuracy of the obtained vehicle road relation information is ensured.
Further, in this embodiment, the process of obtaining the lane information may further specifically be:
Firstly, collecting guideboard information of the lane, and determining a lane route of the lane based on the guideboard information;
Then, dividing the lane route into a plurality of lane segments based on a preset distance to obtain a lane queue, wherein the arrangement sequence of the lane segments in the lane queue corresponds to the positions of the lane segments in the lane route;
And finally, respectively marking the center line position and the edge position of each lane segment, and determining the lane width of the lane segment based on the center line position and the edge position to obtain the lane information.
In this embodiment, the guideboard information may actually hover at a fixed position above the acquisition scene using the unmanned aerial vehicle, capture a specific traffic road section, and acquire the lane route based on cloud data. After determining the lane route, the route may be segmented, i.e. divided according to a predetermined distance, to obtain a plurality of lane segments, and then a queue, i.e. a lane queue, is formed based on the lane segments. Specifically, the specific content and form related to the lane queue may be: in this example, the positions of the lanes within the video screenshot taken by the unmanned aerial vehicle are used for labeling, and a transfer matrix between the labeling map coordinate system and the track data coordinate system is calculated. A map scene is typically made up of multiple lanes, in this example, each lane is split into multiple segments, each segment is represented by a vector, and the final extracted map lane information can be represented as follows: where i is the number of the lane segment,/> For the number of all lane segments in the scene,/>For the start point coordinates of the ith lane segment,/>Is the vector of the i-th lane segment.
205. And based on the driving state information, the vehicle road relation information and the historical decision information, performing prediction operation by using a vehicle state prediction model to obtain a prediction result.
The vehicle state prediction model is trained through an attention mechanism algorithm through historical behavior state information, historical vehicle road relation information and historical decision information.
It should be noted that, in this embodiment, the history decision information is specifically a vehicle road relation queue; the vehicle road relation in the historical decision information is specifically a lane segment relation, wherein each vehicle segment relation is determined based on the lane segment and the driving state information.
Specifically, as can be seen from the foregoing description about the steps, the prediction model of the present embodiment is actually divided into a plurality of different modules, and based on this, the step may be performed specifically when the prediction behavior is performed:
Based on the foregoing, the function of the module is to store the output of the vehicle road interaction module at a time step prior to time t Vectors are calculated in the sequential encoding module by means of a self-attention mechanism in the sequential encoding module. In the coding module corresponding to the history decision module, a 64-dimensional randomly initialized vector is added at the last of the history coding vector sequence, the sequence is stacked to be used as q vector, k vector and v vector of an attention mechanism, a 64-dimensional coding vector sequence is obtained after calculation, and the last element of the sequence is taken and recorded as/>As a result of the computation by the timing encoding module.
In particular, the execution module is predicted toAnd/>As input, the information of two coding vectors is synthesized through the structure of the multi-layer perceptron, and finally the acceleration/>, which should be adopted by the target vehicle at the time t, is output。
And when the prediction is finished, outputting the target vehicle code by the vehicle road interaction moduleThe queue module is added, and if the queue is full at this time, the code which enters the queue earliest is deleted.
It should be noted that, the history decision module of this embodiment stores the prediction result made by the model at each moment, and the specific stored information isVector.
By adding the codes on the stored historical time steps and the codes at the current moment, a decision queue is obtained, and the decision queue reflects the complete process of deciding the behavior of the model on the target vehicle and controlling the running of the vehicle. The purpose of calculating the queue by using a self-attention mechanism is to construct interactions of different elements in the sequence, so that a decision result at the current moment can comprehensively consider decision information at the historical moment, and consistency of vehicle running are ensured.
In addition, the function of the additional random initialization vector described above is to store all the information of the whole queue, the calculation of the self-attention mechanism will integrate the information of the decision queue into the additional random vector, and the final timing encoding module outputs the additional vector integrating all the information.
Further, as an implementation of the method shown in fig. 1 and fig. 2, another embodiment of the present application further provides a device for predicting a vehicle state. The embodiment of the device corresponds to the embodiment of the method, and for convenience of reading, details of the embodiment of the method are not repeated one by one, but it should be clear that the device in the embodiment can correspondingly realize all the details of the embodiment of the method. In order to implement a method for predicting a vehicle state, as shown in fig. 4, the apparatus includes:
An obtaining unit 31, configured to obtain driving state information of a vehicle, vehicle road relation information, and history decision information, where the vehicle road relation information may be used to characterize a positional relation between the vehicle and a road, and the vehicle road relation information is calculated based on the driving state information; the historical decision information may be used to characterize vehicle road relationship information for the vehicle over a historical target period;
The execution unit 32 may be configured to execute a prediction operation by using a vehicle state prediction model based on the driving state information, the vehicle road relation information, and the historical decision information, to obtain a prediction result, where the vehicle state prediction model is trained by using an attention mechanism algorithm through historical behavior state information, historical vehicle road relation information, and historical decision information.
Further, as shown in fig. 5, the apparatus further includes:
A setting unit 33 that may be used to set a model frame of the vehicle state prediction model; the model framework comprises a first input module, a second input module, a third input module and a prediction execution module; the first input module can be used for inputting vehicle road relation information, the second input module can be used for inputting driving state information, and the third input module can be used for inputting historical decision information; the prediction execution module may be configured to input results of the first input module, the second input module, and the third input module and execute prediction;
the training set obtaining unit 34 may be configured to obtain a target training set, where a training sample in the target training set includes historical behavior state information, historical vehicle road information, historical decision information, and corresponding actual running state information;
The training unit 35 may be configured to perform a training operation on the target training set and the model frame based on the attention mechanism algorithm, and generate the vehicle state prediction model.
Further, as shown in fig. 5, the training unit 35 includes:
a setting subunit 351, configured to set a loss function according to an actual result and the predicted result, and determine a distortion ratio parameter of the loss function;
The input subunit 352 may be configured to input, respectively, historical vehicle road information of the training sample to a first input module of the model framework, input the historical driving state information of the training sample to a second input module of the model framework, input the historical decision information of the training sample to the third input module, and control the prediction module to perform prediction in combination with an attention mechanism algorithm, so as to obtain a simulated prediction result;
a determining subunit 353, configured to compare the simulated prediction result with the corresponding actual running state information, and determine a prediction parameter of the model framework based on the comparison result and the loss function;
A generation subunit 354 may be configured to generate the vehicle state prediction model based on the model framework and the prediction parameters.
Further, as shown in fig. 5, the apparatus further includes:
A lane information acquisition unit 36 operable to acquire lane information, wherein the lane information includes at least a lane route and the lane width;
a running state information acquisition unit 37 operable to acquire running state information of the vehicle, the running state information including at least a speed, an acceleration, a running time, a start position, and a distance track width;
A first determining unit 38 operable to calculate a position coordinate of the vehicle based on the speed, acceleration, travel time, and start position, and determine a course position of the vehicle based on the position coordinate and the lane course;
A second determining unit 39 operable to calculate a euclidean distance from a vehicle to a lane center line based on the lane width and the distance-to-lane-edge width, and determine a lane position of the vehicle based on the euclidean distance;
The third determination unit 40 may be configured to determine the course position and the lane position as the vehicle road relation information.
Further, as shown in fig. 5, the lane information acquiring unit 36 includes:
An acquisition subunit 361, configured to acquire guideboard information of the lane, and determine a lane route of the lane based on the guideboard information;
The dividing sub-unit 362 may be configured to divide the lane route into a plurality of lane segments based on a preset distance, so as to obtain a lane queue, where an arrangement order of the lane segments in the lane queue corresponds to a position of the lane segments in the lane route;
A determining subunit 363 may be configured to mark a center line position and an edge position of each of the lane segments, respectively, determine a lane width of the lane segment based on the center line position and the edge position, and obtain the lane information.
Further, as shown in fig. 5, the historical decision information is specifically a vehicle road relation queue; the vehicle road relation in the historical decision information is specifically a lane segment relation, wherein each vehicle segment relation is determined based on the lane segment and the driving state information.
In order to achieve the above object, according to another aspect of the present application, there is further provided a storage medium, where the storage medium includes a stored program, and when the program runs, a device where the storage medium is controlled to execute the above method for predicting a vehicle state.
In order to achieve the above object, according to another aspect of the present application, an embodiment of the present application further provides a prediction apparatus of a vehicle state, the apparatus including a storage medium; and one or more processors coupled to the storage medium, the processors configured to execute the program instructions stored in the storage medium; and executing the method for predicting the vehicle state when the program instructions run.
Compared with the prior art, the vehicle state prediction method and device provided by the embodiment of the application are used for predicting based on the vehicle state prediction model in the vehicle state prediction process, wherein the vehicle state prediction model is trained based on historical behavior state information, historical vehicle road relation information and historical decision information, that is to say, the vehicle state prediction model is actually trained on the basis of taking the historical running condition of the vehicle as a sample, and the model is different from the parameters of the expression based on manual setting in the prior art, and the parameters are not required to be set manually, so that the threshold requirement on setting experience in the manual setting process is avoided, the problem that the accuracy of a follow-up prediction result is influenced due to insufficient experience is avoided, and the accuracy of the prediction result can be improved. Meanwhile, in the application, the method for predicting the vehicle state can be used as a prediction evidence from three aspects of vehicle running state information, vehicle road relation and historical decision information, and compared with the prior art for predicting the vehicle running state only through acceleration, speed and the like of the vehicle, the method increases the consideration of the two aspects of the interrelationship between the vehicle and the road and the vehicle historical decision, thereby comprehensively analyzing and predicting the finally predicted vehicle state from more aspects of angles, and further improving the accuracy of a prediction result.
The prediction device for the vehicle state comprises a processor and a memory, wherein the transmission unit, the calculation verification unit, the application verification unit and the like are all stored in the memory as program units, and the processor executes the program units stored in the memory to realize corresponding functions.
The processor includes a kernel, and the kernel fetches the corresponding program unit from the memory. The kernel can be provided with one or more than one kernel, and a prediction method of the vehicle state is realized by adjusting kernel parameters, so that the problem that the accuracy of a prediction result is lower when the existing vehicle state is predicted is solved.
The embodiment of the application provides a prediction device of a vehicle state, which comprises a storage medium; and one or more processors coupled to the storage medium, the processors configured to execute the program instructions stored in the storage medium; the program instructions, when executed, perform the method of predicting a vehicle state of any one of the preceding claims.
The embodiment of the application provides a storage medium which comprises a stored program, wherein equipment where the storage medium is located is controlled to execute the method for predicting the vehicle state when the program runs.
The storage medium may include volatile memory, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM), among other forms in computer readable media, the memory including at least one memory chip.
The embodiment of the application provides equipment, which comprises a processor, a memory and a program stored in the memory and capable of running on the processor, wherein the processor realizes the following steps when executing the program: acquiring running state information, vehicle road relation information and historical decision information of a vehicle, wherein the vehicle road relation information is used for representing the position relation between the vehicle and a road, and the vehicle road relation information is calculated based on the running state information; the historical decision information is used for representing vehicle road relation information of the vehicle in a historical target period; and based on the driving state information, the vehicle road relation information and the historical decision information, performing prediction operation by using a vehicle state prediction model to obtain a prediction result, wherein the vehicle state prediction model is trained by a attention mechanism algorithm through the historical behavior state information, the historical vehicle road relation information and the historical decision information.
Further, before the predicting operation is performed by using the vehicle state predicting model based on the driving state information, the vehicle road relation information and the historical decision information, the method further includes:
Setting a model frame of the vehicle state prediction model; the model framework comprises a first input module, a second input module, a third input module and a prediction execution module; the first input module is used for inputting vehicle road relation information, the second input module is used for inputting driving state information, and the third input module is used for inputting historical decision information; the prediction execution module is used for inputting the input results of the first input module, the second input module and the third input module and executing prediction;
acquiring a target training set, wherein training samples in the target training set comprise historical behavior state information, historical vehicle road information, historical decision information and corresponding actual running state information;
and training the target training set and the model framework based on the attention mechanism algorithm to generate the vehicle state prediction model.
Further, the training operation is performed on the target training set and the model framework based on the attention mechanism algorithm, and the vehicle state prediction model is generated, which includes:
Setting a loss function through an actual result and the predicted result, and determining a distortion proportion parameter of the loss function;
Respectively inputting the historical vehicle road information of the training sample into a first input module of the model framework, inputting the historical driving state information of the training sample into a second input module of the model framework, inputting the historical decision information of the training sample into a third input module, and controlling a prediction module to predict by combining an attention mechanism algorithm to obtain a simulation prediction result;
Comparing the simulation prediction result with the corresponding actual running state information, and determining the prediction parameters of the model framework based on the comparison result and the loss function;
the vehicle state prediction model is generated based on the model framework and the prediction parameters.
Further, the method further comprises:
obtaining lane information, wherein the lane information at least comprises a lane route and the lane width;
acquiring running state information of a vehicle, wherein the running state information at least comprises speed, acceleration, running time, starting position and distance track width;
calculating position coordinates of the vehicle based on the speed, the acceleration, the running time and the starting position, and determining a route position of the vehicle based on the position coordinates and the lane route;
Calculating the Euclidean distance from a vehicle to a lane center line based on the lane width and the distance lane edge width, and determining the lane position of the vehicle based on the Euclidean distance;
The course position and the lane position are determined as the vehicle road relation information.
Further, the obtaining lane information includes:
collecting guideboard information of the lane, and determining a lane route of the lane based on the guideboard information;
Dividing the lane route into a plurality of lane segments based on a preset distance to obtain a lane queue, wherein the arrangement sequence of the lane segments in the lane queue corresponds to the positions of the lane segments in the lane route;
And respectively marking the center line position and the edge position of each lane segment, and determining the lane width of the lane segment based on the center line position and the edge position to obtain the lane information.
Further, the historical decision information is specifically a vehicle road relation queue; the vehicle road relation in the historical decision information is specifically a lane segment relation, wherein each vehicle segment relation is determined based on the lane segment and the driving state information.
The present application also provides a computer program product that can perform corresponding functions, comprising: transmitting metadata to a computing environment after the metadata is generated in a generation environment; acquiring running state information, vehicle road relation information and historical decision information of a vehicle, wherein the vehicle road relation information is used for representing the position relation between the vehicle and a road, and the vehicle road relation information is calculated based on the running state information; the historical decision information is used for representing vehicle road relation information of the vehicle in a historical target period; and based on the driving state information, the vehicle road relation information and the historical decision information, performing prediction operation by using a vehicle state prediction model to obtain a prediction result, wherein the vehicle state prediction model is trained by a attention mechanism algorithm through the historical behavior state information, the historical vehicle road relation information and the historical decision information.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, etc., such as Read Only Memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.
Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises an element.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The foregoing is merely exemplary of the present application and is not intended to limit the present application. Various modifications and variations of the present application will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. which come within the spirit and principles of the application are to be included in the scope of the claims of the present application.
Claims (10)
1. A method of predicting a vehicle state, the method comprising:
Acquiring running state information, vehicle road relation information and historical decision information of a vehicle, wherein the vehicle road relation information is used for representing the position relation between the vehicle and a road, and the vehicle road relation information is calculated based on the running state information; the historical decision information is used for representing vehicle road relation information of the vehicle in a historical target period;
And based on the driving state information, the vehicle road relation information and the historical decision information, performing prediction operation by using a vehicle state prediction model to obtain a prediction result, wherein the vehicle state prediction model is trained by a attention mechanism algorithm through the historical behavior state information, the historical vehicle road relation information and the historical decision information.
2. The method according to claim 1, wherein before said performing a prediction operation using a vehicle state prediction model based on said travel state information, said vehicle road relation information, and said history decision information, said method further comprises:
Setting a model frame of the vehicle state prediction model; the model framework comprises a first input module, a second input module, a third input module and a prediction execution module; the first input module is used for inputting vehicle road relation information, the second input module is used for inputting driving state information, and the third input module is used for inputting historical decision information; the prediction execution module is used for inputting the input results of the first input module, the second input module and the third input module and executing prediction;
acquiring a target training set, wherein training samples in the target training set comprise historical behavior state information, historical vehicle road information, historical decision information and corresponding actual running state information;
and training the target training set and the model framework based on the attention mechanism algorithm to generate the vehicle state prediction model.
3. The method of claim 2, wherein the training the target training set and the model framework based on the attention mechanism algorithm to generate the vehicle state prediction model comprises:
Setting a loss function through an actual result and the predicted result, and determining a distortion proportion parameter of the loss function;
Respectively inputting the historical vehicle road information of the training sample into a first input module of the model framework, inputting the historical driving state information of the training sample into a second input module of the model framework, inputting the historical decision information of the training sample into a third input module, and controlling a prediction module to predict by combining an attention mechanism algorithm to obtain a simulation prediction result;
Comparing the simulation prediction result with the corresponding actual running state information, and determining the prediction parameters of the model framework based on the comparison result and the loss function;
the vehicle state prediction model is generated based on the model framework and the prediction parameters.
4. A method according to claim 1 or 3, characterized in that the method further comprises:
obtaining lane information, wherein the lane information at least comprises a lane route and the lane width;
acquiring running state information of a vehicle, wherein the running state information at least comprises speed, acceleration, running time, starting position and distance track width;
calculating position coordinates of the vehicle based on the speed, the acceleration, the running time and the starting position, and determining a route position of the vehicle based on the position coordinates and the lane route;
Calculating the Euclidean distance from a vehicle to a lane center line based on the lane width and the distance lane edge width, and determining the lane position of the vehicle based on the Euclidean distance;
The course position and the lane position are determined as the vehicle road relation information.
5. The method of claim 4, wherein the acquiring lane information comprises:
collecting guideboard information of the lane, and determining a lane route of the lane based on the guideboard information;
Dividing the lane route into a plurality of lane segments based on a preset distance to obtain a lane queue, wherein the arrangement sequence of the lane segments in the lane queue corresponds to the positions of the lane segments in the lane route;
And respectively marking the center line position and the edge position of each lane segment, and determining the lane width of the lane segment based on the center line position and the edge position to obtain the lane information.
6. The method according to claim 4, wherein the historical decision information is in particular a vehicle road relation queue; the vehicle road relation in the historical decision information is specifically a lane segment relation, wherein each vehicle segment relation is determined based on the lane segment and the driving state information.
7. A prediction apparatus of a vehicle state, characterized by comprising:
the system comprises an acquisition unit, a calculation unit and a control unit, wherein the acquisition unit is used for acquiring running state information of a vehicle, vehicle road relation information and historical decision information, the vehicle road relation information is used for representing the position relation between the vehicle and a road, and the vehicle road relation information is calculated based on the running state information; the historical decision information is used for representing vehicle road relation information of the vehicle in a historical target period;
And the execution unit is used for executing prediction operation by utilizing a vehicle state prediction model based on the driving state information, the vehicle road relation information and the historical decision information to obtain a prediction result, wherein the vehicle state prediction model is trained by a attention mechanism algorithm through the historical behavior state information, the historical vehicle road relation information and the historical decision information.
8. The apparatus of claim 7, wherein the apparatus further comprises:
A setting unit configured to set a model frame of the vehicle state prediction model; the model framework comprises a first input module, a second input module, a third input module and a prediction execution module; the first input module is used for inputting vehicle road relation information, the second input module is used for inputting driving state information, and the third input module is used for inputting historical decision information; the prediction execution module is used for inputting the input results of the first input module, the second input module and the third input module and executing prediction;
the training set acquisition unit is used for acquiring a target training set, wherein a training sample in the target training set comprises historical behavior state information, historical vehicle road information, historical decision information and corresponding actual running state information;
and the training unit is used for training the target training set and the model frame based on the attention mechanism algorithm to generate the vehicle state prediction model.
9. A storage medium comprising a stored program, wherein the program, when run, controls a device in which the storage medium is located to perform the method of predicting a vehicle state according to any one of claims 1-6.
10. A prediction apparatus of a vehicle state, characterized in that the apparatus includes a storage medium; and one or more processors coupled to the storage medium, the processors configured to execute the program instructions stored in the storage medium; the program instructions, when executed, perform the method of predicting a vehicle state of any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410425010.3A CN118013868A (en) | 2024-04-10 | 2024-04-10 | Vehicle state prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410425010.3A CN118013868A (en) | 2024-04-10 | 2024-04-10 | Vehicle state prediction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118013868A true CN118013868A (en) | 2024-05-10 |
Family
ID=90948948
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410425010.3A Pending CN118013868A (en) | 2024-04-10 | 2024-04-10 | Vehicle state prediction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118013868A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220363259A1 (en) * | 2019-11-27 | 2022-11-17 | Momenta (suzhou) Technology Co., Ltd. | Method for generating lane changing decision-making model, method for lane changing decision-making of unmanned vehicle and electronic device |
| CN115991196A (en) * | 2023-03-01 | 2023-04-21 | 云控智行(上海)汽车科技有限公司 | LSTM-based vehicle lane change strategy control algorithm and software |
| CN116252813A (en) * | 2022-09-09 | 2023-06-13 | 中国第一汽车股份有限公司 | Vehicle driving track prediction method, device and storage medium |
| CN117325865A (en) * | 2023-11-10 | 2024-01-02 | 湖北汽车工业学院 | Intelligent vehicle lane change decision method and system for LSTM track prediction |
| CN117585017A (en) * | 2023-12-11 | 2024-02-23 | 西部科学城智能网联汽车创新中心(重庆)有限公司 | Automatic driving vehicle lane change decision method, device, equipment and storage medium |
-
2024
- 2024-04-10 CN CN202410425010.3A patent/CN118013868A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220363259A1 (en) * | 2019-11-27 | 2022-11-17 | Momenta (suzhou) Technology Co., Ltd. | Method for generating lane changing decision-making model, method for lane changing decision-making of unmanned vehicle and electronic device |
| CN116252813A (en) * | 2022-09-09 | 2023-06-13 | 中国第一汽车股份有限公司 | Vehicle driving track prediction method, device and storage medium |
| CN115991196A (en) * | 2023-03-01 | 2023-04-21 | 云控智行(上海)汽车科技有限公司 | LSTM-based vehicle lane change strategy control algorithm and software |
| CN117325865A (en) * | 2023-11-10 | 2024-01-02 | 湖北汽车工业学院 | Intelligent vehicle lane change decision method and system for LSTM track prediction |
| CN117585017A (en) * | 2023-12-11 | 2024-02-23 | 西部科学城智能网联汽车创新中心(重庆)有限公司 | Automatic driving vehicle lane change decision method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112015847B (en) | Obstacle trajectory prediction method and device, storage medium and electronic equipment | |
| KR102548732B1 (en) | Apparatus and Method for learning a neural network | |
| CN111152796B (en) | Vehicle motion state prediction method and device | |
| JP2022547611A (en) | Simulation of various long-term future trajectories in road scenes | |
| CN112489497B (en) | Airspace operation complexity evaluation method based on deep convolutional neural network | |
| CN111290370B (en) | Automatic driving performance detection method and device | |
| KR102429372B1 (en) | Apparatus and method for calculating ratio of negligence based on 3d simulator | |
| CN113238970B (en) | Training method, evaluation method, control method and device of automatic driving model | |
| CN112465199A (en) | Airspace situation evaluation system | |
| CN113128381A (en) | Obstacle trajectory prediction method, system and computer storage medium | |
| JP2019074849A (en) | Drive data analyzer | |
| KR102087889B1 (en) | A deep-learning eduacation system using arrow images | |
| CN116306790B (en) | Offshore ship track real-time prediction method, system, equipment and medium based on CNN-GRU and attention mechanism | |
| KR102014288B1 (en) | Development pressure prediction method based on artificial intelligence using drone | |
| CN118013868A (en) | Vehicle state prediction method and device | |
| CN116859931A (en) | Training method of track planning model, vehicle control mode and device | |
| CN111950753A (en) | Scenic spot passenger flow prediction method and device | |
| Ge et al. | Deep reinforcement learning navigation via decision transformer in autonomous driving | |
| CN114510468A (en) | Data processing method and computer program product | |
| CN113822293A (en) | Model processing method, device and equipment for graph data and storage medium | |
| Toubeh et al. | Risk-aware planning by confidence estimation using deep learning-based perception | |
| CN117058498B (en) | Training method of segmentation map evaluation model, and segmentation map evaluation method and device | |
| CN117132958B (en) | Road element identification method and related device | |
| CN117615359B (en) | Bluetooth data transmission method and system based on multiple rule engines | |
| CN117971399B (en) | Container number adjusting method, intelligent computing cloud operating system and computing platform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |