CN115086375A - Method, device, system and medium for compensating motion state information delay of networked vehicle - Google Patents

Abstract

The application provides a method, a device, a system and a medium for compensating motion state information delay of networked vehicles, wherein the method comprises the following steps: the method comprises the steps of obtaining current motion state information and current delay information of a sending vehicle, obtaining historical motion state information and corresponding historical delay information within preset time of the sending vehicle, inputting the current motion state information, the current delay information, the historical motion state information and the historical delay information into a long-short term memory network model, enabling the long-short term memory network model to output prediction compensation information of the sending vehicle, and calculating according to the prediction compensation information to obtain prediction compensation displacement of the sending vehicle. Therefore, according to the long-short term memory network model, the vehicle motion state information and the delay information can be modeled and predicted to compensate the vehicle motion state information in the time period of vehicle-mounted communication delay, so that the aim of reducing the influence of vehicle-mounted communication delay is fulfilled, and the running safety of the networked vehicle is improved.

Description

Method, device, system and medium for delay compensation of motion state information of connected vehicle

technical field

The present application relates to the field of computers, and in particular, to a method, device, system and medium for delay compensation of motion state information of a networked vehicle.

Background technique

Over the past few decades, the emergence of Connected Autonomous Vehicle (CAV) technology has brought new changes to the transportation system, benefiting our daily driving in terms of safety, mobility and sustainability. The experience in other aspects has been significantly improved.

Networked autonomous vehicles equipped with communication equipment can operate cooperatively. The cooperation between vehicles mainly refers to the realization of all-round communication between the vehicle and surrounding vehicles, the environment and the network through the V2X (vehicle-to-everything) technology of the Internet of Vehicles. Vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), vehicle-to-pedestrian (V2P), vehicle-to-network (V2N) It provides environmental perception, information interaction and collaborative control capabilities for automotive driving and traffic management applications.

Vehicles rely on onboard perception sensors, such as cameras, radar, and lidar, to measure the state of nearby vehicles. With the introduction of V2X communication, connected autonomous vehicles can obtain data beyond their immediate sensing range and obtain information that cannot be detected by remote sensors, which helps to improve the sensing range of connected autonomous vehicles. However, when it comes to vehicle communication technology, issues such as communication delays will inevitably be introduced, which will degrade the performance of any connected autonomous vehicle application.

That is to say, after the receiving vehicle receives the motion state information sent by the sending vehicle, the sending vehicle has already displaced a certain amount within the delay time. This vehicle communication delay problem makes the information (position, speed and acceleration, etc.) of the sending vehicle obtained by the receiving vehicle. ) is not real-time, and the safety of the receiving vehicle is bound to be affected to some extent.

SUMMARY OF THE INVENTION

In view of this, the purpose of the present application is to provide a method, device, system and medium for delay compensation of motion state information of a connected vehicle, which can reduce the influence of delay in vehicle communication and improve the operation safety of connected vehicles.

To achieve the above object, the application has the following technical solutions:

In a first aspect, an embodiment of the present application provides a method for delay compensation of motion state information of a connected vehicle, including:

Obtain the current motion state information and current delay information of the sending vehicle; the current motion state information includes the current position, the current speed and the current acceleration; the current delay information includes the delay from the previous time corresponding to the current time to the current time duration;

Obtain the historical motion state information and corresponding historical delay information within the preset duration of the sending vehicle; the historical motion state information includes the position at each moment, the speed at each moment and the acceleration at each moment; the historical delay information Including the corresponding delay time from each moment to the previous moment of each moment;

Inputting the current motion state information, the current delay information, the historical motion state information, and the historical delay information into a long-term and short-term memory network model, so that the long-term and short-term memory network model outputs the prediction of the sending vehicle Compensation information; the predicted compensation information includes the predicted speed after delay, the predicted acceleration after delay, and the predicted delay time;

The predicted compensation displacement of the sending vehicle is obtained by calculating according to the predicted compensation information.

In a possible implementation manner, before the current motion state information, the current delay information, the historical motion state information and the historical delay information are input into the long short-term memory network model, the Methods also include:

Acquire a training set of the long short-term memory network model, the training set includes: actual motion state information at a known time, actual delay information at a known time, actual motion state information at the next time at a known time, and past information. The actual delay information of the next moment after the known moment;

The actual motion state information at the known time includes the actual position, actual speed and actual acceleration at the known time; the actual delay information at the known time includes the actual delay from the previous time corresponding to the known time to the known time. duration; the actual motion state information at the next moment at the known moment includes the actual position, actual speed and actual acceleration at the next moment at the known moment; the actual delay information at the next moment at the known moment includes The actual delay time from the known moment to the next moment from the known moment;

Use the training set to learn the motion state information of the known time and the actual delay information of the known time, and the actual motion state information of the next time of the known time and the next time of the known time The mapping relationship of the actual delay information at the moment;

The model parameters of the long short-term memory network model are determined according to the mapping relationship.

In a possible implementation manner, inputting the current motion state information, the current delay information, the historical motion state information and the historical delay information into a long short-term memory network model, including:

The current motion state information, the current delay information, the historical motion state information and the historical delay information are encoded and input into the long short-term memory network model;

The encoding includes uniformly standardizing the current motion state information, the current delay information, the historical motion state information, and the historical delay information.

In a possible implementation manner, the calculating and obtaining the predicted compensation displacement of the sending vehicle according to the predicted compensation information includes:

The predicted compensated displacement of the sending vehicle is equal to the product of the predicted delayed speed and the predicted delayed duration.

In a second aspect, an embodiment of the present application provides a device for compensating for motion state information delay of a connected vehicle, including:

The first obtaining unit is used to obtain the current motion state information and current delay information of the sending vehicle; the current motion state information includes the current position, the current speed and the current acceleration; the current delay information includes the last time corresponding to the current moment. The delay time from time to current time;

The second acquiring unit is configured to acquire the historical motion state information and the corresponding historical delay information within the preset duration of the sending vehicle; the historical motion state information includes the position at each moment, the speed at each moment and the acceleration at each moment ; Described historical delay information includes the respective corresponding delay time lengths from each moment to the previous moment of each moment;

The input unit is used to input the current motion state information, the current delay information, the historical motion state information and the historical delay information into the long-term and short-term memory network model, so that the long-term and short-term memory network model outputs all the information. The predicted compensation information of the sending vehicle; the predicted compensation information includes the predicted delayed speed, the predicted delayed acceleration and the predicted delayed time duration;

A calculation unit, configured to calculate and obtain the predicted compensation displacement of the sending vehicle according to the predicted compensation information.

In a possible implementation, the apparatus further includes:

The third obtaining unit is configured to obtain a training set of the long short-term memory network model, the training set includes: actual motion state information at a known moment, actual delay information at a known moment, and the next moment at a known moment The actual motion state information and the actual delay information of the next moment at the known moment;

The actual motion state information at the known time includes the actual position, actual speed and actual acceleration at the known time; the actual delay information at the known time includes the actual delay from the previous time corresponding to the known time to the known time. duration; the actual motion state information at the next moment at the known moment includes the actual position, actual speed and actual acceleration at the next moment at the known moment; the actual delay information at the next moment at the known moment includes The actual delay time from the known moment to the next moment from the known moment;

The learning unit is configured to use the training set to learn the motion state information of the known time and the actual delay information of the known time, and the actual motion state information of the next time of the known time and the already known time. The mapping relationship of the actual delay information at the next moment at the known moment;

A determining unit, configured to determine model parameters of the long short-term memory network model according to the mapping relationship.

In a possible implementation manner, the input unit is specifically used for:

The current motion state information, the current delay information, the historical motion state information and the historical delay information are encoded and input into the long short-term memory network model;

The encoding includes uniformly standardizing the current motion state information, the current delay information, the historical motion state information, and the historical delay information.

In a possible implementation manner, the computing unit is specifically used for:

The predicted compensated displacement of the sending vehicle is equal to the product of the predicted delayed speed and the predicted delayed duration.

In a third aspect, an embodiment of the present application provides a system for delay compensation of motion state information of a connected vehicle, including:

memory for storing computer programs;

The processor is configured to implement the steps of the above-mentioned method for compensating for the delay of motion state information of a connected vehicle when executing the computer program.

In a fourth aspect, an embodiment of the present application provides a computer-readable medium, where a computer program is stored on the computer-readable medium, and when the computer program is processed and executed, the delay of the motion state information of the connected vehicle as described above is realized. The steps of the compensation method.

Compared with the prior art, the embodiments of the present application have the following beneficial effects:

Embodiments of the present application provide a method, device, system and medium for delay compensation of motion state information of a connected vehicle. The method includes: acquiring current motion state information and current delay information of a sending vehicle, where the current motion state information includes a current position , the current speed and the current acceleration, the current delay information includes the delay time from the previous time corresponding to the current time to the current time; obtain the historical motion state information and the corresponding historical delay information within the preset time period of the sending vehicle, and the historical motion state The information includes the position of each moment, the speed of each moment and the acceleration of each moment, and the historical delay information includes the corresponding delay time from each moment to the previous moment of each moment; the current motion state information, current delay information, The historical motion state information and historical delay information are input to the long-term and short-term memory network model, so that the long-term and short-term memory network model outputs the predicted compensation information of the vehicle. The predicted compensation information includes the predicted delay speed, the predicted acceleration after the delay, and the predicted delay. The time duration, and the predicted compensation displacement of the sending vehicle is calculated according to the predicted compensation information. Therefore, according to the long-term and short-term memory network model, the vehicle motion state information and delay information can be modeled and predicted to compensate for the vehicle motion state information during the time period of the vehicle communication delay, so as to reduce the impact of the vehicle communication delay. The purpose is to improve the safety of the operation of connected vehicles.

Description of drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are For some embodiments of the present application, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without any creative effort.

The above and other features, advantages and aspects of various embodiments of the present disclosure will become more apparent when taken in conjunction with the accompanying drawings and with reference to the following detailed description. Throughout the drawings, the same or similar reference numbers refer to the same or similar elements. It should be understood that the drawings are schematic and that the originals and elements are not necessarily drawn to scale.

FIG. 1 shows a flowchart of a method for delay compensation of motion state information of a connected vehicle provided by an embodiment of the present application;

FIG. 2 shows a schematic diagram of each device included in an application scenario provided by an embodiment of the present application;

3 shows a schematic diagram of each node unit of an LSTM model provided by an embodiment of the present application;

FIG. 4 shows a flowchart of a method for delay compensation of motion state information of a connected vehicle provided by an embodiment of the present application;

5 shows a schematic diagram of the relationship between historical information and predicted future information under the network architecture of the embodiment of the present application;

6 shows a schematic diagram of the number of neurons in each layer of the LSTM model provided by an embodiment of the present application;

FIG. 7 shows a schematic diagram of a delay compensation device for motion state information of a connected vehicle provided by an embodiment of the present application.

Detailed ways

In order to make the above objects, features and advantages of the present application more clearly understood, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

Many specific details are set forth in the following description to facilitate a full understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can do so without departing from the connotation of the present application. Similar promotion, therefore, the present application is not limited by the specific embodiments disclosed below.

As described in the Background Art, in the past few decades, the emergence of Connected Autonomous Vehicle (CAV) technology has brought new changes to the transportation system, which is beneficial to our daily driving in safety. Experiences such as sex, mobility and sustainability have improved significantly.

Networked autonomous vehicles equipped with communication equipment can operate cooperatively. The cooperation between vehicles mainly refers to the realization of all-round communication between the vehicle and surrounding vehicles, the environment and the network through the V2X (vehicle-to-everything) technology of the Internet of Vehicles. Vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), vehicle-to-pedestrian (V2P), vehicle-to-network (V2N) It provides environmental perception, information interaction and collaborative control capabilities for automotive driving and traffic management applications.

Vehicles rely on onboard perception sensors, such as cameras, radar, and lidar, to measure the state of nearby vehicles. With the introduction of V2X communication, connected autonomous vehicles can obtain data beyond their immediate sensing range and obtain information that cannot be detected by remote sensors, which helps to improve the sensing range of connected autonomous vehicles. However, when it comes to vehicle communication technology, issues such as communication delays will inevitably be introduced, which will degrade the performance of any connected autonomous vehicle application.

That is to say, after the receiving vehicle receives the motion state information sent by the sending vehicle, the sending vehicle has already displaced a certain amount within the delay time. This vehicle communication delay problem makes the information (position, speed and acceleration, etc.) of the sending vehicle obtained by the receiving vehicle. ) is not real-time, and the safety of the receiving vehicle is bound to be affected to some extent.

The applicant's research found that researchers in different directions have conducted research from different angles on how to reduce the impact of in-vehicle communication delays:

The communication direction focuses on reducing the impact of delay from the perspective of communication protocol optimization or routing optimization: from the perspective of protocol optimization, the service channel time slot medium access control mechanism can be designed to improve channel utilization and increase throughput. At the same time, it also reduces the delay of the channel; from the perspective of routing optimization, the routing algorithm can be improved, and the information such as node position and movement speed can be used to predict the link failure time, which is used to improve the end-to-end delay of data packets and transmission throughput. performance in terms of rate and packet delivery rate. Although it has been done well to deal with delay from the perspective of communication, from the perspective of the underlying communication mechanism, the problems of delay and packet loss in V2X communication cannot be permanently eliminated.

The control direction focuses on designing the controller based on the physical motion model. The controller has a feedback unit and a motion estimation unit, which can compensate the motion state of the vehicle according to the real-time error. At the same time, the design process of the controller will consider the existence of communication delay. , to explore its impact on overall stability and security. But this kind of motion model is unreliable for long-term prediction, and the vehicle trajectory is often nonlinear due to the decisions made by the driver during driving, and this model is often not so perfect for dealing with nonlinear situations.

For example, some examples of application of delay compensation for in-vehicle communication with respect to control direction are the motion estimation of vehicles using dynamic model methods such as Extended Kalman Filter, Unscented Kalman Filter and Particle Filter. These filters are modified versions of Kalman filters and they are based on first-order Markov chains, i.e. the state at the current time step only depends on the state at the previous time step. While they are useful for real-time implementations, a key limitation is that sensor data prior to the previous time step cannot be used directly. And the performance of eg Kalman filters depends on the accuracy of the parameter matrices, in particular the process noise covariance matrix Q and the measurement noise covariance matrix R. In practice, the choice of Q and R plays an important role in the evaluation of the Kalman filter. Since the measurement noise is device-dependent, different hardware platforms have different noise characteristics, and the theoretical derivation of these covariance matrices May not be accurate enough for all platforms.

The direction of deep learning focuses on the direct modeling, analysis and prediction of the data transmitted by in-vehicle communication, which can avoid the problem of noise modeling and focus on the data itself. At the same time, it can effectively solve the nonlinear problem that the control direction cannot solve. Commonly used are BP (Back Propagation, Back Propagation) model and RNN (Recurrent Neural Network, Recurrent Neural Network) model. The BP neural network model processes the input data by dividing the input layer, the hidden layer and the output layer. The weights in the neural network are continuously updated and optimized through feedback intermediaries. The output data approximates the actual value with arbitrary precision for the purpose of prediction. However, the problem faced by BP neural network in the field of vehicle trajectory prediction is that the number of input features is fixed, and historical trajectory data cannot be used. At the same time, BP neural network is based on an assumption that the input data is uniformly distributed and independent. But in order to calculate the motion state information of the vehicle, the input time-series data needs to be correlated. Therefore, the recurrent neural network RNN can be used to design the vehicle motion state prediction model. On the basis of the current moment input, the memory information from the previous moment can be added to make it have a certain memory, but RNN does have a disadvantage of gradient disappearance.

In order to solve the above technical problems, the embodiments of the present application provide a method, device, system and medium for delay compensation of motion state information of a connected vehicle. The method includes: acquiring the current motion state information and current delay information of the sending vehicle. The motion state information includes the current position, current speed and current acceleration, and the current delay information includes the delay time from the previous time corresponding to the current time to the current time; obtain and send the historical motion state information and the corresponding historical delay within the preset time period of the vehicle. Time information, historical motion status information includes the position at each moment, speed at each moment and acceleration at each moment, historical delay information includes the corresponding delay time from each moment to the previous moment at each moment; , Current delay information, historical motion state information and historical delay information are input into the long-term and short-term memory network model, so that the long-term and short-term memory network model outputs the predicted compensation information of the vehicle. The predicted compensation information includes the speed after the predicted delay, the predicted delay After calculating the acceleration and the predicted delay time, the predicted compensation displacement of the sending vehicle is calculated according to the predicted compensation information. Therefore, according to the long-term and short-term memory network model, the vehicle motion state information and delay information can be modeled and predicted to compensate for the vehicle motion state information during the time period of the vehicle communication delay, so as to reduce the impact of the vehicle communication delay. The purpose is to improve the safety of the operation of connected vehicles.

Exemplary method

Referring to FIG. 1 , a flowchart of a method for delay compensation of motion state information of a connected vehicle provided by an embodiment of the present application includes:

S101: Obtain current motion state information and current delay information of the sending vehicle; the current motion state information includes the current position, the current speed and the current acceleration; the current delay information includes the time from the previous time corresponding to the current time to the current time Delay time.

In the embodiment of the present application, the current motion state information and current delay information of the sending vehicle can be obtained; the current motion state information can include the current position, the current speed and the current acceleration; the current delay information can include the last time corresponding to the current time The delay time to the current moment.

Specifically, the data can come from the basic safety message set (Basic Safety Messages, BSM) and delay information sent by the sender vehicle through the v2x communication device. The BSM information includes the current position, current speed, and current acceleration of the sending vehicle at the time of sending. Key motion state information, this method focuses on data-driven models, making it possible to use recurrent neural networks to predict sequence information such as vehicle motion states. Subject to the change of vehicle motion state information such as road scene and vehicle speed within a certain range, it is possible to use the historical motion state information of the vehicle to predict its future motion state information.

It should be noted that the location information in the BSM is the latitude, longitude and height information, which does not change much within a certain period of time, and the latitude and longitude information cannot be directly embedded into the model for calculation, and coordinate transformation needs to be performed, which will increase the redundancy of the algorithm. This method does not directly use historical position information for trajectory prediction, but focuses on predicting motion state variables such as velocity and acceleration for delay compensation.

For example, a possible application scenario provided by this embodiment of the present application may be, as shown in FIG. 2 , the scenario includes a data processing terminal 101 sending vehicle j; a v2x communication device 102 sending vehicle j; receiving vehicle i The v2x communication device 103 and the data processing terminal 104 of the receiving vehicle i.

The sending vehicle j sends the motion state information at time t, and the sending vehicle j has shifted from the time t to t+1. At this time, the information about the sending vehicle j in the receiving vehicle i still stays at the time before t and t, so it needs to be delayed. time compensation to make up for the information in the receiving vehicle i about the sending vehicle j from time t to t+1.

The v2x communication device 103 of the receiving vehicle i receives the BSM information about the sending vehicle j at time k from the v2x communication device 102 of the sending vehicle j. The BSM information includes key information such as position, speed, acceleration, etc., and records the time to the previous time. The delay value at the current moment, and then the data processing terminal 104 that receives the vehicle i performs the delay compensation for the motion information of the vehicle j.

Optionally, information preprocessing can be performed on the current motion state information and the current delay information, and the data required for subsequent neural network training can be packaged. Specifically, the packaged current motion state information and the current delay information set can be: :

X _j (t) = {v _j (t), a _j (t), d _j (t)};

Among them, v _j (t) is the current speed, a _j (t) is the current acceleration, and d _j (t) is the delay time from the previous moment corresponding to the current moment to the current moment.

S102: Acquire historical motion state information and corresponding historical delay information within the preset duration of the sending vehicle; the historical motion state information includes the position at each moment, the speed at each moment, and the acceleration at each moment; the historical delay information The time information includes respective delay durations corresponding to each time to the previous time of each time.

In the embodiment of the present application, the historical motion state information and corresponding historical delay information within the preset time period of the vehicle can be obtained; the historical motion state information includes the position at each moment, the speed at each moment, and the acceleration at each moment; historical delay information The time information includes respective delay durations corresponding to each time to the previous time of each time.

Specifically, X _j (t-1) may be the set of historical motion state information at time t-1 and the corresponding historical delay information, X _j (t-1)={v _j (t-1), a _j (t-1), d _j (t-1)}; where v _j (t-1) is the velocity at the historical time t-1, a _j (t-1) is the acceleration at the historical time t-1, d _j (t-1) is the delay time from time t-1 to time t-2 in the history.

Thus, a total set containing current motion state information, current delay information, historical motion state information and corresponding historical delay information can be formed:

X={ _Xj (t), _Xj (t-1)...}.

It should be noted that the speed and acceleration information here have two directions, horizontal and vertical, which are used to calculate the lateral and vertical displacements of the vehicle respectively. For the convenience of description, they are all represented by a symbol.

S103: Input the current motion state information, the current delay information, the historical motion state information, and the historical delay information into a long-term and short-term memory network model, so that the long-term and short-term memory network model outputs the sending vehicle The predicted compensation information includes the predicted speed after delay, the predicted acceleration after delay, and the predicted delay time.

S104: Calculate and obtain the predicted compensation displacement of the sending vehicle according to the predicted compensation information.

In this embodiment of the present application, current motion state information, current delay information, historical motion state information, and historical delay information may be input into a long short-term memory network model (LSTM, Long Short-Term Memory).

Specifically, as shown in FIG. 3 , which is a structure diagram of each LSTM unit node provided in this embodiment of the application, at each time step t, g(x _t ) is the input of the LSTM subunit, and h _t is the LSTM The output of the element node. Combine the unit node output h _t-1 of the previous time step with the current system input g(x _t ) to form the input of the current unit. The state of each cell node is C _t , which records system memory. _Ct is updated at each time step. To control the flow of information through the cell nodes, several gates are applied, including an input gate (it ), an output gate (o _t ), and a forget gate ( _{f t )} . Each gate produces an output between 0 and 1, where the value of the output is computed by a sigmoid(σ) function. An output of 0 means that the input to the gate is completely blocked, while an output of 1 means that all information of the input is kept in the cell node. The input, output and forget gate are calculated as follows:

i _t =σ(W _i g(x _t )+U _i h _t-1 +b _i )

o _t =σ(W _o g(x _t )+U _o h _t-1 +b _o )

f _t =σ(W _f g(x _t )+U _f h _t-1 +b _f );

Among them, W _i , W _o and W _f are the weights of the three gates; U _i , U _o and U _f are the corresponding cycle weights; b _i , b _o and b _f are the bias values of the three gates.

Similar to the gate function, the cell state is updated by combining the current input and the previous cell state C _t-1 . The difference is that the input will be processed by a hyperbolic tangent function instead of a sigmoid, which produces an output between -1 and 1:

乘以输入门的输出，然后用作更新单元格状态的第一个组件。用于更新单元状态的另一个组件是前一个单元状态，它由遗忘门处理，以确定如何使用过去的数据。对于这两个组件，t时的单元格状态将更新为：Updated,

The output of the input gate is multiplied and used as the first component to update the cell state. Another component used to update the cell state is the previous cell state, which is processed by a forget gate to determine how to use past data. For both components, the cell state at time t will be updated to:

Calculate the output of the unit h _t used at t+1 by multiplying the output gate with the tanh function of the current unit state:

h _t =o _t tanh(c _t ).

In a possible implementation, in order to improve the accuracy of model prediction, a semi-supervised learning strategy can be adopted to obtain a training set of the long short-term memory network model. The training set includes: the actual motion state information at the known time, the known time The actual delay information, the actual motion state information at the next moment at the known moment, and the actual delay information at the next moment at the known moment.

The actual motion state information at the known time includes the actual position, actual speed and actual acceleration at the known time; the actual delay information at the known time includes the actual delay time from the previous time corresponding to the known time to the known time; The actual motion state information at the next moment at the known moment includes the actual position, actual speed and actual acceleration at the next moment at the known moment; the actual delay information at the next moment at the known moment includes the time from the known moment to the known moment. The actual delay time at the next moment;

Use the training set to learn the motion state information at the known time and the actual delay information at the known time, and the mapping between the actual motion state information at the next time at the known time and the actual delay information at the next time at the known time relation;

The model parameters of the long and short-term memory network model are determined according to the mapping relationship, so that the state information and delay information of several historical moments are used as training data, and the state information of the next moment is used as data labels for training to improve the accuracy of model prediction.

In a possible implementation, as shown in FIG. 4 , in order to improve the prediction stability and performance of the long-term and short-term memory network, the current motion state information, current delay information, historical motion state information and historical delay information can be After encoding, it is input to the long short-term memory network model.

Specifically, the encoding includes performing unified standardization processing on the current motion state information, the current delay information, the historical motion state information and the historical delay information. Then train and use the network model, which helps to improve the stability and performance of the network, and the encoding can be performed by using an encoder.

In addition, the decoder can also be used to de-normalize the data according to the relevant coding of the encoder, so as to achieve the purpose of data scaling.

Optionally, use the following criteria:

是输入数据的第n个分量，如t时刻的延时、速度或加速度，亦即x_t，n的标准化输入，μ_N、σ_N是根据总样本得出的均值与方差。in

is the nth component of the input data, such as the delay, velocity or acceleration at time t, that is, the standardized input of x _{t, n} , μ _N , σ _N are the mean and variance obtained from the total sample.

Referring to FIG. 5 , the relationship between historical information and predicted future information under the network architecture of the method provided by the embodiment of the present application is shown, and the input Input at each time T _t-h+1 to T _t is encoded and then input to the LSTM model, and finally decode and output to predict the prediction compensation information Output at time T _t+1 in the future.

The length of the input sequence, i.e. the historical time step, is an important factor affecting the prediction performance. According to the inventor's preliminary experiments, the length of 5 to 15 is suitable.

It should be noted that the sequences used in this method for neural network training are of equal step size, but the motion state information in the sequence is the result of changes after a number of unequal delay values have occurred. This is also not done directly with position information. One of the reasons for trajectory prediction.

Specifically, for the LSTM model, see Figure 6, from the input layer (Input Layer) to the fully connected layer (Full connect Layer) to the LSTM layer to the LSTM layer to the fully connected layer, and finally do regression prediction ( Regression). In parentheses is the number of neurons in each sublayer.

Optionally, taking the uniformly variable speed motion as an example, a detailed description of the compensation motion state information is given.

The state information after prediction delay is: X _j (t+1)={v _j (t+1), a _j (t+1), d _j (t+1)}

Among them, X _j (t+1) is the information set after the delay, v _j (t+1) is the predicted speed after the delay, a _j (t+1) is the predicted acceleration, d _j (t+ 1) is the predicted delay value from time t to time t+1.

That is, the predicted compensation displacement of the sending vehicle can be calculated according to the predicted compensation information. Specifically, the predicted compensation displacement Δp of the sending vehicle is equal to the product of the predicted delay speed and the predicted delay time.

Δp=v _j (t+1)*d _j (t+1);

Predicted speed after delay: v _j (t+1)=a _j (t+1)*d _j (t+1).

An embodiment of the present application provides a method for delay compensation of motion state information of a connected vehicle. The method includes: acquiring current motion state information and current delay information of a sending vehicle, where the current motion state information includes current position, current speed, and current acceleration , the current delay information includes the delay time from the previous time corresponding to the current time to the current time; obtain the historical motion state information and the corresponding historical delay information within the preset time period of the sending vehicle, and the historical motion state information includes the position of each moment , the speed at each moment and the acceleration at each moment, the historical delay information includes the corresponding delay time from each moment to the previous moment at each moment; the current motion state information, current delay information, historical motion state information and history The delay information is input to the long-term and short-term memory network model, so that the long-term and short-term memory network model outputs the predicted compensation information of the sent vehicle. The predicted compensation information includes the predicted delay speed, the predicted acceleration after the delay, and the predicted delay time. According to the predicted compensation The information is calculated to obtain the predicted compensated displacement of the sending vehicle. Therefore, according to the long-term and short-term memory network model, the vehicle motion state information and delay information can be modeled and predicted to compensate for the vehicle motion state information during the time period of the vehicle communication delay, so as to reduce the impact of the vehicle communication delay. The purpose is to improve the safety of the operation of connected vehicles.

Exemplary device

Referring to FIG. 2 , a device for delay compensation of motion state information of a connected vehicle provided by an embodiment of the present application includes:

The first obtaining unit 201 is used to obtain the current motion state information and current delay information of the sending vehicle; the current motion state information includes the current position, the current speed and the current acceleration; the current delay information includes the upper The delay time from one moment to the current moment;

The second obtaining unit 202 is configured to obtain the historical motion state information and the corresponding historical delay information within the preset duration of the sending vehicle; the historical motion state information includes the position of each moment, the speed of each moment and the speed of each moment. acceleration; the historical delay information includes the respective delay durations from each moment to the previous moment of each moment;

The input unit 203 is configured to input the current motion state information, the current delay information, the historical motion state information and the historical delay information into the long-term and short-term memory network model, so that the long-term and short-term memory network model outputs The predicted compensation information of the sending vehicle; the predicted compensation information includes the predicted speed after delay, the predicted acceleration after delay, and the predicted delay time;

The calculating unit 204 is configured to calculate and obtain the predicted compensation displacement of the sending vehicle according to the predicted compensation information.

In a possible implementation, the apparatus further includes:

The third obtaining unit is configured to obtain a training set of the long short-term memory network model, the training set includes: actual motion state information at a known moment, actual delay information at a known moment, and the next moment at a known moment The actual motion state information and the actual delay information of the next moment at the known moment;

The actual motion state information at the known time includes the actual position, actual speed and actual acceleration at the known time; the actual delay information at the known time includes the actual delay from the previous time corresponding to the known time to the known time. duration; the actual motion state information at the next moment at the known moment includes the actual position, actual speed and actual acceleration at the next moment at the known moment; the actual delay information at the next moment at the known moment includes The actual delay time from the known moment to the next moment from the known moment;

The learning unit is configured to use the training set to learn the motion state information of the known time and the actual delay information of the known time, and the actual motion state information of the next time of the known time and the already known time. The mapping relationship of the actual delay information at the next moment at the known moment;

A determining unit, configured to determine model parameters of the long short-term memory network model according to the mapping relationship.

In a possible implementation manner, the input unit is specifically used for:

The current motion state information, the current delay information, the historical motion state information and the historical delay information are encoded and input into the long short-term memory network model;

The encoding includes uniformly standardizing the current motion state information, the current delay information, the historical motion state information, and the historical delay information.

In a possible implementation manner, the computing unit is specifically used for:

The predicted compensated displacement of the sending vehicle is equal to the product of the predicted delayed speed and the predicted delayed duration.

An embodiment of the present application provides a device for compensating for motion state information delay of a connected vehicle. A method for using the device includes: acquiring current motion state information and current delay information of a sending vehicle, where the current motion state information includes current position, current speed and the current acceleration, the current delay information includes the delay time from the previous time corresponding to the current time to the current time; obtain the historical motion state information and the corresponding historical delay information within the preset time period of the sending vehicle, and the historical motion state information includes each The position of the moment, the speed of each moment and the acceleration of each moment, the historical delay information includes the corresponding delay time from each moment to the previous moment of each moment; the current motion status information, current delay information, historical motion status The information and historical delay information are input into the long-term and short-term memory network model, so that the long-term and short-term memory network model outputs and sends the predicted compensation information of the vehicle. The predicted compensation information includes the predicted delay speed, the predicted acceleration after the delay, and the predicted delay time. The predicted compensation displacement of the sending vehicle is calculated according to the predicted compensation information. Therefore, according to the long-term and short-term memory network model, the vehicle motion state information and delay information can be modeled and predicted to compensate for the vehicle motion state information during the time period of the vehicle communication delay, so as to reduce the impact of the vehicle communication delay. The purpose is to improve the safety of the operation of connected vehicles.

On the basis of the foregoing embodiments, the embodiments of the present application provide a system for delay compensation of motion state information of connected vehicles, including:

memory for storing computer programs;

The processor is configured to implement the steps of the above-mentioned method for compensating for the delay of motion state information of a connected vehicle when executing the computer program.

On the basis of the foregoing embodiments, the embodiments of the present application further provide a computer-readable medium, where a computer program is stored on the computer-readable medium, and when the computer program is processed and executed, the motion state of the connected vehicle as described above is realized. The steps of the information delay compensation method.

It should be noted that the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples of computer readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Programmable read only memory (EPROM or flash memory), fiber optics, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. In this disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present disclosure, however, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave with computer-readable program code embodied thereon. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device . Program code embodied on a computer readable medium may be transmitted using any suitable medium including, but not limited to, electrical wire, optical fiber cable, RF (radio frequency), etc., or any suitable combination of the foregoing.

The above-mentioned computer-readable medium may be included in the above-mentioned system; or may exist alone without being assembled into the system.

In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a non-transitory computer readable medium, the computer program containing program code for performing the method illustrated in the flowchart.

Each embodiment in this specification is described in a progressive manner, and the same and similar parts between the various embodiments may be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the apparatus embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and reference may be made to the partial description of the method embodiment for related parts.

The above descriptions are only the preferred embodiments of the present application. Although the present application has been disclosed above with preferred embodiments, it is not intended to limit the present application. Any person skilled in the art, without departing from the scope of the technical solution of the present application, can use the methods and technical contents disclosed above to make many possible changes and modifications to the technical solution of the present application, or be modified into equivalents of equivalent changes. Example. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments according to the technical essence of the present application without departing from the content of the technical solutions of the present application still fall within the protection scope of the technical solutions of the present application.

Claims (10)

1. A method for compensating time delay for networked vehicle motion state information, characterized in that, comprising: Obtain the current motion state information and current delay information of the sending vehicle; the current motion state information includes the current position, the current speed and the current acceleration; the current delay information includes the delay from the previous time corresponding to the current time to the current time duration; Obtain the historical motion state information and corresponding historical delay information within the preset duration of the sending vehicle; the historical motion state information includes the position at each moment, the speed at each moment and the acceleration at each moment; the historical delay information Including the corresponding delay time from each moment to the previous moment of each moment; Inputting the current motion state information, the current delay information, the historical motion state information, and the historical delay information into a long-term and short-term memory network model, so that the long-term and short-term memory network model outputs the prediction of the sending vehicle Compensation information; the predicted compensation information includes the predicted speed after delay, the predicted acceleration after delay, and the predicted delay time; The predicted compensation displacement of the sending vehicle is obtained by calculating according to the predicted compensation information. 2. The method according to claim 1, wherein the current motion state information, the current delay information, the historical motion state information and the historical delay information are input into long short-term memory Before the network model, the method further includes: Acquire a training set of the long short-term memory network model, the training set includes: actual motion state information at a known time, actual delay information at a known time, actual motion state information at the next time at a known time, and past information. The actual delay information of the next moment after the known moment; The actual motion state information at the known time includes the actual position, actual speed and actual acceleration at the known time; the actual delay information at the known time includes the actual delay from the previous time corresponding to the known time to the known time. duration; the actual motion state information at the next moment at the known moment includes the actual position, actual speed and actual acceleration at the next moment at the known moment; the actual delay information at the next moment at the known moment includes The actual delay time from the known moment to the next moment from the known moment; Use the training set to learn the motion state information of the known time and the actual delay information of the known time, and the actual motion state information of the next time of the known time and the next time of the known time The mapping relationship of the actual delay information at the moment; The model parameters of the long short-term memory network model are determined according to the mapping relationship. 3. The method according to claim 1, wherein the current motion state information, the current delay information, the historical motion state information and the historical delay information are input into a long short-term memory network model, include: The current motion state information, the current delay information, the historical motion state information and the historical delay information are encoded and input into the long short-term memory network model; The encoding includes uniformly standardizing the current motion state information, the current delay information, the historical motion state information, and the historical delay information. 4 . The method according to claim 1 , wherein the calculating and obtaining the predicted compensation displacement of the sending vehicle according to the predicted compensation information comprises: 5 . The predicted compensated displacement of the sending vehicle is equal to the product of the predicted delayed speed and the predicted delayed duration. 5. A delay compensation device for motion state information of a networked vehicle, characterized in that it comprises: The first obtaining unit is used to obtain the current motion state information and current delay information of the sending vehicle; the current motion state information includes the current position, the current speed and the current acceleration; the current delay information includes the last time corresponding to the current moment. The delay time from time to current time; The second acquiring unit is configured to acquire the historical motion state information and the corresponding historical delay information within the preset duration of the sending vehicle; the historical motion state information includes the position at each moment, the speed at each moment and the acceleration at each moment ; Described historical delay information includes the respective corresponding delay time lengths from each moment to the previous moment of each moment; The input unit is configured to input the current motion state information, the current delay information, the historical motion state information and the historical delay information into the long-term and short-term memory network model, so that the long-term and short-term memory network model outputs all the information. The predicted compensation information of the sending vehicle; the predicted compensation information includes the predicted delayed speed, the predicted delayed acceleration and the predicted delayed time length; A calculation unit, configured to calculate and obtain the predicted compensation displacement of the sending vehicle according to the predicted compensation information. 6. The apparatus according to claim 5, wherein the apparatus further comprises: The third obtaining unit is configured to obtain a training set of the long short-term memory network model, the training set includes: actual motion state information at a known moment, actual delay information at a known moment, and the next moment at a known moment The actual motion state information and the actual delay information of the next moment at the known moment; The actual motion state information at the known time includes the actual position, actual speed and actual acceleration at the known time; the actual delay information at the known time includes the actual delay from the previous time corresponding to the known time to the known time. duration; the actual motion state information at the next moment at the known moment includes the actual position, actual speed and actual acceleration at the next moment at the known moment; the actual delay information at the next moment at the known moment includes The actual delay time from the known moment to the next moment from the known moment; The learning unit is configured to use the training set to learn the motion state information of the known time and the actual delay information of the known time, and the actual motion state information of the next time of the known time and the already known time. The mapping relationship of the actual delay information at the next moment at the known moment; A determining unit, configured to determine model parameters of the long short-term memory network model according to the mapping relationship. 7. The device according to claim 5, wherein the input unit is specifically used for: The current motion state information, the current delay information, the historical motion state information and the historical delay information are encoded and then input into the long short-term memory network model; The encoding includes uniformly standardizing the current motion state information, the current delay information, the historical motion state information, and the historical delay information. 8. The device according to claim 5, wherein the computing unit is specifically used for: The predicted compensated displacement of the sending vehicle is equal to the product of the predicted delayed speed and the predicted delayed duration. 9. A networked vehicle motion state information delay compensation system, characterized in that it comprises: memory for storing computer programs; The processor is configured to implement the steps of the method for delay compensation of motion state information of a connected vehicle according to any one of claims 1-4 when executing the computer program. 10. A computer-readable medium, characterized in that, a computer program is stored on the computer-readable medium, and when the computer program is processed and executed, the motion state of the connected vehicle according to any one of claims 1-4 is realized. The steps of the information delay compensation method.

Images