Cyber-physical twin probabilistic interactive prediction model identification method

Abstract

The application relates to the technical field of automatic driving application, in particular to an information physical twin probabilistic vehicle interaction prediction model identification method, which comprises the following steps: in the establishment stage of the probabilistic vehicle interaction prediction model, preset data sets are used to fit the probabilistic vehicle interaction prediction model parameters, so that the initialized probabilistic vehicle interaction prediction model parameters and the intention parameters of the target vehicle after initialization are obtained; in the running stage of the probabilistic vehicle interaction prediction model, based on a preset parameter and intention updating algorithm and a preset updating trigger mechanism, real-time sensing data are used to update the probabilistic vehicle interaction prediction model parameters, the intention of the target vehicle and uncertainty. Therefore, the problem that the prior art cannot effectively model and update the complex uncertainty in long-time domain interaction prediction is solved, the vehicle interaction behavior can be predicted in a longer time domain, and definite interaction basis is provided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic driving application, and in particular to an information-physical twin probabilistic surrounding vehicle interactive prediction model identification method. BACKGROUND

[0002] Vehicle-road-cloud integration and intelligent cyber-physical system (IVCPS) based thereon are a new way to solve the above problems. Vehicle-road-cloud integration fusion control can support and supplement single vehicle automatic driving, and use the advantages of wide-area perception based on roadside in cloud control system, large parallel computing power that can be scheduled by cloud server in a short time, and centralized decision-making for coordination of multiple vehicles or even entire intelligent connected vehicle traffic flow to assist single vehicle to make more energy-efficient decisions and planning.

[0003] Under the vehicle-road-cloud integration architecture, real-time twins of traffic conditions and all traffic participants need to be established in the cloud for more long-term predictive information-based applications, so as to better protect the safety of vehicles and make better decisions. However, the current cloud twin establishment is generally deterministic, or only contains simple probability information, and it is still difficult to model and real-time adjust the complex uncertainty and its propagation in long-term interactive prediction. Therefore, how to establish a model that can simply and effectively represent long-term interactive prediction, and real-time update the uncertainty of model parameters and intentions, has become a problem to be solved.

[0004] With the rapid development of automatic driving technology, behavior decision-making has become an indispensable important link in realizing automatic driving function. The key to behavior decision-making is how to effectively handle the interaction between the ego vehicle and surrounding vehicles. The traditional rule-based scheme has the logical fallacy of predicting first and then deciding, because the prediction results of surrounding vehicles should be different under different decisions of the ego vehicle. At present, single vehicle automatic driving is mainly divided into two categories: modular automatic driving and end-to-end automatic driving. Among them, the latest modular automatic driving scheme handles the interaction between the ego vehicle and surrounding vehicles through multiple strategies such as explicit interaction, implicit interaction, and comprehensive interaction; while end-to-end automatic driving solves the interaction problem through overall collaborative optimization from perception to planning results, and multi-objective, multi-modal, etc.

[0005] Although the above method performs well in handling the interaction between the ego vehicle and the surrounding vehicles in a short time domain, single-vehicle automatic driving is usually limited by the ego vehicle's perception and computing limitations, and still easily falls into a local optimal decision, making it difficult to consider the dynamic traffic conditions in a longer time domain to make a better decision and planning. The reason for this limitation is that the ego vehicle's perception range is short, making it difficult to obtain perception results over a longer distance. Even if a larger range of perception is obtained through a cloud information platform such as vehicle-road cooperation or map navigation, it is limited by the small computing power of the vehicle and the difficulty in well representing the uncertainty changes of the surrounding vehicles' motion over time.

[0006] Vehicle-road cloud integration and the intelligent cyber-physical system (IVCPS) based thereon are a new way to solve the above problems. Vehicle-road cloud integration can support and supplement single-vehicle automatic driving, and use the advantages of cloud control systems, such as wide-area perception based on roadside, large parallel computing power that can be scheduled by cloud servers for a short time, and centralized decision-making that can be coordinated with multiple vehicles or even the entire intelligent connected vehicle traffic flow, to assist single vehicles in making more energy-efficient and efficient decisions and planning.

[0007] Under the vehicle-road cloud integration architecture, real-time twins of traffic conditions and all traffic participants need to be established in the cloud to implement more long-time domain predictive information-based applications, thereby better protecting the safety of vehicles and making better decisions. However, the current cloud twin establishment is generally deterministic, or only contains simple probability information, and it is still difficult to model and real-time adjust the complex uncertainty and its propagation in long-time domain interactive prediction. Therefore, how to establish a model that can simply and effectively represent long-time domain interactive prediction, and real-time update the uncertainty of model parameters and intentions, has become a problem to be solved. SUMMARY

[0008] The present application provides an information physical twin probabilistic interactive prediction model identification method to solve the problem that the prior art cannot effectively model and real-time update the complex uncertainty in long-time domain interactive prediction, which can predict the interactive behavior of surrounding vehicles in a longer time domain while providing clear interaction basis.

[0009] The first aspect embodiment of the present application provides an information physical twin probabilistic interactive prediction model identification method, comprising the following steps:

[0010] obtaining real-time perception data of a target vehicle;

[0011] In the establishment stage of the probabilistic vehicle interaction prediction model, preset data sets are used to fit the probabilistic vehicle interaction prediction model parameters, to obtain initialized probabilistic vehicle interaction prediction model parameters and intention parameters of the target vehicle after initialization, wherein the probabilistic vehicle interaction prediction model includes a driver model, a probabilistic lane changing model and an uncertainty accumulation mechanism, and the probabilistic vehicle interaction prediction model parameters include driver model parameters and probabilistic lane changing model parameters.

[0012] In the running stage of the probabilistic vehicle interaction prediction model, based on a preset parameter and intention update algorithm and a preset update triggering mechanism, the real-time perception data are used to update the probabilistic vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty.

[0013] According to an embodiment of the present application, the fitting of the probabilistic vehicle interaction prediction model parameters using the preset data sets to obtain the initialized probabilistic vehicle interaction prediction model parameters and the intention parameters of the target vehicle after initialization includes:

[0014] Based on a weighted two-norm loss function, the driver model parameters are fitted, and based on a cross-entropy loss function, the probabilistic lane changing model parameters are fitted, to obtain the initialized probabilistic vehicle interaction prediction model parameters and the intention parameters of the target vehicle after initialization.

[0015] According to an embodiment of the present application, the updating of the probabilistic vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty using the real-time perception data based on the preset parameter and intention update algorithm and the preset update triggering mechanism includes:

[0016] Based on a preset posterior distribution calculation method, the driver model parameters and the intention parameters are jointly sampled, and a first posterior probability is calculated according to a first sampling result;

[0017] Based on a preset posterior distribution calculation method, the probabilistic lane changing model parameters are separately sampled, and a second posterior probability is calculated according to a second sampling result;

[0018] The probabilistic vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty are updated according to the first posterior probability and / or the second posterior probability;

[0019] The calculation formula of the first posterior probability is:

[0020]

[0021] wherein, is the sampling time, is based on the driver model parameters and the intention The predicted state, is the actual observed state, is the prior variance between the predicted state and the observed state, are the parameters of the driver model, The driver's intention;

[0022] The calculation formula of the second posterior probability is:

[0023]

[0024] in, is the number of samples, For the The probability of left lane change predicted by the model under the premise of samples, For the Based on the parameters of the sample The calculated probability of changing lanes left is, For the The probability of not changing lanes predicted by the model under the premise of samples is, For the Based on the parameters of the sample The calculated probability of not changing lanes is For the The probability of right lane change predicted by the model under the premise of samples, For the Based on the parameters of the sample The calculated probability of right lane change is are the parameters of the probabilistic weekly vehicle interaction prediction model adopted.

[0025] According to one embodiment of the present application, the updating of the probabilistic vehicle-to-vehicle interaction prediction model parameters, the target vehicle's intention, and the uncertainty using the real-time perception data based on the preset parameter and intention update algorithm and the preset update trigger mechanism further includes:

[0026] If the joint sampling of the driver model parameters and the intention parameters meets a preset time interval condition, the probabilistic vehicle-to-vehicle interaction prediction model parameters, the intention and uncertainty of the target vehicle are updated based on a time update trigger mechanism;

[0027] If the target vehicle meets the preset lane-changing conditions or the prediction result of the probabilistic vehicle-to-vehicle interaction prediction model meets the preset result, the parameters of the probabilistic vehicle-to-vehicle interaction prediction model, the intention and uncertainty of the target vehicle are updated based on an event update trigger mechanism.

[0028] According to one embodiment of the present application, before acquiring the real-time perception data of the target vehicle, further comprising:

[0029] A representation of longitudinal motion, lateral motion, and uncertainty accumulation is established for each vehicle to be interactively predicted.

[0030] According to the information physical twin probabilistic interactive prediction model identification method of the embodiment of the present application, in the establishment stage of the probabilistic interactive prediction model, the preset data set is used to fit the probabilistic interactive prediction model parameters, to obtain the initialized probabilistic interactive prediction model parameters and the intention parameters of the initialized target vehicle. In the running stage of the probabilistic interactive prediction model, based on the preset parameter and intention update algorithm and the preset update trigger mechanism, the real-time perception data is used to update the probabilistic interactive prediction model parameters, the intention of the target vehicle, and the uncertainty. Thus, the problem that the prior art is difficult to effectively model and update the complex uncertainty in long-time domain interactive prediction is solved, the interactive behavior of the vehicle can be predicted in a longer time domain, and clear interaction basis is provided.

[0031] The second aspect embodiment of the present application provides an information physical twin probabilistic interactive prediction model identification device, comprising:

[0032] The acquisition module is configured to acquire real-time perception data of a target vehicle.

[0033] The fitting module is configured to, in the establishment stage of the probabilistic interactive prediction model, use a preset data set to fit probabilistic interactive prediction model parameters, to obtain initialized probabilistic interactive prediction model parameters and intention parameters of an initialized target vehicle. The probabilistic interactive prediction model includes a driver model, a probabilistic lane changing model, and an uncertainty accumulation mechanism. The probabilistic interactive prediction model parameters include driver model parameters and probabilistic lane changing model parameters.

[0034] The update module is configured to, in the running stage of the probabilistic interactive prediction model, based on a preset parameter and intention update algorithm and a preset update trigger mechanism, use the real-time perception data to update the probabilistic interactive prediction model parameters, the intention of the target vehicle, and the uncertainty.

[0035] According to one embodiment of the present application, the fitting module is configured to:

[0036] Based on a weighted two-norm loss function, the driver model parameters are fitted, and based on a cross-entropy loss function, the probabilistic lane changing model parameters are fitted, to obtain the initialized probabilistic interactive prediction model parameters and the intention parameters of the initialized target vehicle.

[0037] According to one embodiment of the present application, the updating module is configured to:

[0038] jointly sample the driver model parameters and the intention parameters based on a preset posterior distribution calculation method, and calculate a first posterior probability according to a first sampling result;

[0039] individually sample the parameterized lane-changing model based on a preset posterior distribution calculation method, and calculate a second posterior probability according to a second sampling result;

[0040] update the parameterized car-following interaction prediction model, the intention and uncertainty of the target vehicle according to the first posterior probability and / or the second posterior probability;

[0041] wherein the calculation formula of the first posterior probability is:

[0042]

[0043] wherein, is a sampling time, is a state predicted based on the driver model parameters and the intention of the driver, is an actually observed state, is a prior variance of the predicted state and the observed state, is a parameter of the driver model, is an intention of the driver;

[0044] the calculation formula of the second posterior probability is:

[0045]

[0046] wherein, is a number of samples, is a probability of left lane-changing predicted by the model under the premise of the th sample, is a probability of left lane-changing calculated based on the parameter under the premise of the th sample, is a probability of no lane-changing predicted by the model under the premise of the th sample, is a probability of no lane-changing calculated based on the parameter under the premise of the th sample, is a probability of right lane-changing predicted by the model under the premise of the th sample, is a probability of right lane-changing calculated based on the parameter under the premise of the th sample. parameters of the probabilistic surrounding vehicle interaction prediction model.

[0047] According to an embodiment of the present application, the updating module is further configured to:

[0048] if joint sampling of the driver model parameters and the intention parameters meets a preset time interval condition, then updating the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty based on a time updating trigger mechanism;

[0049] if the target vehicle meets a preset lane changing condition or the prediction result of the probabilistic surrounding vehicle interaction prediction model meets a preset result, then updating the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty based on an event updating trigger mechanism.

[0050] According to an embodiment of the present application, before acquiring real-time perception data of the target vehicle, the acquiring module is further configured to:

[0051] establishing a representation of longitudinal motion, lateral motion and uncertainty accumulation for each vehicle to be subjected to interactive prediction.

[0052] The information-physical twin probabilistic surrounding vehicle interactive prediction model recognition device according to the embodiment of the present application, in the establishment stage of the probabilistic surrounding vehicle interaction prediction model, uses a preset data set to fit the probabilistic surrounding vehicle interaction prediction model parameters, to obtain initialized probabilistic surrounding vehicle interaction prediction model parameters and initialized intention parameters of the target vehicle, in the running stage of the probabilistic surrounding vehicle interaction prediction model, based on a preset parameter and intention updating algorithm and a preset updating trigger mechanism, using real-time perception data to update the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty. Thus, the problem that the prior art is difficult to effectively model and update the complex uncertainty in long-time domain interactive prediction is solved, and the surrounding vehicle interaction behavior prediction in a longer time domain can be performed, while providing clear interaction basis.

[0053] The third aspect embodiment of the present application provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the program to implement the information-physical twin probabilistic surrounding vehicle interactive prediction model recognition method as described in the above embodiments.

[0054] The fourth aspect embodiment of the present application provides a computer readable storage medium having a computer program stored thereon, which is executed by a processor to implement the information-physical twin probabilistic surrounding vehicle interactive prediction model recognition method as described in the above embodiments.

[0055] The fifth aspect of the present application provides a computer program product comprising a computer program which, when executed by a processor, is configured to implement the information-physical twin probabilistic interactive prediction model identification method according to the above-mentioned embodiments.

[0056] Additional aspects and advantages of the present application will be made apparent by the following description and the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0057] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description, taken in conjunction with the accompanying drawings, in which:

[0058] Figure 1 A structural schematic diagram of an information-physical twin probabilistic interactive prediction model identification system according to an embodiment of the present application;

[0059] Figure 2 A schematic diagram of a longitudinal driver model construction method according to an embodiment of the present application;

[0060] Figure 3 A schematic diagram of a lateral probabilistic lane change model construction method according to an embodiment of the present application;

[0061] Figure 4 A schematic diagram of an illustration of the propagation of uncertainty according to an embodiment of the present application;

[0062] Figure 5 A flowchart of an information-physical twin probabilistic interactive prediction model identification method according to an embodiment of the present application;

[0063] Figure 6 A schematic diagram of a parameter uncertainty model establishment stage identification method according to an embodiment of the present application;

[0064] Figure 7 A schematic diagram of an intention and its uncertainty model establishment stage identification method according to an embodiment of the present application;

[0065] Figure 8 A schematic diagram of a parameter and intention uncertainty model running stage identification method according to an embodiment of the present application;

[0066] Figure 9 A block schematic diagram of an information-physical twin probabilistic interactive prediction model identification apparatus according to an embodiment of the present application;

[0067] Figure 10 A structural schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0068] Embodiments of the present application are described below in detail, examples of which are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0069] The information physical twin probabilistic surrounding vehicle interactive prediction model identification method according to the embodiments of the present application is described below with reference to the drawings.

[0070] Before introducing the information physical twin probabilistic surrounding vehicle interactive prediction model identification method of the present application, first introduce the information physical twin probabilistic surrounding vehicle prediction model establishment and identification system and the information physical twin probabilistic surrounding vehicle interactive prediction model establishment method involved in the information physical twin probabilistic surrounding vehicle interactive prediction model identification method of the present application, wherein the probabilistic surrounding vehicle interactive prediction model establishment and identification method used by the system is the probabilistic surrounding vehicle interactive prediction model establishment and identification method as proposed in the present application, and the specific implementation steps of the method will be described in detail below.

[0071] First, in order to realize the establishment and identification of the information physical twin probabilistic surrounding vehicle interactive prediction model, the present disclosure proposes an information physical twin probabilistic surrounding vehicle interactive prediction model establishment and identification system based on the vehicle-road cloud integrated architecture, as shown in Figure 1 The system includes an information space and a physical space, and the physical space contains user vehicles using services and other traffic participants; the information space contains an information mapping layer and a fusion application layer, both of which are distributed in the vehicle end and the cloud end. Among them, in the information mapping layer, the cloud end contains the digital twin of the current traffic participant; the fusion application layer is distributed in the vehicle end controller and the cloud end server, and the cloud end fusion application layer can establish and identify the surrounding vehicle probability model and be used in other applications.

[0072] Further, the information physical twin probabilistic surrounding vehicle interactive prediction model establishment and identification system of the embodiments of the present application further includes roadside devices, intelligent connected vehicles, a cloud end and users. Among them, the roadside devices include environment perception function and communication function, can identify the global ID of the vehicle on the road and the state information of the vehicle and transmit the information; the number of intelligent connected vehicles is arbitrary, which contains environment perception function and communication function, can perceive the surrounding environment and transmit the perception raw data or perception target level result; the cloud end contains traffic real-time twin and historical database, the traffic real-time twin is obtained by fusing and perceiving the perception data uploaded by the roadside devices and intelligent connected vehicles, and the historical database is obtained by recording and storing the traffic real-time twin; the user can communicate with the cloud end, upload the interested area, and obtain the probabilistic surrounding vehicle interactive prediction model establishment and identification service provided by the cloud end.

[0073] Secondly, the method for establishing the information-physical twin probabilistic weekly vehicle interactive prediction model of this application is introduced.

[0074] Specifically, the cyber-physical twin probabilistic weekly vehicle interactive prediction model establishes a representation of longitudinal motion, lateral motion and uncertainty accumulation for each vehicle that needs to be interactively predicted. The model for one of the vehicles specifically includes: for longitudinal motion, a rule-based model is used and its parameters are considered to conform to the normal distribution; for lateral motion, a model is used to output the probability of predefined semantic behavior decisions (left lane change, right lane change, lane keeping); the uncertainty accumulation mechanism takes into account parameter uncertainty, lane change uncertainty of the vehicle in front of the vehicle, and uncertainty of the vehicle's intention.

[0075] Among them, the rule-based model is the driver model, and the driver model is continuously and partially differentiable everywhere. A modified intelligent driver model (IDM) can be used, which has the form:

[0076] (1)

[0077] (2)

[0078] in, For the ideal following distance, is the minimum distance, is the vehicle speed, is the expected headway, is the speed of the preceding vehicle, is the maximum acceleration, For comfortable deceleration, For the driver model, is the expected speed, is the driver's aggressiveness, is the actual distance.

[0079] in, The function is:

[0080] (3)

[0081] like Figure 2 As shown, use Alternative The IDM model can be made continuously differentiable everywhere, so as to obtain the analytical gradient of each parameter. To include Expected headway, Maximum acceleration, Comfortable deceleration, Minimum distance, Desired speed, driver aggressiveness , composed of vectors.

[0082] Furthermore, a model that outputs the probability of predefined semantic behavior decisions is called a probabilistic lane change model. This model outputs the probabilities of different lane change decisions. Specifically, it is a modification of the MOBIL model. Using the widely recognized MOBIL model as a foundation, the process of defining the probabilities of a vehicle's L, K, and R (left lane change, lane keeping, and right lane change) involves the following steps:

[0083] Step 1: Use the MOBIL model to obtain the lane change acceleration vector ,Include (acceleration gain of the vehicle changing lanes to the left), (acceleration gain of the ego vehicle without changing lanes), (acceleration gain of the vehicle changing lanes to the right), (the algebraic sum of the acceleration gains of the vehicle behind it in the original lane and the new vehicle behind it in the left lane after the vehicle changes lanes to the left), (the benefits of other vehicles when the vehicle keeps its lane), (The algebraic sum of the acceleration gains of the vehicle behind the ego vehicle and the new vehicle behind the ego vehicle in the right lane after the ego vehicle changes lanes to the right) The expected acceleration gain of the ego vehicle and the following vehicle after the lane change corresponding to the ego vehicle's LKR action (calculated using the identified driver model of each vehicle). For the ego vehicle, this is the difference between the acceleration after the lane change and the acceleration before the lane change. For the following vehicle, this is the sum of the loss to the new following vehicle after the lane change due to the ego vehicle's lane change and the gain to the original following vehicle before the lane change due to the ego vehicle's lane change.

[0084] Step 2: Design evaluation function: Evaluation function It needs to be able to reflect the vehicle's preference for different lane-changing options. Generally, the greater the acceleration, the more likely the vehicle is to choose that behavior. It can be defined as:

[0085] (4)

[0086] in, There are 5 parameters in total, which are model parameters to be identified. The common parameters are It is the politeness coefficient corresponding to the MOBIL model, reflecting the importance that the car attaches to the impact of its own car on other cars.

[0087] Step 3: Get the lane-changing probability of the car, using The function normalizes the output of the evaluation function to obtain the probability of changing lanes to the left or right or not changing lanes at this moment.

[0088] (5)

[0089] in, is the probability of changing lanes left, is the probability of changing lanes right, is the probability of not changing lanes.

[0090] Specifically, if Figure 3 As shown in Figure 1, this is equivalent to connecting the output of the MOBIL model to the fully connected layer and the Softmax layer, generating the probabilities of the three decisions of changing lanes left, right, and not changing lanes.

[0091] Furthermore, the parameter uncertainty considered by the uncertainty accumulation mechanism is the parameter uncertainty of the driver model. The lane-changing uncertainty of the vehicle in front of the vehicle is output through the lateral motion model of the probabilistic vehicle-to-vehicle interactive prediction model established for the vehicle in front of the vehicle. The uncertainty of the vehicle's intention is obtained by calculating the entropy value of the intention probability, and the above uncertainty is accumulated over time through the motion equation (including two state quantities: longitudinal speed and longitudinal position).

[0092] Specifically, the uncertainty accumulation mechanism of the embodiment of the present application is an additive uncertainty accumulation mechanism, and the uncertainty of the output acceleration prediction result is the weighted product of the uncertainty of the driver parameters of the vehicle, the uncertainty of the lane change of the preceding vehicle and the uncertainty of the state parameters of the two vehicles, and the sum of the uncertainty of the vehicle's intention.

[0093] The propagation of uncertainty between vehicles is considered by Taylor expanding the influence of the driver model and lane change model parameters on the model output results. Figure 4 As shown, the specific implementation is as follows:

[0094] The driver model can be expressed as:

[0095] (6)

[0096] in, is the output of the driver model, i.e. the predicted acceleration, For the driver model, is the speed of the preceding vehicle, is the vehicle speed, is the distance between the vehicle and the preceding vehicle, is the parameter vector that needs to be estimated.

[0097] Alternatively, it can be simplified to free flow driving when there is no vehicle ahead:

[0098] (7)

[0099] The parameters of the driver model will have variance when identified , through Taylor expansion, the uncertainty of the leading vehicle state and the driver parameters of the following vehicle The effect of uncertainty on speed can be approximated as:

[0100] (8)

[0101] where, is the variance of the predicted acceleration, is the variance of the lead vehicle speed, is the variance of the inter-vehicle distance, is the variance of the parameter is the variance of the parameter is the parameter is the parameter.

[0102] where the uncertainty of the inter-vehicle distance is the sum of the uncertainty of the two vehicle positions:

[0103] (9)

[0104] where, is the variance of the ego vehicle position, is the variance of the lead vehicle position.

[0105] Further, the longitudinal uncertainty including the uncertainty of lane changing can be calculated as follows. Since the lead vehicle can either change lane to leave ( ) or stay in the same lane ( ), the longitudinal uncertainty is calculated cumulatively as:

[0106] (10)

[0107] where the subscript is the lead vehicle of the lead vehicle, is the of the lead vehicle, is the inter-vehicle distance of the lead vehicle of the lead vehicle.

[0108] When considering the intention uncertainty, since it can be reasonably assumed that the intentions of different vehicles are independent of each other, the calculation of the uncertainty adds the product of the intention entropy. Let:

[0109] (11)

[0110] where, is the corrected variance of the lead vehicle speed, is the corrected inter-vehicle distance variance, is the ego vehicle parameter uncertainty.

[0111] then the following formula is true:

[0112] (12)

[0113] where, is the lane changing decision entropy of the lead vehicle, is the lane changing decision entropy of the ego vehicle.

[0114] where the intent entropy of a certain car is computed as follows, the intent of this car contains

[0115] (13)

[0116] The reason why the intents of different cars can be assumed to be independent of each other is that the macroscopic intent depends on the path planning of this car to the destination, which basically does not produce mutual influences between cars.

[0117] Further, in the absence of state updates, the propagation of uncertainty depends on the prediction step, which is achieved by the following steps: at each time step the state of the vehicle is predicted through an acceleration model, the state vector comprises the position and velocity, the state transition is as follows:

[0118] (14)

[0119] i.e.:

[0120] (15)

[0121] ZOH discretization is performed to obtain a discrete state equation:

[0122] (16)

[0123] where is the state quantity at time , is the state transition matrix, is the state quantity at time , is the predicted acceleration at time

[0124] At each time step, the uncertainty of the state propagates, and this uncertainty influences the uncertainty of the position and velocity through the state transition equation, therefore, the uncertainty propagation of the preceding car is described by the following equation:

[0125] (17)

[0126] where is the uncertainty at time , is the uncertainty at time

[0127] where is the process noise covariance matrix, which is defined as:​​​​​

[0128] (18)

[0129] where, denotes the effect of acceleration on position variance, since position is the second integral of acceleration, the variance is proportional to the fourth power of the sampling time. denotes the effect of acceleration on the unknown and velocity covariance, since velocity is the first integral of acceleration, and position is the second integral of acceleration, resulting in a cubic time dependence. denotes the dependence of velocity on the square of acceleration.

[0130] The information physical twin probabilistic interactive prediction model identification method proposed in the present application is described below.

[0131] Specifically, Figure 5 A flowchart of an information physical twin probabilistic interactive prediction model identification method provided by an embodiment of the present application is shown.

[0132] As Figure 5 shown, the information physical twin probabilistic interactive prediction model identification method includes the following steps:

[0133] In step S501, real-time perception data of a target vehicle is obtained.

[0134] Specifically, the real-time perception data of the target vehicle can be obtained by a vehicle sensor, which is not specifically limited here.

[0135] In step S502, in the establishment stage of the probabilistic interactive prediction model of surrounding vehicles, the probabilistic interactive prediction model parameters are fitted using a preset data set to obtain initialized probabilistic interactive prediction model parameters and initialized intention parameters of the target vehicle, wherein the probabilistic interactive prediction model of surrounding vehicles includes a driver model, a probabilistic lane changing model, and an uncertainty accumulation mechanism, and the probabilistic interactive prediction model parameters include driver model parameters and probabilistic lane changing model parameters.

[0136] The preset data set contains the historical driving state data of a certain or certain traffic vehicle and the interaction record with surrounding vehicles. The intention parameters of the initialized target vehicle include intention and probability.

[0137] Further, in some embodiments, the probabilistic surrounding vehicle interaction prediction model parameters are fitted with the preset data set to obtain initialized probabilistic surrounding vehicle interaction prediction model parameters and initialized intention parameters of the target vehicle, including: fitting the driver model parameters based on a weighted two-norm loss function, and fitting the probabilistic lane changing model parameters based on a cross-entropy loss function to obtain the initialized probabilistic surrounding vehicle interaction prediction model parameters and the initialized intention parameters of the target vehicle.

[0138] Specifically, as shown in FIG. 3, the parameter identification in the model establishment stage is implemented by fitting the average behavior in the data set. The weighted two-norm loss and the cross-entropy loss are used for the driver model and the lane changing model, respectively. Figure 6

[0139] The weighted two-norm loss function is as follows:

[0140] (19)

[0141] wherein, is the weighted two-norm loss, is the weight of the error between the predicted acceleration and the actual acceleration, is the acceleration output by the driver model, is the actually observed acceleration, is the weight of the error between the predicted speed obtained by integrating the predicted acceleration and the actually observed speed, is the predicted speed obtained by integrating the predicted acceleration, is the actually observed speed, is the weight of the error between the predicted position obtained by integrating the predicted speed and the actual position, is the predicted position obtained by integrating the predicted speed, is the actual position.

[0142] The cross-entropy loss function is as follows:

[0143] (20)

[0144] wherein, is the cross-entropy loss, is the probability of left lane changing predicted by the model, is the probability of actually left lane changing (since the actual behavior is determined to be observed, it is a non-zero quantity, , same), is the probability of no lane changing predicted by the model, is the probability of actually no lane changing, is the probability of right lane changing predicted by the model, ​The probability of actual right lane changing.

[0145] As Figure 7 shown, the intention recognition in the model establishment stage determines the intention branch point of the vehicle in the future given long distance through map information and traffic rules (K possible intentions are identified, such as currently driving on a highway, there is a ramp at 500m in front, and the possible intentions include straight driving and ramp; and based on the identified driver model, the vehicle is set to different intentions (each intention has its own prior information, which can be reasonably inferred and given through road structure, vehicle position, such as at the ramp, the traffic flow according to the usual traffic flow + current lane position of the vehicle + current acceleration trend of the vehicle can be given , ).

[0146] In step S503, in the running stage of the probabilistic surrounding vehicle interaction prediction model, based on the preset parameter and intention update algorithm and the preset update triggering mechanism, the real-time perception data is used to update the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty.

[0147] Further, in some embodiments, based on the preset parameter and intention update algorithm and the preset update triggering mechanism, the real-time perception data is used to update the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty, including: based on the preset posterior distribution calculation method, the driver model parameters and the intention parameters are jointly sampled, and the first posterior probability is calculated according to the first sampling result; based on the preset posterior distribution calculation method, the probabilistic lane changing model parameters are sampled alone, and the second posterior probability is calculated according to the second sampling result; the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty are updated according to the first posterior probability and / or the second posterior probability.

[0148] Specifically, in the model running stage of the information physical twin probabilistic surrounding vehicle interactive prediction model identification method, the parameter and intention update algorithm is Metropolis Hastings sampling, the driver model parameters and the intention parameters are jointly sampled according to the defined posterior distribution calculation method, and the lane changing probability model parameters are updated alone. As Figure 8 shown, the MCMC (Markov Chain Monte Carlo) method in Bayesian estimation is used for sampling to update the uncertainty of the driver model parameters and the lane changing model parameters, i.e. the intention. The MCMC method constructs a Markov chain to sample a series of from the posterior distribution to construct the probability distribution of the parameters. The distribution of the parameters can be estimated by the sample values obtained by sampling. As shown in the figure, the sample values obtained by sampling , we can use the sample mean as the estimated value of the parameter and the sample variance as the uncertainty of the parameter. The remaining question is how to obtain the posterior probability.

[0149] Further, if Figure 8 As shown, the preset posterior distribution calculation method is to sample the driver model parameters and intention parameters together, and a certain set of parameters and intention The posterior probability The calculation method is based on parameters and intention at N sampling moments (N≥1) Predicted status The error normal distribution probability product of Indicates the vehicle status, including at least the horizontal and vertical position and speed, is the prior value.

[0150] (twenty one)

[0151] in, is the sampling time, Based on the driver model parameters and intention The predicted state, is the actual observed state, is the prior variance between the predicted state and the observed state, are the parameters of the driver model, The driver's intention.

[0152] Furthermore, for the separate sampling of lane-changing model parameters, the parameters The posterior probability calculation method is based on the parameter The calculated probability of lane-changing decisions corresponds to the probability of the actual decision.

[0153] (twenty two)

[0154] in, is the number of samples, For the The probability of left lane change predicted by the model under the premise of samples, For the Based on the parameters of the sample The calculated probability of changing lanes left is, For the The probability of not changing lanes predicted by the model under the premise of samples is, For the Based on the parameters of the sample The calculated probability of not changing lanes is For the The probability of right lane change predicted by the model under the premise of samples, the probability of right lane changing calculated based on parameters of the first sample, parameters of the probabilistic surrounding vehicle interaction prediction model used.

[0155] Further, in some embodiments, based on the preset parameters, the intention updating algorithm and the preset updating trigger mechanism, the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty are updated using real-time perception data, and further comprising: if the joint sampling of the driver model parameters and the intention parameters satisfies the preset time interval condition, the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty are updated based on the time updating trigger mechanism; if the target vehicle satisfies the preset lane changing condition or the prediction result of the probabilistic surrounding vehicle interaction prediction model satisfies the preset result, the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty are updated based on the event updating trigger mechanism.

[0156] Specifically, the updating trigger mechanism is a time and time joint trigger mechanism, the updating mechanism of the joint sampling of the driver model parameters and the intention is a time trigger mechanism satisfying a certain time interval condition, and the parameter updating mechanism of the probabilistic lane changing model is an event trigger mechanism, which is updated when the vehicle changes lane or the model predicts that the vehicle changes lane but actually does not change lane.

[0157] Therefore, in the information physical twin probabilistic surrounding vehicle interactive prediction model identification method of the present application, the rule-based longitudinal and lateral model modeling is adopted in the model establishment method, and the probabilistic attribute is given to consider the uncertainty of the model parameters and the driver's intention, the uncertainty accumulation mechanism considers the model parameters, the front vehicle lane changing and the intention uncertainty, and the model is used for long-time domain interactive deduction; in the identification method, the model parameters are initialized by fitting the average value of the data set in the model establishment stage, and the model is updated by the parameter and intention updating algorithm in the model running stage. In the system, the cloud establishes the traffic real-time twin and historical database through the cooperative perception with the roadside perception equipment and the intelligent networked vehicle, and according to the user application, the prediction model is established and the parameters are updated for the vehicles in a specific area. The method and system provided by the present application provide long-time domain interactive deduction basic services for users in the cloud.

[0158] According to the information physical twin probabilistic interactive prediction model identification method of the embodiment of the present application, in the establishment stage of the probabilistic interactive prediction model, the preset data set is used to fit the probabilistic interactive prediction model parameters, and the initialized probabilistic interactive prediction model parameters and the intention parameters of the target vehicle after initialization are obtained. In the running stage of the probabilistic interactive prediction model, based on the preset parameter and intention update algorithm and the preset update trigger mechanism, the real-time perception data is used to update the probabilistic interactive prediction model parameters, the intention of the target vehicle and the uncertainty. Thus, the problem of the update time and the update method of the long-time domain interactive prediction model is solved. Compared with the interactive scheme of single vehicle automatic driving, the method proposed in the present application can solve the problem of uncertainty of interactive behavior prediction in a longer time domain, and can give a clear interaction basis, and can be used as one of the basic functions of more cloud long-time domain predictive decision tasks.

[0159] Secondly, the information physical twin probabilistic interactive prediction model identification device proposed in the embodiment of the present application is described with reference to the accompanying drawings.

[0160] Figure 9 The block schematic diagram of the information physical twin probabilistic interactive prediction model identification device of the embodiment of the present application is shown in the figure.

[0161] As shown in the figure, Figure 9 The information physical twin probabilistic interactive prediction model identification device 10 includes an acquisition module 100, a fitting module 200 and an update module 300.

[0162] The acquisition module 100 is configured to acquire real-time perception data of a target vehicle. The fitting module 200 is configured to, in the establishment stage of the probabilistic interactive prediction model, use a preset data set to fit the probabilistic interactive prediction model parameters, and obtain the initialized probabilistic interactive prediction model parameters and the intention parameters of the target vehicle after initialization. The probabilistic interactive prediction model includes a driver model, a probabilistic lane changing model and an uncertainty accumulation mechanism. The probabilistic interactive prediction model parameters include the driver model parameters and the probabilistic lane changing model parameters. The update module 300 is configured to, in the running stage of the probabilistic interactive prediction model, based on the preset parameter and intention update algorithm and the preset update trigger mechanism, use the real-time perception data to update the probabilistic interactive prediction model parameters, the intention of the target vehicle and the uncertainty.

[0163] Further, in some embodiments, the fitting module 200 is configured to fit the driver model parameters based on a weighted two-norm loss function, and fit the probabilistic lane changing model parameters based on a cross-entropy loss function, to obtain the initialized probabilistic interactive prediction model parameters and the intention parameters of the target vehicle after initialization.

[0164] Further, in some embodiments, the updating module 300 is configured to: based on a preset posterior distribution calculation method, jointly sample the driver model parameters and the intention parameters, and calculate a first posterior probability according to a first sampling result; based on the preset posterior distribution calculation method, separately sample the parameterized lane-changing model parameters, and calculate a second posterior probability according to a second sampling result; and update the parameterized car-following interaction prediction model parameters, the intention of the target vehicle and the uncertainty according to the first posterior probability and / or the second posterior probability.

[0165] The calculation formula of the first posterior probability is:

[0166]

[0167] wherein, is the sampling time, is the state predicted based on the driver model parameters and the intention of the driver, is the actually observed state, is the prior variance of the predicted state and the observed state, is the parameter of the driver model, is the intention of the driver.

[0168] The calculation formula of the second posterior probability is:

[0169]

[0170] wherein, is the number of samples, is the probability of left lane-changing predicted by the model under the premise of the th sample, is the probability of left lane-changing calculated based on the parameter under the premise of the th sample, is the probability of no lane-changing predicted by the model under the premise of the th sample, is the probability of no lane-changing calculated based on the parameter under the premise of the th sample, is the probability of right lane-changing predicted by the model under the premise of the th sample, is the probability of right lane-changing calculated based on the parameter under the premise of the th sample, is the parameter of the parameterized car-following interaction prediction model used.

[0171] Further, in some embodiments, the updating module 300 is further configured to: if the joint sampling of the driver model parameters and the intention parameters meets a preset time interval condition, update the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty based on a time update trigger mechanism; and if the target vehicle meets a preset lane changing condition or a prediction result of the probabilistic surrounding vehicle interaction prediction model meets a preset result, update the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty based on an event update trigger mechanism.

[0172] Further, in some embodiments, before obtaining the real-time perception data of the target vehicle, the obtaining module 100 is further configured to: establish a representation of longitudinal motion, lateral motion and uncertainty accumulation for each vehicle to be subjected to interactive prediction.

[0173] It should be noted that the foregoing explanation of the information physical twin probabilistic surrounding vehicle interactive prediction model identification method embodiment is also applicable to the information physical twin probabilistic surrounding vehicle interactive prediction model identification device of this embodiment, which will not be described here.

[0174] According to the information physical twin probabilistic surrounding vehicle interactive prediction model identification device of the embodiments of the present application, in the establishment stage of the probabilistic surrounding vehicle interaction prediction model, the probabilistic surrounding vehicle interaction prediction model parameters are fitted using a preset data set to obtain initialized probabilistic surrounding vehicle interaction prediction model parameters and intention parameters of the target vehicle after initialization, and in the running stage of the probabilistic surrounding vehicle interaction prediction model, the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and the uncertainty are updated using real-time perception data based on a preset parameter and intention update algorithm and a preset update trigger mechanism. Thus, the problem that the prior art is difficult to effectively model and update the complex uncertainty in long-time domain interactive prediction is solved, the surrounding vehicle interaction behavior prediction can be performed in a longer time domain, and clear interaction basis is provided.

[0175] Figure 10 A structural schematic diagram of an electronic device provided by the embodiments of the present application is provided. The electronic device can include:

[0176] The memory 1001, the processor 1002 and the computer program stored in the memory 1001 and executable on the processor 1002.

[0177] The processor 1002 implements the information physical twin probabilistic surrounding vehicle interactive prediction model identification method provided in the above embodiments when executing the program.

[0178] Further, the electronic device further includes:

[0179] The communication interface 1003 is configured to communicate between the memory 1001 and the processor 1002.

[0180] The memory 1001 is used to store computer programs that can be run on the processor 1002 .

[0181] The memory 1001 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0182] If the memory 1001, processor 1002, and communication interface 1003 are implemented independently, the communication interface 1003, memory 1001, and processor 1002 can be connected to each other via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 10 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0183] Optionally, in a specific implementation, if the memory 1001, the processor 1002 and the communication interface 1003 are integrated on a chip, the memory 1001, the processor 1002 and the communication interface 1003 can communicate with each other through an internal interface.

[0184] The processor 1002 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0185] An embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, it implements the above-mentioned information-physical twin probabilistic weekly vehicle interactive prediction model identification method.

[0186] An embodiment of the present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned information-physical twin probabilistic weekly vehicle interactive prediction model identification method.

[0187] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the description of the specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or N embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples, without contradiction.

[0188] In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is at least two, for example, two, three, etc., unless otherwise specifically limited.

[0189] Although the embodiments of the present application have been shown and described above, it is understood that the above-described embodiments are exemplary and cannot be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above-described embodiments within the scope of the present application.

Claims

1. An information-physical twin probabilistic roundabout interactive prediction model identification method, characterized in that, The method comprises the following steps: obtaining real-time perception data of a target vehicle; in the establishment stage of the probabilistic surrounding vehicle interaction prediction model, fitting the probabilistic surrounding vehicle interaction prediction model parameters by using a preset data set to obtain initialized probabilistic surrounding vehicle interaction prediction model parameters and initialized intention parameters of the target vehicle, wherein the probabilistic surrounding vehicle interaction prediction model comprises a driver model, a probabilistic lane-changing model and an uncertainty accumulation mechanism, and the probabilistic surrounding vehicle interaction prediction model parameters comprise driver model parameters and probabilistic lane-changing model parameters; in the running stage of the probabilistic surrounding vehicle interaction prediction model, updating the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and uncertainty based on a preset parameter and intention updating algorithm and a preset updating trigger mechanism by using the real-time perception data; wherein the updating of the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and uncertainty based on the preset parameter and intention updating algorithm and the preset updating trigger mechanism by using the real-time perception data comprises: jointly sampling the driver model parameters and the intention parameters based on a preset posterior distribution calculation method, and calculating a first posterior probability according to a first sampling result; separately sampling the probabilistic lane-changing model parameters based on a preset posterior distribution calculation method, and calculating a second posterior probability according to a second sampling result; and updating the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and uncertainty according to the first posterior probability and / or the second posterior probability; wherein the calculation formula of the first posterior probability is: wherein, is the sampling time, is the predicted state based on the driver model parameters and intentions , is the actual observed state, is the prior variance of the predicted state and the observed state, is the parameter of the driver model, is the intention of the driver; the calculation formula of the second posterior probability is: in, is the number of samples, For the The probability of left lane change predicted by the model under the premise of samples, For the Based on the parameters of the sample The calculated probability of changing lanes left is, For the The probability of not changing lanes predicted by the model under the premise of samples is, For the Based on the parameters of the sample The calculated probability of not changing lanes is For the The probability of right lane change predicted by the model under the premise of samples, For the Based on the parameters of the sample The calculated probability of right lane change is, are the parameters of the probabilistic weekly vehicle interaction prediction model adopted; the updating of the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and uncertainty based on the preset parameter and intention updating algorithm and the preset updating trigger mechanism by using the real-time perception data further comprises: if the joint sampling of the driver model parameters and the intention parameters meets a preset time interval condition, updating the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and uncertainty based on a time updating trigger mechanism; and if the target vehicle meets a preset lane-changing condition or the prediction result of the probabilistic surrounding vehicle interaction prediction model meets a preset result, updating the probabilistic surrounding vehicle interaction prediction model parameters, the intention of the target vehicle and uncertainty based on an event updating trigger mechanism.

2. The method of claim 1, wherein, the fitting of the probabilistic surrounding vehicle interaction prediction model parameters by using the preset data set to obtain the initialized probabilistic surrounding vehicle interaction prediction model parameters and the initialized intention parameters of the target vehicle comprises: fitting the driver model parameters based on a weighted two-norm loss function, and fitting the probabilistic lane-changing model parameters based on a cross-entropy loss function to obtain the initialized probabilistic surrounding vehicle interaction prediction model parameters and the initialized intention parameters of the target vehicle.

3. The method of claim 1, wherein, Before obtaining the real-time perception data of the target vehicle, the method further comprises: establishing a representation of longitudinal motion, lateral motion and uncertainty accumulation for each vehicle to be subjected to interactive prediction.

4. An information-physical twin probabilistic roundabout interactive prediction model identification device, characterized in that, comprises: An acquisition module is configured to acquire real-time perception data of a target vehicle; A fitting module is configured to, in a building stage of a probabilistic surrounding vehicle interaction prediction model, fit a probabilistic surrounding vehicle interaction prediction model parameter by using a preset data set, to obtain an initialized probabilistic surrounding vehicle interaction prediction model parameter and an intention parameter of the target vehicle after initialization, wherein the probabilistic surrounding vehicle interaction prediction model comprises a driver model, a probabilistic lane-changing model, and an uncertainty accumulation mechanism, and the probabilistic surrounding vehicle interaction prediction model parameter comprises a driver model parameter and a probabilistic lane-changing model parameter; An updating module is configured to, in a running stage of the probabilistic surrounding vehicle interaction prediction model, update the probabilistic surrounding vehicle interaction prediction model parameter, the intention of the target vehicle, and uncertainty by using the real-time perception data based on a preset parameter and intention updating algorithm and a preset updating trigger mechanism. The updating module is configured to: based on a preset posterior distribution calculation method, jointly sample the driver model parameter and the intention parameter, and calculate a first posterior probability according to a first sampling result; based on the preset posterior distribution calculation method, separately sample the probabilistic lane-changing model parameter, and calculate a second posterior probability according to a second sampling result; and update the probabilistic surrounding vehicle interaction prediction model parameter, the intention of the target vehicle, and the uncertainty according to the first posterior probability and / or the second posterior probability. The calculation formula of the first posterior probability is as follows: wherein, is the sampling time, is the predicted state based on the driver model parameters and intentions , is the actual observed state, is the prior variance of the predicted state and the observed state, is the parameter of the driver model, is the intention of the driver; The calculation formula of the second posterior probability is as follows: in, is the number of samples, For the The probability of left lane change predicted by the model under the premise of samples, For the Based on the parameters of the sample The calculated probability of changing lanes left is, For the The probability of not changing lanes predicted by the model under the premise of samples is, For the Based on the parameters of the sample The calculated probability of not changing lanes is For the The probability of right lane change predicted by the model under the premise of samples, For the Based on the parameters of the sample The calculated probability of right lane change is, are the parameters of the probabilistic weekly vehicle interaction prediction model adopted; The updating of the probabilistic surrounding vehicle interaction prediction model parameter, the intention of the target vehicle, and the uncertainty by using the real-time perception data based on the preset parameter and intention updating algorithm and the preset updating trigger mechanism further comprises: if the joint sampling of the driver model parameter and the intention parameter meets a preset time interval condition, updating the probabilistic surrounding vehicle interaction prediction model parameter, the intention of the target vehicle, and the uncertainty based on a time updating trigger mechanism; and if the target vehicle meets a preset lane-changing condition or a prediction result of the probabilistic surrounding vehicle interaction prediction model meets a preset result, updating the probabilistic surrounding vehicle interaction prediction model parameter, the intention of the target vehicle, and the uncertainty based on an event updating trigger mechanism.

5. The apparatus of claim 4, wherein, The fitting module is configured to: fit the driver model parameter based on a weighted two-norm loss function, and fit the probabilistic lane-changing model parameter based on a cross-entropy loss function, to obtain the initialized probabilistic surrounding vehicle interaction prediction model parameter and the intention parameter of the target vehicle after initialization.

6. An electronic device, comprising: The computer program is executed by the processor to implement the information-physical twin probabilistic surrounding vehicle interaction prediction model identification method according to any one of claims 1-3. The program is executed by the processor to implement the information-physical twin probabilistic surrounding vehicle interaction prediction model identification method according to any one of claims 1-3.

7. A computer storage medium having a computer program stored thereon, characterized in that: ​ 8. A computer program product, characterised in that, The computer program comprises a computer program which, when executed by a processor, is used to implement the information physical twin probabilistic roundabout interactive prediction model identification method of any one of claims 1-3.

Images