Vehicle cloud cooperation based driving trajectory planning method and device

Abstract

The application relates to the technical field of vehicle auxiliary driving, in particular to a driving track planning method and device based on vehicle-cloud cooperation, wherein the method comprises the following steps: updating an initial surrounding vehicle probability model based on obtained target perception data of a self vehicle and surrounding vehicles, obtaining a surrounding vehicle probability model, obtaining driver model and vehicle lane-changing model parameters and parameter uncertainty, and obtaining the intention of the self vehicle and intention uncertainty; constructing a self vehicle closed-loop strategy tree based on the parameters and parameter uncertainty, the intention of the self vehicle and intention uncertainty, searching the self vehicle closed-loop strategy tree in combination with the self vehicle closed-loop strategy tree and a parallel computing mode, and generating a behavior sequence decision of the self vehicle; and generating a space-time track of the self vehicle based on the behavior sequence decision and a track optimization model. Thus, the problems that, in the related art, the single vehicle is limited by the field of view, the observable range is small, the single vehicle has limited computing power, and it is difficult to effectively solve the long-time-domain planning problem, resulting in low availability of the planning result and low prediction accuracy are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle assisted driving, and in particular relates to a driving trajectory planning method and device based on vehicle-cloud cooperation. BACKGROUND

[0002] Vehicle assisted driving technology is crucial for ensuring safe driving of vehicles and improving vehicle passing efficiency, especially in an era when the penetration rate of intelligent and networked vehicles is not high. As a kind of vehicle assisted driving technology, long-time domain can plan the driving lane and speed of the vehicle in advance through the front road and surrounding vehicle information, so that the vehicle can avoid possible large system emergence in advance, and ensure safe and efficient driving of the vehicle, but there are still some problems that can be solved by vehicle-cloud cooperation.

[0003] In related technologies, there are mainly two types: a predictive driving planning method based on single vehicle intelligence and a predictive driving planning method based on vehicle-cloud cooperation. Among them, the predictive driving planning method based on single vehicle intelligence obtains surrounding vehicle information and road information through sensors mounted on the ego vehicle, performs predictive planning and solving, and then performs further short-time domain trajectory planning or driving control; the predictive driving planning method based on vehicle-cloud cooperation models the surrounding vehicles through the perception information of the cloud, predicts the surrounding vehicles, and then predicts the behavior of the ego vehicle, and the surrounding vehicles are not affected by the ego vehicle during the planning process, and the planning time domain is generally less than 10s.

[0004] However, in related technologies, the predictive driving planning method based on single vehicle intelligence is limited by the single vehicle's field of view, and the observable range is generally small, and the single vehicle's computing power is limited, so it is generally difficult to effectively solve the long-time domain predictive planning problem, resulting in unreasonable problem design, low usability of planning results, low prediction accuracy, and other situations, and the predictive driving planning method based on vehicle-cloud cooperation does not consider the interaction between the ego vehicle and the surrounding vehicles, or implicitly considers such interaction, making the long-time domain planning unavailable. SUMMARY

[0005] The present application provides a driving trajectory planning method and device based on vehicle-cloud cooperation to solve the problems in related technologies, such as limited single vehicle's field of view, small observable range, limited single vehicle's computing power, difficulty in effectively solving long-time domain planning problems, and low usability of planning results and low prediction accuracy.

[0006] The first aspect embodiment of the application provides a driving trajectory planning method based on vehicle-cloud cooperation, the method comprising the following steps: obtaining target perception data of a host vehicle and surrounding vehicles, and updating an initial surrounding vehicle probability model based on the target perception data to obtain a surrounding vehicle probability model, so as to obtain parameters and parameter uncertainties in a driver model and a vehicle lane-changing model, an intention and intention uncertainty of the host vehicle based on the surrounding vehicle probability model; constructing a host vehicle closed-loop strategy tree based on the parameters and parameter uncertainties, the intention and intention uncertainty of the host vehicle, and combining the host vehicle closed-loop strategy tree and a parallel computing method to search the host vehicle closed-loop strategy tree to generate a behavior sequence decision of the host vehicle; and generating a space-time trajectory of the host vehicle based on the behavior sequence decision and a trajectory optimization model.

[0007] Optionally, in an embodiment of the application, the target perception data of the host vehicle and the surrounding vehicles is obtained, and the initial surrounding vehicle probability model is updated based on the target perception data to obtain a surrounding vehicle probability model, comprising: performing segmentation and extraction of road map information along a route according to a destination of the host vehicle and global path planning information, so as to generate discrete road graph information of the host vehicle according to different positions of the host vehicle, for establishing the initial surrounding vehicle probability model; obtaining the target perception data of the host vehicle and the surrounding vehicles according to real-time traffic twins at the beginning of each planning period; in each planning period, the discrete road graph information is called according to the position of the host vehicle, and the initial surrounding vehicle probability model is updated based on the discrete road graph information and a parameter and intention recognition algorithm to obtain a surrounding vehicle probability model, wherein the surrounding vehicle probability model comprises at least one of surrounding vehicle intention and probability, a driver model, and a vehicle lane-changing model.

[0008] Optionally, in an embodiment of the application, the parameter and intention recognition algorithm is based on Bayesian inference, wherein the Bayesian inference comprises: selecting a prior distribution of a parameter in the driver model based on an initial hypothesis; constructing a posterior distribution of the parameter in the intention, the driver model, and the surrounding vehicle lane-changing model based on the prior distribution; when the posterior distribution cannot be analyzed, using a MCMC (Markov Chain Monte Carlo) method to sample a to-be-recognized parameter to obtain sample data of the to-be-recognized parameter; and obtaining a to-be-recognized intention probability, a parameter, and an uncertainty thereof in the to-be-recognized parameter based on the sample data.

[0009] Optionally, in an embodiment of the present application, the self-vehicle closed-loop strategy tree is constructed, and the search of the self-vehicle closed-loop strategy tree is performed in combination with the self-vehicle closed-loop strategy tree and a parallel computing mode to generate the behavior sequence decision of the self-vehicle, including: constructing a self-vehicle decision sequence containing multiple lane-changing behaviors in a long time domain; screening a surrounding vehicle interacting with the self-vehicle in combination with the self-vehicle decision sequence and a confidence reachable set; obtaining interaction information of the self-vehicle and the surrounding vehicle by using a forward deduction simulation mechanism; performing task allocation of the parallel computing mode based on the interaction information and the self-vehicle closed-loop strategy tree, and dynamically decomposing the task according to the number of available cores to obtain the behavior sequence decision.

[0010] Optionally, in an embodiment of the present application, the parameters and parameter uncertainties in the driver model and the vehicle lane-changing model, and the intention and intention uncertainty of the self-vehicle include: first-order uncertainty of the parameters in the driver model; first-order uncertainty of one lane-changing of the preceding vehicle; uncertainty of the intention entropy of the self-vehicle; cumulative effect of uncertainty over time.

[0011] Optionally, in an embodiment of the present application, the generation of the space-time trajectory of the self-vehicle based on the behavior sequence decision and a trajectory optimization model includes: selecting a high-order polynomial based on the number of start point and end point constraints of the self-vehicle, wherein the high-order polynomial is greater than 5th order; generating boundary constraint conditions of the self-vehicle according to the behavior sequence decision; optimizing the remaining degree of freedom coefficients of the boundary constraint conditions by using the trajectory optimization model to construct a trajectory optimization model for a rollover state index; solving the trajectory optimization model to obtain an optimal trajectory of the self-vehicle, and obtaining the space-time trajectory based on the optimal trajectory.

[0012] The second aspect embodiment of the present application provides a driving trajectory planning device based on vehicle-cloud cooperation, the device includes: a generation module configured to obtain target perception data of a self-vehicle and surrounding vehicles, and update an initial surrounding vehicle probability model based on the target perception data to obtain a surrounding vehicle probability model, so as to obtain parameters and parameter uncertainties in a driver model and a vehicle lane-changing model, and intention and intention uncertainty of the self-vehicle based on the surrounding vehicle probability model; a construction module configured to construct a self-vehicle closed-loop strategy tree based on the parameters and parameter uncertainties, and the intention and intention uncertainty of the self-vehicle, and perform a search of the self-vehicle closed-loop strategy tree in combination with the self-vehicle closed-loop strategy tree and a parallel computing mode to generate a behavior sequence decision of the self-vehicle; and a planning module configured to generate a space-time trajectory of the self-vehicle based on the behavior sequence decision and a trajectory optimization model.

[0013] Optionally, in an embodiment of the present application, the generating module comprises: a first generating unit configured to perform segmentation and extraction of road map information along a route of the ego vehicle according to a destination of the ego vehicle and global path planning information, to generate discrete road map information of the ego vehicle according to different positions of the ego vehicle, for establishing the initial surrounding vehicle probability model; an obtaining unit configured to obtain target perception data of the ego vehicle and the surrounding vehicle according to real-time traffic twins at the beginning of each planning period; and an updating unit configured to, in each planning period, retrieve the discrete road map information according to a position of the ego vehicle, and update the initial surrounding vehicle probability model in combination with the discrete road map information and a parameter and intention recognition algorithm, to obtain a surrounding vehicle probability model, wherein the surrounding vehicle probability model comprises at least one of a surrounding vehicle intention and a probability thereof, a driver model, and a vehicle lane-changing model.

[0014] Optionally, in an embodiment of the present application, the parameter and intention recognition algorithm is based on Bayesian inference, wherein the Bayesian inference comprises: selecting a prior distribution of a parameter in the driver model based on an initial hypothesis; constructing a posterior distribution of parameters in the intention, the driver model, and the surrounding vehicle lane-changing model based on the prior distribution; when the posterior distribution cannot be solved, sampling a to-be-recognized parameter by using a Markov Chain Monte Carlo (MCMC) method to obtain sample data of the to-be-recognized parameter; and obtaining a to-be-recognized intention probability, a parameter, and uncertainty thereof in the to-be-recognized parameter based on the sample data.

[0015] Optionally, in an embodiment of the present application, the constructing module comprises: a first constructing unit configured to construct a decision sequence of the ego vehicle containing multiple lane-changing behaviors in a long time domain; a first screening unit configured to screen a surrounding vehicle interacting with the ego vehicle in combination with the decision sequence of the ego vehicle and a confidence reachable set; a second generating unit configured to obtain interaction information of the ego vehicle and the surrounding vehicle by using a forward reasoning simulation mechanism; and a third generating unit configured to perform task allocation of the parallel computing mode based on the interaction information and a closed-loop strategy tree of the ego vehicle, and dynamically decompose the task according to a number of available cores, to obtain the behavior sequence decision.

[0016] Optionally, in an embodiment of the present application, the parameters and parameter uncertainties in the driver model and the vehicle lane-changing model, and the intention and intention uncertainty of the ego vehicle comprise: a first-order uncertainty of a parameter in the driver model; a first-order uncertainty of a lane-changing behavior of a preceding vehicle; an uncertainty of an intention entropy of the ego vehicle; and a cumulative effect of uncertainty over time.

[0017] Optionally, in one embodiment of the present application, the planning module includes: a second screening unit, used to select a high-order polynomial based on the number of starting point and end point constraints of the ego vehicle, wherein the high-order polynomial is greater than 5th order; a fourth generation unit, used to generate the boundary constraints of the ego vehicle according to the behavior sequence decision; a second construction unit, used to optimize the residual degree of freedom coefficient of the boundary constraints using the trajectory optimization model to construct a trajectory optimization model for the rollover state index; a solving unit, used to solve the trajectory optimization model to obtain the optimal trajectory of the ego vehicle, and obtain the spatiotemporal trajectory based on the optimal trajectory.

[0018] The third aspect of the present application provides a server, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the driving trajectory planning method based on vehicle-cloud collaboration as described in the above embodiment.

[0019] The fourth aspect of the present application provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, it implements the above-mentioned driving trajectory planning method based on vehicle-cloud collaboration.

[0020] The fifth aspect of the present application provides a computer program product, including a computer program, which, when executed, implements the above-mentioned driving trajectory planning method based on vehicle-cloud collaboration.

[0021] The embodiment of the present application can obtain a surrounding vehicle probability model based on the acquired target perception data of the ego vehicle and surrounding vehicles, and then obtain the parameters and parameter uncertainties in the driver model and the vehicle lane change model, the ego vehicle's intention and intention uncertainty, and construct an ego vehicle closed-loop policy tree based on the parameters and parameter uncertainties, the ego vehicle's intention and intention uncertainty. The ego vehicle closed-loop policy tree is searched using a parallel computing method to obtain behavior sequence decisions, generate a spatiotemporal trajectory, and plan the ego vehicle's spatiotemporal trajectory. The uncertainty of the driver's parameters and intentions is clearly reflected through the surrounding vehicle probability model. The uncertainty propagation mechanism avoids the logical error of "prediction first, decision planning later" while considering uncertainty information, thereby improving the credibility and feasibility of the planning results. The ego vehicle closed-loop policy tree achieves the highest solution efficiency and ensures real-time performance. Good performance can be obtained under long-term predictive decision-making tasks. The trajectory optimization model is used to ensure the rollover safety of the ego vehicle, solve the lateral manipulation safety problem of the ego vehicle from the planning layer, and output a feasible, safe and stable trajectory. This solves the problems in related technologies, such as being restricted by the field of view of a single vehicle, having a small observable range, and limited computing power of a single vehicle, making it difficult to effectively solve long-term planning problems, resulting in low availability of planning results and low prediction accuracy.

[0022] Additional aspects and advantages of the present application will be made apparent by the following description and the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0023] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, including the accompanying drawings.

[0024] Figure 1 A flow chart of a driving trajectory planning method based on vehicle-cloud cooperation according to an embodiment of the present application;

[0025] Figure 2 A schematic diagram of an example of discretized road graph information according to an embodiment of the present application;

[0026] Figure 3 A flow chart of constructing a round vehicle probability model according to an embodiment of the present application;

[0027] Figure 4 A flow chart of acquiring a lane change probability using a MLE (Maximum Likelihood Estimation) and MOBIL (minimize overall braking induced by lane change) model according to an embodiment of the present application;

[0028] Figure 5 A flow chart of a predictive driving planning algorithm according to an embodiment of the present application;

[0029] Figure 6 A flow chart of constructing a variable time step vehicle closed-loop strategy tree according to an embodiment of the present application;

[0030] Figure 7 A schematic diagram of an example of a running scenario according to an embodiment of the present application;

[0031] Figure 8 A block schematic diagram of interactive round vehicle extraction results and real-time round vehicle deduction according to an embodiment of the present application;

[0032] Figure 9 A block schematic diagram of scenario division in a trajectory configuration according to an embodiment of the present application;

[0033] Figure 10 A flow chart of assigning a trajectory configuration according to an embodiment of the present application;

[0034] Figure 11A flow chart for generating a spatiotemporal trajectory of a vehicle according to an embodiment of the present application;

[0035] Figure 12 A block diagram of a rollover-preventing lane-changing trajectory planning result according to an embodiment of the present application;

[0036] Figure 13 A block diagram of a predictive safe and efficient driving trajectory planning result according to an embodiment of the present application;

[0037] Figure 14 A block diagram of a driving trajectory planning system according to an embodiment of the present application;

[0038] Figure 15 A block diagram of a driving trajectory planning device based on vehicle-cloud cooperation according to an embodiment of the present application;

[0039] Figure 16 A structural diagram of a server according to an embodiment of the present application. DETAILED DESCRIPTION

[0040] Embodiments of the present application are described in detail below with reference to the attached drawings, which are meant to be exemplary and are not intended to limit the present application.

[0041] A method and device for planning a driving trajectory based on vehicle-cloud cooperation according to an embodiment of the present application are described below with reference to the accompanying drawings. In view of the small observable range and limited computing power of a single vehicle, it is difficult to effectively solve a long-time-domain planning problem, resulting in low availability and low prediction accuracy of the planning result. To address this issue, the present application provides a method for planning a driving trajectory based on vehicle-cloud cooperation. In this method, a vehicle probability model is obtained based on target perception data of a host vehicle and surrounding vehicles, and then parameters and parameter uncertainty in a driver model and a vehicle lane-changing model, and the intention and intention uncertainty of the host vehicle are obtained. A host vehicle closed-loop strategy tree is constructed based on the parameters and parameter uncertainty, and the intention and intention uncertainty of the host vehicle. The host vehicle closed-loop strategy tree is searched using a parallel computing method, and then a behavior sequence decision is obtained to generate a space-time trajectory and plan a space-time trajectory for the host vehicle. The uncertainty of the driver's parameters and intentions is explicitly reflected by the vehicle probability model. The uncertainty propagation mechanism avoids the logical error of "predicting first and then planning" while considering uncertainty information, thereby improving the credibility and feasibility of the planning result. The host vehicle closed-loop strategy tree achieves the highest solution efficiency and ensures real-time performance, and can achieve good performance in long-time-domain predictive decision tasks. The trajectory optimization model ensures the safety of the host vehicle from rolling over, and solves the lateral control safety problem of the host vehicle at the planning level, and outputs a feasible and safe trajectory. Thus, the problems of low availability and low prediction accuracy of the planning result are solved.

[0042] Specifically, Figure 1 A flowchart of a method for planning a driving trajectory based on vehicle-cloud cooperation according to an embodiment of the present application is provided.

[0043] As Figure 1 shown, the method for planning a driving trajectory based on vehicle-cloud cooperation includes the following steps:

[0044] In step S101, target perception data of a host vehicle and surrounding vehicles is obtained, and an initial surrounding vehicle probability model is updated based on the target perception data to obtain a surrounding vehicle probability model. Parameters and parameter uncertainty in a driver model and a vehicle lane-changing model, and the intention and intention uncertainty of the host vehicle are obtained based on the surrounding vehicle probability model.

[0045] It can be understood that the target perception data according to the present application can include, but is not limited to, perception data of a roadside perception unit, perception data of an intelligent connected vehicle, etc. The specific settings can be made by those skilled in the art according to actual conditions, and the present application does not make specific limitations.

[0046] In the embodiments of the present application, the target perception data of the ego vehicle and surrounding vehicles can be obtained through roadside perception units, intelligent networked vehicles, and information sharing of traffic control platforms, and the overall traffic twin in each small area can be established in parallel after the map is partitioned, and then the global number of the ego vehicle and surrounding vehicles can be obtained by means of vehicle ReID (Person Re-identification, pedestrian re-identification) technology. The global number is unique, and the driving data in the past short period of time can be retrieved from the database without infringing the personal privacy of road users to update the parameters and intentions for dynamic deduction.

[0047] In addition, it should be noted that the surrounding vehicle probability model in the embodiments of the present application can include, but is not limited to, a driver model and a vehicle lane changing model. The specific settings can be made by those skilled in the art according to actual conditions, and the present application does not make specific limitations.

[0048] As a possible implementation, after obtaining the target perception data of the ego vehicle and surrounding vehicles, the initial surrounding vehicle probability model can be updated based on the target perception data to obtain the surrounding vehicle probability model, and then the parameters and parameter uncertainties in the driver model and the vehicle lane changing model, the intention and intention uncertainty of the ego vehicle can be obtained.

[0049] Optionally, in an embodiment of the present application, the target perception data of the ego vehicle and surrounding vehicles is obtained, and the initial surrounding vehicle probability model is updated based on the target perception data to obtain the surrounding vehicle probability model, which includes: dividing and extracting the road map information along the route according to the destination of the ego vehicle and the global path planning information, to generate discrete road graph information of the ego vehicle according to different positions of the ego vehicle, for establishing the initial surrounding vehicle probability model; at the beginning of each planning period, obtaining the target perception data of the ego vehicle and surrounding vehicles according to the real-time traffic twin; in each planning period, the discrete road graph information is retrieved according to the position of the ego vehicle, and the initial surrounding vehicle probability model is updated by combining the discrete road graph information and the parameter and intention recognition algorithm to obtain the surrounding vehicle probability model, wherein the surrounding vehicle probability model includes at least one of the surrounding vehicle intention and its probability, the driver model, and the vehicle lane changing model.

[0050] In the embodiments of the present application, the discrete road graph information can be understood as lane-level structured map information, which can include, but is not limited to, the number of drivable lanes, the driving restrictions of lanes for different types of vehicles, the lane width, the scatter point position information of the lane center line, the road level, the road speed limit or the lane-level speed limit, the road connection relationship and traffic rule information, and the like. The specific settings can be made by those skilled in the art according to actual conditions. For example, the discrete road graph information in the embodiments of the present application can be expressed in the form of a data dictionary, as shown in Figure 2

[0051] ​As a possible implementation manner, the embodiment of the application constructs a vehicle surrounding probability model process as shown in the following table Figure 3 , including the following steps:

[0052] Step S301: according to the destination of the ego vehicle and the global path planning information, the road map information along the route is segmented and extracted, so as to generate the discretized road map information of the ego vehicle according to different positions of the ego vehicle, and to establish an initial vehicle surrounding probability model.

[0053] Step S302: at the beginning of each planning period, the target perception data of the ego vehicle and the surrounding vehicles are obtained according to real-time traffic twins.

[0054] Step S303: in each planning period, the discretized road map information is called according to the position of the ego vehicle, and the initial vehicle surrounding probability model is updated by combining the discretized road map information and the parameter and intention recognition algorithm, so as to obtain the vehicle surrounding probability model.

[0055] It can be understood that the embodiment of the application updates the initial vehicle surrounding probability model, such as updating the driver parameter information of the driver model (such as the parameters and parameter variances of the driver model), the lane changing parameter information of the vehicle lane changing model (such as the parameters and parameter variances of the vehicle lane changing model) and the ego vehicle intention information of the ego vehicle (the intention and intention variance of the ego vehicle), so as to obtain the vehicle surrounding probability model, wherein the vehicle surrounding probability model can include but is not limited to at least one of the vehicle surrounding intention and its probability, the driver model and the vehicle lane changing model.

[0056] Among them, the expression of the driver model in the embodiment of the application can be but not limited to:

[0057]

[0058] Among them, T hw represents the expected headway, a max represents the maximum acceleration, b comf represents the comfortable deceleration, s min represents the minimum distance, v des,ego represents the expected speed.

[0059] In addition, it should be noted that the use of softplus (*) instead of ReLU (*) in the embodiment of the application can make the driver model continuous and derivable everywhere, so that the analytical gradient of each parameter can be obtained, wherein the parameters θ to be estimated are a vector composed of T hw , a max , b comf , s min , v des,ego , and the expression of the softplus (*) function can be but not limited to:

[0060] softplus(x) = log(1 + e x )

[0061] , (3)

[0062] Optionally, in an embodiment of the present application, the parameters and the intention recognition algorithm are based on Bayesian inference, wherein the Bayesian inference comprises: selecting a prior distribution of the parameters in the driver model based on an initial hypothesis; constructing a posterior distribution of the parameters in the intention, the driver model and the car-following lane-changing model based on the prior distribution; when the posterior distribution cannot be solved, sampling the to-be-identified parameters by using a MCMC method to obtain sample data of the to-be-identified parameters; and obtaining a to-be-identified intention probability, the parameters and their uncertainty in the to-be-identified parameters based on the sample data.

[0063] In some embodiments, the parameters and the intention recognition algorithm of the driver model in the embodiments of the present application can be but are not limited to Bayesian inference, wherein in the embodiments of the present application, the steps of the Bayesian inference comprise: first, selecting a prior distribution of the parameters in the driver model based on an initial hypothesis; second, constructing a likelihood function reflecting the observation data under the given parameters; third, obtaining a posterior distribution of the parameters by combining the prior distribution and the likelihood function through the Bayesian theorem; fourth, when the posterior distribution cannot be solved, sampling the to-be-identified parameters by using a MCMC method to obtain sample data of the to-be-identified parameters; and finally, calculating the variance of the model parameters to quantify the uncertainty and further determine the confidence interval to represent the potential variation range of the parameters through the sample data.

[0064] In the embodiments of the present application, the expression of the prior distribution p(θ) can be but is not limited to:

[0065]

[0066] wherein μ θ is the mean of the prior distribution of the parameters, is the prior variance of the parameters.

[0067] The expression of the constructed likelihood function p(data|θ) can be but is not limited to:

[0068]

[0069] wherein x i is the i th observation value, σ 2 is the variance of the observation noise, and N is the number of observation data.

[0070] Further, according to the Bayesian theorem, the posterior distribution p(θ|data) is obtained from the product of the prior distribution and the likelihood function, and the expression thereof can be but is not limited to:

[0071]

[0072] Among them, p(data) is a normalization constant, which is ignored in actual processing and does not depend on the specific model parameters θ.

[0073] In addition, when the posterior distribution cannot be analyzed, the embodiment of the present application uses the MCMC method to sample it, wherein the MCMC method of the embodiment of the present application constructs a Markov chain and samples a series of θ from the posterior distribution to construct the probability distribution of the parameters. The θ1, θ2, ..., θ obtained by sampling M , and then estimate the distribution of the parameters. For example, if the embodiment of the present application samples T hw Sample value of The sample mean can be used as the estimated value of the parameter, the sample variance can be used as the uncertainty of the parameter, and the sample covariance matrix can be used to represent the linear dependence between the parameters. That is, the embodiment of the present application calculates the covariance between each pair of samples sampled for each parameter, and then obtains the sample covariance matrix. Among them, the element C of the covariance matrix is ij The calculation formula can be but not limited to:

[0074]

[0075] Where M is the number of samples in Markov Chain Monte Carlo, is the i-th parameter value of the m-th sample, is the sample mean of the i-th parameter.

[0076] In addition, it should be noted that the diagonal elements of the covariance matrix in the embodiment of the present application are the variances of the parameters, and the diagonal elements represent the correlation between different parameters.

[0077] In some embodiments, the parameter and intention recognition algorithm for the driver model of the embodiments of the present application can be, but is not limited to, a heuristic algorithm. The variance of the parameters is determined by simulating the driver model after the parameters have been identified. The preceding vehicle's speed and relative distance information are input at a fixed time, initialized with real historical speed data, and the simulated vehicle's speed is obtained. The squared difference between the simulated speed data and the real speed data is then calculated frame by frame. The average of the squared differences is calculated and multiplied by a defined conversion factor to determine the overall parameter variance. The fixed time can be set by those skilled in the art based on actual circumstances and is not specifically limited in this application.

[0078] In some embodiments, the parameters of the driver model and the intention recognition algorithm of the embodiments of the present application can be, but are not limited to, a combination of rule-based intention likelihood recognition and Bayesian inference, determining the intention bifurcation point of the ego vehicle within a set distance (which can be set by a person skilled in the art according to actual conditions, and the present application does not make specific limitations) through map information and traffic rules, and setting different intentions (each intention has its own prior information, which can be inferred through road structure and ego vehicle position) based on the recognized driver model. The past window of a set time (which can be set by a person skilled in the art according to actual conditions, and the present application does not make specific limitations) is used as input to simulate the acceleration output of the ego vehicle under different intention assumptions. Bayesian inference is still used for parameter updating, and the likelihood function of each different intention under the given recognized driver model parameters and observation data is calculated. After inputting the evaluation function, the probability increment of different intentions is obtained, which is added to the intention probability of the previous control period to update the new intention information.

[0079] Further, the vehicle lane changing model of the embodiments of the present application can use the MLE method or Bayesian inference for probability calculation and parameter recognition updating.

[0080] For example, the embodiments of the present application are based on the MLE method and the MOBIL model, and the steps of obtaining the left lane changing L, lane keeping K and right lane changing R probabilities of the ego vehicle are as shown in Figure 4 The main steps include:

[0081] Step S401: Obtain the lane changing acceleration.

[0082] It can be understood that the embodiments of the present application can use the MOBIL model to obtain the lane changing acceleration vector a, which can include, but is not limited to, Δa L,ego , a K,ego = 0, Δa R,ego , Δa L,f , Δa K,f = 0, Δa R,f , the expected acceleration benefits of the ego vehicle and the vehicle behind after lane changing corresponding to the LKR action of the ego vehicle (calculated by the recognized driver model of the ego vehicle and the surrounding vehicle), which is the difference between the acceleration after lane changing and the acceleration before lane changing for the ego vehicle; for the vehicle behind, it is the sum of the loss of the new vehicle behind after lane changing due to the lane changing of the ego vehicle and the benefit of the original vehicle behind before lane changing due to the lane changing of the ego vehicle.

[0083] Step S402: Design the evaluation function.

[0084] In the embodiments of the present application, the evaluation function f(a, p) can reflect the preferences of the ego vehicle and the surrounding vehicle for different lane changing options. Generally, the greater the acceleration, the more the ego vehicle and the surrounding vehicle tend to choose this behavior, which can be represented as:

[0085]

[0086] wherein, α L ,β L ,α R ,β R , p are five parameters to be identified, and the common parameter p is a courtesy coefficient corresponding to the MOBIL model, reflecting the importance of the vehicle to the ego vehicle affecting other vehicles.

[0087] Step S403: Obtain the lane changing probability.

[0088] Further, the embodiments of the present application can normalize the output of the evaluation function by using a softmax(*) function, that is, the probabilities of left lane changing L, lane keeping K and right lane changing R at this moment can be obtained, which can be but not limited to expressed as:

[0089]

[0090] For example, the embodiments of the present application update the parameters by using MLE, and the main content is: the embodiments of the present application can construct a data set by using a historical data set, which can include but is not limited to wherein, d i ∈{L,K,R} represents the actual lane changing decision of the i th sample, and the main target is to find parameters α L ,β L , a R ,β R , p so that the probability predicted by the model is most matched with the actually observed decision.

[0091] It can be understood that in the embodiments of the present application, the target of MLE is to maximize the product of the decision probability predicted by the model under all data samples, and the log-likelihood function can be but not limited to expressed as:

[0092]

[0093] wherein, the embodiments of the present application can use a first-order gradient descent method or a second-order quasi-Newton method to solve and update the parameters.

[0094] Further, in the embodiments of the present application, the parameters of the vehicle lane-changing model and the intention recognition algorithm can be, but are not limited to, a combination of rule-based intention possibility recognition and Bayesian inference, the intention bifurcation point of the vehicle in a given distance (not specifically limited in the present application) in the future is determined through map information and traffic rules (M possible intentions are recognized, for example, the vehicle is currently driving on a highway, there is a ramp at 500 m ahead, and the possible intentions include straight driving and ramp off; for example, the vehicle is currently at a traffic light intersection, and the possible intentions include straight driving, left turning, right turning, and U-turn), and based on the recognized 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 and vehicle position, for example, at a ramp, P(straight driving) = 0.7 and P(ramp off) = 0.3 can be given according to the current traffic flow + the lane position of the vehicle + the acceleration trend of the vehicle), the past vehicle history data in a given time window (not specifically limited in the present application) is used as input for simulation to obtain the acceleration output of the vehicle under different intention assumptions, the Markov chain Monte Carlo under the Bayesian inference given above is still used for parameter updating, the likelihood function of each different intention under the given recognized driver model parameters and observation data is calculated, and the probability increment of different intentions is obtained after inputting the evaluation function, and the intention probability of the previous control period is added to update the new intention information.

[0095] The embodiments of the present application rely on real-time traffic twins, establish a vehicle probability model, fully consider the uncertainty of driver parameters and intentions, accurately simulate the driving environment, and provide more real and accurate environmental information for the ego vehicle, thereby further improving the safety and accuracy of the ego vehicle decision.

[0096] Optionally, in an embodiment of the present application, the parameters and parameter uncertainty in the driver model and the vehicle lane-changing model, and the intention and intention uncertainty of the ego vehicle, include: first-order uncertainty of parameters in the driver model; first-order uncertainty of one lane-changing of the preceding vehicle; uncertainty of intention entropy of the ego vehicle; cumulative effect of uncertainty over time.

[0097] Those skilled in the art can understand that the parameters and parameter uncertainty in the driver model and the vehicle lane-changing model, and the intention and intention uncertainty of the ego vehicle in the embodiments of the present application can be, but are not limited to, first-order uncertainty of parameters in the driver model, first-order uncertainty of one lane-changing of the preceding vehicle, uncertainty of intention entropy of the ego vehicle, and cumulative effect of uncertainty over time, and the specific settings can be made by those skilled in the art according to actual conditions, which are not specifically limited in the present application.

[0098] In step S102, a self-vehicle closed-loop strategy tree is constructed based on the parameters and parameter uncertainties, the intentions of the self-vehicle and intention uncertainties, and a search is performed on the self-vehicle closed-loop strategy tree in combination with the self-vehicle closed-loop strategy tree and a parallel computing mode to generate a behavior sequence decision of the self-vehicle.

[0099] In some embodiments, the embodiments of the present application consider the interaction between the self-vehicle and the surrounding vehicles by using a forward deduction method, and use the uncertainty of the surrounding vehicles to assist in trajectory decision-making by using an uncertainty accumulation mechanism. While considering the uncertainty information, the logical error of "predicting first and then planning" is avoided, the planning result credibility and feasibility are improved, and the long-time domain reliable self-vehicle decision-making can be performed while considering the potential danger caused by the uncertainty of the surrounding vehicles.

[0100] It can be understood that the embodiments of the present application provide an uncertainty propagation mechanism and a self-vehicle and surrounding vehicle interaction method based on the uncertainty propagation mechanism. In the embodiments of the present application, the interaction method can be divided into two types. The first type is a multi-scenario Monte Carlo simulation method based on sampling, which is referred to as Monte Carlo simulation. The second type is a single-scenario simulation method considering uncertainty propagation, which is referred to as single-scenario simulation.

[0101] In addition, it should be noted that, based on the interaction between the self-vehicle and the surrounding vehicles, the simulation of all scenarios is performed based on the selected action sequence of the self-vehicle.

[0102] In some embodiments, the Monte Carlo simulation does not consider the accumulation of uncertainty, and uses a random sampling method to sample the parameter values of different surrounding vehicles and the scenarios of different specific intentions according to the estimation of the mean and variance of the model parameters of the surrounding vehicles. The parameter values with higher probabilities and intentions are more likely to be sampled. Each scenario contains a group of surrounding vehicles with specific parameter values and intentions. Through forward simulation deduction of each surrounding vehicle, each specific parameterized scenario instance with the same probability obtains a deterministic surrounding vehicle prediction result. After weighting the surrounding vehicle position prediction results of all scenarios according to the same probability, the probability position distribution of the surrounding vehicles under the selected self-vehicle behavior sequence is obtained.

[0103] In some embodiments, the single-scenario simulation simulates the highest possible parameters and intentions identified, but the accumulation and propagation of uncertainty are calculated in each step of the forward simulation process, and the deterministic events (such as static obstacles on the road, with a probability of 100% existing here) make the uncertainty gradually converge.

[0104] In some embodiments, the uncertainty propagation mechanism is a linearized vehicle motion uncertainty propagation mechanism, which considers the propagation of uncertainty between vehicles by Taylor expanding the influence of the driver model and lane changing model parameters on the model output results. The specific implementation is as follows:

[0105] wherein, the driver model of embodiments of the present application can but is not limited to be expressed as:

[0106] a pred = f(v p , v ego , d intvl , θ)

[0107] , (11)

[0108] When there is no leading vehicle, it can be simplified as free flow driving, which can but is not limited to be expressed as:

[0109] a pred = f(θ)

[0110] , (12)

[0111] The parameter identification of the driver model will have variance Through Taylor expansion, the influence of the leading vehicle state uncertainty and the uncertainty of the driver parameter θ of the following vehicle itself on the speed can be approximated as:

[0112]

[0113] Wherein, the uncertainty of the distance variance is the sum of the position state uncertainty of the two vehicles, which can but is not limited to be expressed as:

[0114]

[0115] Because the leading vehicle can change lanes to leave (1-P(K)) or stay in the lane (P(K)), the calculation formula of the longitudinal uncertainty containing the lane change uncertainty can but is not limited to be expressed as:

[0116]

[0117] Wherein, the subscript pp represents the leading vehicle of the leading vehicle.

[0118] When considering the intention uncertainty, because it can be assumed that the intentions of different vehicles are independent of each other for sufficient reasons, the calculation of the uncertainty adds the product of the intention entropy. Let:

[0119]

[0120] Then the following formula is established:

[0121]

[0122]

[0123] The intention entropy H of a certain vehicle is calculated according to formula (18), and the intention can include but is not limited to N possible intentions I = {I1, I2, …, In} of the vehicle. N}。

[0124]

[0125] Further, the embodiment of the present application can assume that the reason why the intentions of different vehicles are independent of each other is that the macro intention depends on the path planning of the vehicle to the destination, and basically no mutual influence between vehicles is generated.

[0126] In addition, the propagation value of the uncertainty of the embodiment of the present application in the case of no state update depends on the prediction step, which is realized by the following steps: in each time step Δt, the state of the vehicle is predicted by the acceleration model, the state vector x includes the position and the speed, and the state transition can be but is not limited to represented as:

[0127]

[0128] That is

[0129] x = Ax + Ba pred

[0130] , (20)

[0131] ZOH discretization is performed to obtain a discrete state equation:

[0132] x k+1 = Gx k + Ha pred,k

[0133] , (21)

[0134] At each time step, the uncertainty of the state is propagated, and this uncertainty affects the uncertainty of the position and the speed through the state transition equation, therefore, the uncertainty propagation of the preceding vehicle can be but is not limited to represented as:

[0135] P k+1 = GP k G + Q

[0136] , (22)

[0137] Wherein, Q is a process noise covariance matrix, which can be but is not limited to represented as:

[0138]

[0139] Wherein, Indicates the influence of acceleration on the variance of position, because the position is the integral of acceleration twice, and the variance is proportional to the fourth power of the sampling time; represents 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; Δt 2 represents the dependence of velocity on the square of acceleration.

[0140] The embodiment of the present application comprehensively considers the interaction mode of the ego vehicle and the surrounding vehicles by a forward deduction mode, introduces an uncertainty accumulation mechanism, fully utilizes the uncertainty information of the surrounding vehicles, and then assists in trajectory decision-making, avoids the logical error of the traditional "first prediction, then decision-making and planning", ensures the correctness of the planning result, and improves the credibility of the decision-making and planning.

[0141] Optionally, in an embodiment of the present application, an ego vehicle closed-loop strategy tree is constructed, and a search of the ego vehicle closed-loop strategy tree is performed in combination with the ego vehicle closed-loop strategy tree and a parallel computing mode to generate a behavior sequence decision of the ego vehicle, including: constructing an ego vehicle decision sequence containing multiple lane-changing behaviors in a long time domain; screening surrounding vehicles interacting with the ego vehicle in combination with the ego vehicle decision sequence and a confidence reachable set; obtaining interaction information of the ego vehicle and the surrounding vehicles by using a forward deduction simulation mechanism; performing task allocation in a parallel computing mode based on the interaction information and the ego vehicle closed-loop strategy tree, and dynamically decomposing tasks according to the number of available cores to obtain the behavior sequence decision.

[0142] In some embodiments, the embodiment of the present application constructs an ego vehicle closed-loop strategy tree based on parameters and parameter uncertainty, ego vehicle intention and intention uncertainty, considers an ego vehicle decision sequence containing multiple lane-changing behaviors in a long time domain, screens interacting surrounding vehicles by using a confidence reachable set, designs a forward deduction simulation mechanism to focus on the interaction of the ego vehicle and the related surrounding vehicles, and designs a load adaptive ego vehicle behavior tree search algorithm based on a parallel computing mode, dynamically decomposes tasks according to the number of available cores to realize the highest solving efficiency and ensure real-time performance, and then improves the solving efficiency under the premise of ensuring safe and efficient driving by using a heuristic pruning method and the like, and can achieve good performance under the long time domain predictive decision-making task.

[0143] For example, the embodiment of the present application can generate a behavior sequence decision by constructing an ego vehicle closed-loop strategy tree, screening interacting surrounding vehicles, pruning an ego vehicle behavior tree, and using a parallel computing mode and task allocation, and the flow thereof is as shown in Figure 5 The main steps can be:

[0144] Step S501: obtaining a generation value by using single-scene simulation of an ego vehicle and surrounding vehicle interaction model, as an upper limit of the cost.

[0145] Step S502: selecting surrounding vehicles satisfying certain interaction conditions.

[0146] The certain interaction conditions can be set by a person skilled in the art according to the actual situation, and the present application does not make specific limitations.

[0147] It can be understood that, in the forward simulation deduction process, the vehicles that do not interact with the ego vehicle are completely adopted with the data obtained in step S501. (For example, the trajectories generated by the same lane-changing operation are taken as a trajectory configuration set. L and R represent left lane-changing and right lane-changing operations respectively. L represents a set of all possible trajectories that complete left lane-changing once in the prediction time domain; RLR represents a set of all possible trajectories that complete right lane-changing, left lane-changing and right lane-changing in sequence once in the prediction time domain, such as [K, R, R, R, R, K, K, K, L, L, L, L, K, R, R, R, R, K, K] in 20s, which represents the right lane-changing, left lane-changing and right lane-changing behaviors with a duration of 4s at 2s, 10s and 15s respectively).

[0148] In addition, it should be noted that the interaction mode of the vehicles that interact with the ego vehicle in the embodiment of the present application can adopt the confidence reachable set, that is, according to a certain percentile of the comfortable acceleration in the driver parameter of the vehicle as the maximum acceleration, a certain percentile of the comfortable deceleration as the maximum deceleration, the speed of the vehicle at the prediction start time as the initial speed, the closest and farthest distances reached by the vehicle in the confidence range are calculated, and the distance contained in the middle is the confidence reachable set. The vehicles that intersect the confidence reachable set of the ego vehicle at the final time of the prediction time domain are selected as the interactive vehicles.

[0149] For example, the embodiment of the present application can take the 85% percentile of the comfortable acceleration in the driver parameter of the vehicle as the maximum acceleration, the 85% percentile of the comfortable deceleration as the maximum deceleration, the speed of the vehicle at the prediction start time as the initial speed, calculate the closest and farthest distances reached by the vehicle in the confidence range, and the distance contained in the middle is the confidence reachable set. The vehicles that intersect the confidence reachable set of the ego vehicle at the final time of the prediction time domain are selected as the interactive vehicles.

[0150] Step S503: constructing the ego vehicle closed-loop strategy tree and obtaining the trajectory configuration set.

[0151] In the embodiment of the present application, the ego vehicle closed-loop strategy tree can be constructed according to the domain knowledge, and the feasible trajectory configuration set can be constructed according to the current state and the traffic rules.

[0152] In addition, the construction of the ego vehicle closed-loop strategy tree in the embodiment of the present application can be performed in a variable time step manner, specifically as shown in Figure 6 A time limit t N is set, and when tN When the discrete step length is Δt1; when time t>t N When , the discrete step length is taken as Δt2. Among them, Δt1<Δt2, which reflects that the uncertainty of the prediction results increases with time. When the prediction time domain is very long, a multi-stage time step can be adopted, and a larger time discrete step length can be used in the longer time domain to save computing resources. The principle of step size selection ensures that the sum of all time steps can converge to a certain constant, that is, when t→∞, Δt i →∞, and

[0153] Step S504: perform task allocation in a parallel computing manner and dynamically decompose tasks according to the number of available cores to obtain a behavior sequence decision.

[0154] It can be understood that the embodiment of the present application performs task allocation in a parallel computing manner, and performs parallel search and pruning at the same time. The parallel search initializes M cores and performs parallel forward deduction based on different trajectory configurations. Each core is responsible for searching a subset of the trajectory configuration.

[0155] In the embodiment of the present application, pruning of the vehicle behavior tree can be performed using both heuristic and rule-based methods. The heuristic method may include, but is not limited to, training a heuristic function using a learning-based approach to estimate the cost of a particular situation; the rule-based method may include, but is not limited to, setting a rule such as automatically pruning a scenario if the incremental cost of a single scenario simulation exceeds a certain value.

[0156] For example, in the embodiment of the present application, the pruning of the vehicle behavior tree is based on a rule-based approach. The rule is set as follows: if the cumulative cost of a scene exceeds the recorded cost upper limit, the scene will be automatically pruned. For example, if the cost increment of a single scene simulation exceeds a value of 100 (related to the weight setting), the scene will be automatically pruned.

[0157] The task allocation in the parallel computing mode in the embodiment of the present application is the computing task allocation in the tree search algorithm based on the trajectory configuration parallelism. The cloud parallel core can adopt a means of parallel computing such as CPU (Central Processing Unit) or GPU (Graphics Processing Unit). Based on the feasibility, N different trajectory configurations are preliminarily screened. When considering the maximum number of lane change decisions, there is an upper limit N for the trajectory configuration. max , the number of parallel cores M≥N maxAccording to the discrete rules of the closed-loop strategy tree of the ego vehicle, the number of scenarios of each trajectory configuration can be calculated in advance, and the following principles are used: Principle one: minimize the maximum of the number of scenarios of all cores; Principle two: as evenly as possible, distribute all cores for parallel core trajectory allocation, and adopt a heuristic greedy algorithm. For each core, a subset of trajectory configurations is allocated, which contains several specific decision behavior possibilities. According to the idea of dynamic programming, the common part of the behavior tree of the ego vehicle in each core can be calculated only once, and the increased calculation caused by different specific behavior decisions in each trajectory configuration is reflected in the increased heterogeneous scenarios relative to the case without the second decision.

[0158] For example, the task allocation of the parallel computing mode in the embodiments of the present application is the calculation task allocation in the trajectory configuration-based parallel tree search algorithm, and the cloud parallel core adopts 24-core Intel@i9 CPU parallel computing. Specifically, as shown in Figure 7 , the scenario at the beginning of a specific planning period is shown, the ego vehicle has just passed the intersection, and there are 11 surrounding vehicles within a radius of 200m. Using the confidence reachable set method, the interactive surrounding vehicles are obtained as 7 vehicles, as shown in Figure 8 , which are labeled respectively. According to the preliminary feasibility screening, N different feasible trajectory configurations are obtained, and in the embodiments of the present application, N = 6 (trimmedSet = {R, RL, RR, RRR, RLR, RRL}). When considering that the maximum number of lane changing decisions is 3, the upper limit of the trajectory configuration is N max = 14, and the number of parallel cores M = 24 > N max = 14, as shown in Figure 9 . According to the discrete rules of the closed-loop strategy tree of the ego vehicle, the number of scenarios of each trajectory configuration can be calculated in advance, and the following principles are used:

[0159] Principle one: minimize the maximum of the number of scenarios of all cores.

[0160] Principle two: as evenly as possible, distribute all cores for parallel core trajectory allocation.

[0161] In the embodiments of the present application, a heuristic greedy algorithm is adopted to allocate a subset of trajectory configurations for each core, which contains several specific decision behavior possibilities. The goal of the algorithm is to evenly distribute the lane changing scenarios (such as "RLR" representing right, left, and right lane changing in turn) to parallel computing cores by reasonable division, so as to achieve the effect of load balancing, and the specific process is shown in Figure 10 , and the main steps include:

[0162] Step S1001: input scenario initialization.

[0163] In the embodiment of the present application, the number of lane changes and the corresponding time period (e.g., the time of each lane change) of each scene are initialized to generate a scene set, and the length of the scene set is discretized based on a fixed time step, and the time sequence length corresponding to each scene is recorded.

[0164] Step S1002: scene segmentation preparation.

[0165] In the embodiment of the present application, the number of lane changes corresponding to each scene is converted into the length of a "plank", and the length represents the calculation workload required by each scene, and the starting and ending time points (indices) of each scene are recorded to facilitate subsequent segmentation.

[0166] Step S1003: calculation of the number of required segmentations.

[0167] In the embodiment of the present application, the target core number M is determined, the difference between the current plank number and the target core number is calculated, and the number of segmentations required is determined.

[0168] Step S1004: greedy algorithm segmentation.

[0169] In the embodiment of the present application, the longest plank (i.e., the scene with the largest calculation amount) is segmented: the possible cutting points (accumulated workload) are calculated; a cutting point closest to the average workload is selected to ensure load balancing; the scene is divided into two parts, and the newly cut part is added to the plank set.

[0170] Step S1005: update of scene information.

[0171] In the embodiment of the present application, the starting and ending indices of the segmented scene are updated to ensure that each core is allocated a scene containing the workload after cutting.

[0172] Step S1006: core allocation.

[0173] In the embodiment of the present application, each sub-scene is allocated to a parallel core according to the segmentation result of the scene, and the allocation result can include but is not limited to the scene information and the corresponding workload of each core.

[0174] Step S1007: output of balanced result.

[0175] In the embodiment of the present application, the scene name processed by each core and its workload are outputted to ensure that the load of each core is balanced and the total calculation amount is optimal.

[0176] For example, the scene name assigned according to the above process in the embodiment of the present application, such as RRR_a00b01, represents the RRR scene in which the first lane change is completed within the time interval [0,1]. For the 6 trajectory configurations of trimmedSet given in this example, a subset of 20 trajectory configurations is refined according to the number of parallel cores 20: balancedSet = 20×1 array {'R_a00b17'}, {'RL_a00b02'}, {'RR_a00b02'}, {'RRR_a00b01'}, {'RLR_a00b01'}, {'RRL_a00b01'}, {'RRR_a01b02'}, {'RLR_a00b02'} '}, {'RRL_a01b02'}, {'RRR_a02b03'}, {'RLR_a02b03'}, {'RRL_a02b03'}, {'RL_a02b05'}, {'RR_a02b05'}, {'RRR_a03b06'}, {'RLR_a03b17'}, {'RRL_a03b17'}, {'RL_a05b17'}, {'RR_a05b17'}, {'RRR_a06b17'}.

[0177] Furthermore, in the embodiments of the present application, through the concept of dynamic programming, the common part of the closed-loop strategy tree of the ego-vehicle in each core can be calculated only once, and the increase in computation brought about by the different specific behavioral decisions in each trajectory configuration is reflected in the increased heterogeneity of scenarios compared to the case without including secondary decisions.

[0178] The embodiment of the present application constructs a closed-loop self-vehicle strategy tree and uses a confidence reachable set to screen interactive vehicles. A load-adaptive self-vehicle closed-loop strategy tree search algorithm for parallel computing is designed. The algorithm can dynamically decompose tasks according to the number of available cores. Heuristic pruning further improves the solution efficiency, ensuring the real-time and high efficiency of the system in predictive decision-making tasks involving multiple lane changes over a long time domain, thereby further improving the efficiency and real-time performance of self-vehicle decision-making.

[0179] In step S103 , the spatiotemporal trajectory of the ego vehicle is generated based on the behavior sequence decision and trajectory optimization model.

[0180] In some embodiments, the embodiments of the present application derive a trajectory optimization model that minimizes the lateral load transfer rate based on a simplified vehicle roll model, which is used to output a predictive driving space-time trajectory based on the results of smooth behavioral decisions, ensure the rollover safety of the vehicle, solve the lateral manipulation safety problem of the vehicle from the planning level, and output a feasible, safe and smooth space-time trajectory.

[0181] As a possible implementation method, the embodiment of the present application can use the behavior sequence decision and trajectory optimization model to generate the spatiotemporal trajectory of the vehicle.

[0182] Optionally, in an embodiment of the present application, the spatiotemporal trajectory of the ego vehicle is generated based on the behavior sequence decision and the trajectory optimization model, including: selecting a high-order polynomial based on the starting point and the termination point of the ego vehicle, wherein the high-order polynomial is greater than 5th order; generating boundary constraint conditions of the ego vehicle according to the behavior sequence decision; optimizing the residual degree of freedom coefficients of the boundary constraint conditions by using the trajectory optimization model to construct a trajectory optimization model for a rollover state index; solving the trajectory optimization model to obtain an optimal trajectory of the ego vehicle, and obtaining the spatiotemporal trajectory based on the optimal trajectory.

[0183] In some embodiments, the flow of generating the spatiotemporal trajectory of the ego vehicle according to the embodiments of the present application is as shown in Figure 11 The main content is as follows:

[0184] Step S1101: Select a high-order polynomial (greater than 5th order).

[0185] For example, the selected high-order polynomial in the embodiments of the present application is a 6th order polynomial, which can be but is not limited to represented as:

[0186]

[0187] Wherein a0, a1, …, a6 and b0, b1, …, b6 represent undetermined coefficients, a total of 14.

[0188] Step S1102: Generating boundary constraint conditions of the ego vehicle according to the behavior sequence decision.

[0189] For example, the embodiments of the present application can generate boundary constraint conditions such as starting point constraint and termination point constraint for each lane changing process curve according to the calculated decision, which is not specifically limited by the present application. In the embodiments of the present application, it is assumed that all lane changing lasts for 4s, i.e. t f = 4 (assuming that the time axis is shifted to t0s to facilitate problem construction).

[0190] In addition, the embodiments of the present application can determine the lateral position of the starting point and the ending point respectively according to the original lane and the target lane of the lane changing, and determine the longitudinal position of the starting point and the ending point respectively according to the speed information, and constrain the lateral speed and acceleration of the starting point and the ending point to be 0 to meet the trajectory feasibility and driving stability.

[0191] Wherein the starting point constraint can be but is not limited to represented as:

[0192]

[0193] The termination point constraint can be but is not limited to represented as:

[0194]

[0195] Step S1103: constructing a trajectory optimization model that minimizes LTR (Lateral-load Transfer Ratio) according to the remaining degrees of freedom;

[0196] Step S1104: solving the trajectory optimization model to obtain a time-space trajectory that satisfies a certain load transfer rate condition.

[0197] Further, the embodiments of the present application bring the 6th order polynomial into the starting point constraint, that is, the coefficients a0, a1, a2 and b0, b1, b2 can be directly solved and obtained, which can be respectively: 0x 0y ,

[0198] Further, the coefficients a3, a4, a5 and b3, b4, b5 of the embodiments of the present application are coupled together with the remaining coefficients a6, b6, and the coefficients a3, a4, a5 and b3, b4, b5 can be expressed by the remaining coefficients a6, b6 by bringing the end point constraint, which can be but not limited to expressed as:

[0199]

[0200] wherein,

[0201]

[0202] Let

[0203]

[0204] [b5 b4 b5] T = A -1 c-A -1 db6

[0205] , (31)

[0206] Therefore, only the remaining coefficients a6, b6 in x(t) and y(t) are to be determined, and once determined, a specific curve can be obtained immediately, and the lateral acceleration a y of the trajectory can be calculated after obtaining the trajectory. That is, the embodiments of the present application can obtain the lateral acceleration according to the velocity curve and the curvature curve.

[0207] wherein, the expressions of the velocity curve and the curvature curve can be but not limited to expressed as:

[0208]

[0209] The expression of the lateral acceleration can be but not limited to expressed as:

[0210] ​

[0211] Further, embodiments of the present application simplify the LTR calculation method as approximating LTR as a second-order inertia link of lateral acceleration of space-time trajectory, which can be but not limited to expressed as:

[0212]

[0213] Further, embodiments of the present application identify the coefficients ζ and ω of the simplified LTR model by recursive least squares method n , and other coefficients generated for given remaining coefficients a6, b6 can be directly obtained a y (t), according to input a y (t) to solve equation (31) to obtain LTR(t).

[0214] It should be noted that embodiments of the present application add vehicle steering actuator constraints:

[0215] The calculation formula of the trajectory front wheel steering angle can be but not limited to expressed as:

[0216] δ(t) = atan(Lk(T))

[0217] , (36)

[0218] The front wheel steering angle constraint and the rotation angular velocity constraint can be but not limited to expressed as:

[0219]

[0220] Further, the trajectory optimization model constructed by embodiments of the present application can be but not limited to expressed as:

[0221]

[0222] s.t. (35) system dynamics

[0223] (31) (34) lateral acceleration calculation

[0224] (37) actuator constraint

[0225] , (38)

[0226] Wherein, embodiments of the present application can use various numerical methods to quickly solve, and after solving, the optimal remaining coefficients a6, b6 are obtained to generate smooth executable and minimize LTR trajectory. Wherein, for the under-damped LTR case, the trajectory solved by embodiments of the present application is as follows: Figure 12As shown, the optimal LTR trajectory tends to have an asymmetric trajectory characteristic of slow first and intense later, and the maximum LTR reduction of the 5th polynomial symmetric trajectory is about 10% to 20% under different speed conditions. For the smoothing of all the trajectories, as shown in Figure 13 As shown, the planned smooth spatiotemporal trajectory.

[0227] The embodiment of the present application can generate a smooth spatiotemporal trajectory based on a trajectory optimization model, thereby guaranteeing the stability and safety of the vehicle during lateral operation, effectively preventing vehicle rollover, improving lateral maneuvering safety at the planning level, and enhancing vehicle maneuvering safety.

[0228] The following will be combined with Figure 14 As shown, a specific embodiment is used to introduce in detail the driving trajectory planning method based on vehicle-cloud collaboration proposed by the embodiment of the present application.

[0229] Among them, Figure 14 A block diagram of a driving trajectory planning system according to an embodiment of the present application is shown.

[0230] In Figure 14 , the driving trajectory planning system of the embodiment of the present application is composed of traffic (self-vehicle, surrounding vehicles, traffic participants, etc.), road infrastructure and roadside intelligent equipment, cloud control platform, communication network, and related support platform information (map platform, traffic control platform, etc., providing related information support). Among them, the self-vehicle (vehicle end) and the cloud control platform (cloud end) each contain an information space and a physical space, the cloud end information space contains a vehicle predictive planning algorithm, and a real-time digital twin of the surrounding traffic is established; the vehicle end continuously uploads the state information of the self-vehicle, and receives the predictive driving information of the cloud end to perform re-planning and trajectory tracking control according to the same.

[0231] According to the driving trajectory planning method based on vehicle-cloud cooperation provided in the embodiments of the present application, the vehicle probability model can be obtained based on the obtained target perception data of the ego vehicle and surrounding vehicles, and then the parameters and parameter uncertainty in the driver model and vehicle lane changing model, the intention and intention uncertainty of the ego vehicle are obtained, and the ego vehicle closed-loop strategy tree is constructed based on the parameters and parameter uncertainty, the intention and intention uncertainty of the ego vehicle. The search of the ego vehicle closed-loop strategy tree is performed in a parallel computing manner, and then the behavior sequence decision is obtained, the space-time trajectory is generated, the space-time trajectory of the ego vehicle is planned, the uncertainty of the driver parameters and intention is clearly reflected through the vehicle probability model, the uncertainty propagation mechanism avoids the logical error of "predicting first and planning second" while considering the uncertainty information, improves the credibility and feasibility of the planning result, realizes the highest solving efficiency through the ego vehicle closed-loop strategy tree and ensures the real-time performance. Good performance can be obtained in the long-time domain predictive decision task, the ego vehicle rollover safety is ensured through the trajectory optimization model, the lateral control safety problem of the ego vehicle is solved at the planning layer, and the feasible and safe smooth trajectory is output. Therefore, the problems of small observable range, limited single vehicle computing power, and difficulty in effectively solving the long-time domain planning problem in the related art are solved, and the problems of low availability of planning results and low prediction accuracy are solved.

[0232] Secondly, the driving trajectory planning device based on vehicle-cloud cooperation provided in the embodiments of the present application is described with reference to the accompanying drawings.

[0233] Figure 15 The block schematic diagram of the driving trajectory planning device based on vehicle-cloud cooperation provided in the embodiments of the present application is shown.

[0234] As shown in Figure 15 The driving trajectory planning device 10 based on vehicle-cloud cooperation includes a generation module 100, a construction module 200 and a planning module 300.

[0235] The generation module 100 is configured to obtain the target perception data of the ego vehicle and surrounding vehicles, update the initial surrounding vehicle probability model based on the target perception data, obtain the surrounding vehicle probability model, and obtain the parameters and parameter uncertainty in the driver model and vehicle lane changing model, and the intention and intention uncertainty of the ego vehicle based on the surrounding vehicle probability model.

[0236] The construction module 200 is configured to construct the ego vehicle closed-loop strategy tree based on the parameters and parameter uncertainty, the intention and intention uncertainty of the ego vehicle, search the ego vehicle closed-loop strategy tree in combination with the ego vehicle closed-loop strategy tree and the parallel computing manner, and generate the behavior sequence decision of the ego vehicle.

[0237] The planning module 300 is configured to generate the space-time trajectory of the ego vehicle based on the behavior sequence decision and the trajectory optimization model.

[0238] Optionally, in an embodiment of the present application, the generating module 100 comprises a first generating unit, an obtaining unit and an updating unit.

[0239] The first generating unit is configured to perform segmentation and extraction of road map information along the route according to the destination of the ego vehicle and global path planning information, so as to generate discrete road map information of the ego vehicle according to different positions of the ego vehicle, and establish an initial surrounding vehicle probability model.

[0240] The obtaining unit is configured to obtain target perception data of the ego vehicle and surrounding vehicles according to real-time traffic twins at the beginning of each planning period.

[0241] The updating unit is configured to, in each planning period, retrieve the discrete road map information according to the position of the ego vehicle, and update the initial surrounding vehicle probability model in combination with the discrete road map information and a parameter and intention recognition algorithm, so as to obtain a surrounding vehicle probability model, wherein the surrounding vehicle probability model comprises at least one of a surrounding vehicle intention and a probability thereof, a driver model and a vehicle lane changing model.

[0242] Optionally, in an embodiment of the present application, the parameter and intention recognition algorithm is based on Bayesian inference, wherein the Bayesian inference comprises: selecting a prior distribution of a parameter in the driver model based on an initial hypothesis; constructing a posterior distribution of the parameter in the intention, driver model and surrounding vehicle lane changing model based on the prior distribution; when the posterior distribution cannot be analyzed, sampling the to-be-recognized parameter by using a MCMC method to obtain sample data of the to-be-recognized parameter; and obtaining a to-be-recognized intention probability, parameter and uncertainty thereof based on the sample data.

[0243] Optionally, in an embodiment of the present application, the constructing module 200 comprises a first constructing unit, a first screening unit, a second generating unit and a third generating unit.

[0244] The first constructing unit is configured to construct a decision sequence of the ego vehicle containing multiple lane changing behaviors in a long time domain.

[0245] The first screening unit is configured to screen out surrounding vehicles interacting with the ego vehicle in combination with the decision sequence of the ego vehicle and the confidence reachable set.

[0246] The second generating unit is configured to obtain interaction information of the ego vehicle and the surrounding vehicles by using a forward reasoning simulation mechanism.

[0247] The third generating unit is configured to perform task allocation in a parallel computing mode based on the interaction information and a closed-loop strategy tree of the ego vehicle, and dynamically decompose tasks according to the number of available cores, so as to obtain a behavior sequence decision.

[0248] Optionally, in an embodiment of the present application, the parameters and parameter uncertainties in the driver model and the vehicle lane-changing model, the ego vehicle's intention and intention uncertainty, include: first-order uncertainty of the parameters in the driver model; first-order uncertainty of the vehicle lane-changing once; uncertainty of the ego vehicle intention entropy; cumulative effect of the uncertainty over time.

[0249] Optionally, in an embodiment of the present application, the planning module 300 includes: a second screening unit, a fourth generation unit, a second construction unit and a solving unit.

[0250] The second screening unit is configured to select a high-order polynomial based on the starting point and the end point constraints of the ego vehicle, wherein the high-order polynomial is greater than 5th order.

[0251] The fourth generation unit is configured to generate boundary constraint conditions of the ego vehicle according to the behavior sequence decision.

[0252] The second construction unit is configured to optimize the remaining degrees of freedom coefficients of the boundary constraint conditions by using the trajectory optimization model, so as to construct a trajectory optimization model for the rollover state index.

[0253] The solving unit is configured to solve the trajectory optimization model to obtain an optimal trajectory of the ego vehicle, and obtain a space-time trajectory based on the optimal trajectory.

[0254] It should be noted that the foregoing explanation and description of the embodiment of the driving trajectory planning method based on vehicle-cloud cooperation also apply to the embodiment of the driving trajectory planning device based on vehicle-cloud cooperation, which will not be described here.

[0255] According to the driving trajectory planning device based on vehicle-cloud cooperation provided in the embodiments of the present application, the vehicle probability model can be obtained based on the obtained target perception data of the ego vehicle and surrounding vehicles, and then the parameters and parameter uncertainty in the driver model and the vehicle lane-changing model, the intention and intention uncertainty of the ego vehicle are obtained, and the ego vehicle closed-loop strategy tree is constructed based on the parameters and parameter uncertainty, the intention and intention uncertainty of the ego vehicle, the search of the ego vehicle closed-loop strategy tree is performed in a parallel computing manner, and then the behavior sequence decision is obtained, the space-time trajectory is generated, the space-time trajectory of the ego vehicle is planned, the uncertainty of the driver parameters and intention is clearly reflected through the vehicle probability model, the uncertainty propagation mechanism avoids the logical error of "predicting first and then planning" while considering the uncertainty information, improves the credibility and feasibility of the planning result, the highest solving efficiency is realized through the ego vehicle closed-loop strategy tree and the real-time performance is ensured, good performance can be obtained in the long-time domain predictive decision task, the ego vehicle rollover safety is ensured through the trajectory optimization model, the lateral control safety of the ego vehicle is solved from the planning layer, and the feasible and safe smooth trajectory is output. Therefore, the problems in the related art that the single vehicle is limited by the field of view, the observable range is small, the single vehicle has limited computing power, and it is difficult to effectively solve the long-time domain planning problem, resulting in low availability of the planning result and low prediction accuracy are solved.

[0256] Figure 16 A structural diagram of a server according to an embodiment of the present application is provided. The server can include:

[0257] The memory 1601, the processor 1602, and the computer program stored in the memory 1601 and executable on the processor 1602.

[0258] The processor 1602 implements the driving trajectory planning method based on vehicle-cloud cooperation provided in the above embodiments when executing the program.

[0259] Further, the server further includes:

[0260] The communication interface 1603 is used for communication between the memory 1601 and the processor 1602.

[0261] The memory 1601 is used to store the computer program executable on the processor 1602.

[0262] The memory 1601 can include a high-speed RAM memory, and can also include a non-volatile memory such as at least one disk memory.

[0263] If the memory 1601, the processor 1602 and the communication interface 1603 are implemented independently, the communication interface 1603, the memory 1601 and the processor 1602 can be connected with each other through a bus and complete communication between 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, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For convenience of representation, Figure 16 Only one thick line is used to represent the bus in the figure, but it does not mean that there is only one bus or only one type of bus.

[0264] Optionally, in a specific implementation, if the memory 1601, the processor 1602 and the communication interface 1603 are integrated on a chip, the memory 1601, the processor 1602 and the communication interface 1603 can complete communication between each other through an internal interface.

[0265] The processor 1602 can be a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0266] The embodiments of the present application further provide a computer readable storage medium, having stored thereon a computer program, which, when executed by a processor, implements the above-mentioned driving trajectory planning method based on vehicle-cloud cooperation.

[0267] The embodiments of the present application further provide a computer program product, comprising a computer program, which, when executed by a processor, implements the above-mentioned driving trajectory planning method based on vehicle-cloud cooperation.

[0268] In the description of the application, reference to "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that a particular feature, structure, material, or characteristic being described is included in at least one embodiment or example of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment or example. Furthermore, the described specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples. In addition, the usage of "N" means at least two, for example, two, three or the like, unless explicitly stated otherwise.

[0269] Furthermore, the terms "first", "second", or the like, are used merely as a designation of certain elements or features of the application, and do not imply or connote relative importance or a specific order of precedence. Thus, features defined with "first", "second", etc. can include at least one of the features, either explicitly or implicitly.

[0270] Any process or method descriptions or blocks in flow charts or otherwise described herein represent embodiments of modules, segments, or portions of code which include one or more executable instructions for implementing specific logic functions or steps, and alternate implementations are possible. In some embodiments, the processes or methods described in flow charts or otherwise described herein are not necessarily performed in the order shown or discussed, including, for example, performing or depending from other operations or stages, in parallel, in reverse order, or in other orders.

[0271] The logic and / or steps represented in the flowcharts and / or described herein, for example, can be considered as a sequence of executable instructions stored in a computer readable medium, which can be executed by an instruction execution system, apparatus or device, such as a computer-based system, a processor-based system, or other system that can fetch the instructions from the instruction execution system, apparatus or device and execute the instructions, or a combination of the above. For the purposes of this specification, a "computer readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus or device. The computer readable medium can be a computer readable storage medium or a computer readable signal medium. The computer readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a propagation medium. The computer readable signal medium can include, but is not limited to, a computer readable medium that facilitates transfer of the program from one place to another. A specific example of a computer readable medium is a non-transitory computer-readable storage medium. A specific example of a computer readable signal medium is a source or destination of the computer readable medium. Another specific example of a computer readable signal medium is a computer readable signal travelling through space. Thus, a computer readable medium can take many forms of hardware to carry out the program for use by or in connection with the instruction execution system, apparatus or device.

[0272] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware and in another embodiment, the hardware can be implemented using any or a combination of the following technologies, which are each well known in the art: a discrete logic circuit(s) having logic gates for implementing logic functions upon an application of data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array(s) (PGA), a field programmable gate array (FPGA), etc.

[0273] Those of skill in the art would understand that the steps of the methods carried out above can be carried out wholly or partly by a program instructing relevant hardware, and the program can be stored in a computer readable storage medium, and when executed, includes one or a combination of the steps of the method embodiments.

[0274] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0275] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A driving track planning method based on vehicle-cloud cooperation, characterized in that, The method comprises the following steps: acquiring target perception data of the ego vehicle and surrounding vehicles, and updating an initial surrounding vehicle probability model based on the target perception data to obtain a surrounding vehicle probability model, so as to obtain parameters and parameter uncertainties in a driver model and a vehicle lane-changing model, intentions and intention uncertainties of the ego vehicle based on the surrounding vehicle probability model; constructing an ego vehicle closed-loop strategy tree based on the parameters and parameter uncertainties, the intentions and intention uncertainties of the ego vehicle, and performing search on the ego vehicle closed-loop strategy tree in combination with the ego vehicle closed-loop strategy tree and a parallel computing mode to generate a behavior sequence decision of the ego vehicle; generating a space-time trajectory of the ego vehicle based on the behavior sequence decision and a trajectory optimization model; wherein the acquiring target perception data of the ego vehicle and surrounding vehicles, and updating an initial surrounding vehicle probability model based on the target perception data to obtain a surrounding vehicle probability model comprises: segmenting and extracting road map information along a route according to a destination of the ego vehicle and global path planning information, so as to generate discrete road graph information of the ego vehicle according to different positions of the ego vehicle, and use the discrete road graph information to establish the initial surrounding vehicle probability model; acquiring target perception data of the ego vehicle and the surrounding vehicles according to real-time traffic twins at the beginning of each planning period; in each planning period, calling the discrete road graph information according to the position of the ego vehicle, and updating the initial surrounding vehicle probability model in combination with the discrete road graph information and a parameter and intention recognition algorithm to obtain a surrounding vehicle probability model, wherein the surrounding vehicle probability model comprises at least one of surrounding vehicle intentions and probabilities thereof, a driver model and a vehicle lane-changing model; the constructing an ego vehicle closed-loop strategy tree based on the parameters and parameter uncertainties, the intentions and intention uncertainties of the ego vehicle, and performing search on the ego vehicle closed-loop strategy tree in combination with the ego vehicle closed-loop strategy tree and a parallel computing mode to generate a behavior sequence decision of the ego vehicle comprises: constructing an ego vehicle decision sequence comprising multiple lane-changing behaviors in a long time domain; screening surrounding vehicles interacting with the ego vehicle in combination with the ego vehicle decision sequence and a confidence reachable set; obtaining interaction information of the ego vehicle and the surrounding vehicles by using a forward deduction simulation mechanism; based on the interaction information and the ego vehicle closed-loop strategy tree, performing task allocation in the parallel computing mode, and dynamically decomposing the tasks according to the number of available cores to obtain the behavior sequence decision; the parameters and parameter uncertainties in the driver model and the vehicle lane-changing model, and the intentions and intention uncertainties of the ego vehicle comprise: first-order uncertainty of parameters in the driver model; first-order uncertainty of a lane-changing behavior of a preceding vehicle; uncertainty of intention entropy of the ego vehicle; cumulative effect of uncertainty over time.

2. The method of claim 1, wherein, the parameter and intention recognition algorithm is based on Bayesian inference, wherein the Bayesian inference comprises: selecting a prior distribution of parameters in the driver model based on an initial hypothesis; constructing a posterior distribution of parameters in the intention, the driver model and the surrounding vehicle lane-changing model based on the prior distribution; when the posterior distribution cannot be analyzed, using a Markov Chain Monte Carlo (MCMC) method to sample a to-be-recognized parameter to obtain sample data of the to-be-recognized parameter; An intention probability to be identified, parameters to be identified, and uncertainties thereof are obtained based on the sample data.

3. The method of claim 1, wherein, The time-space trajectory of the ego vehicle is generated based on the behavior sequence decision and the trajectory optimization model, including: A high-order polynomial is selected based on the number of constraints of the starting point and the ending point of the ego vehicle, wherein the high-order polynomial is greater than 5th order; A boundary constraint condition of the ego vehicle is generated according to the behavior sequence decision; A trajectory optimization model for a rollover state index is constructed by optimizing the residual degree of freedom coefficients of the boundary constraint condition by using the trajectory optimization model; The optimal trajectory of the ego vehicle is obtained by solving the trajectory optimization model, and the time-space trajectory is obtained based on the optimal trajectory.

4. A driving track planning device based on vehicle-cloud cooperation, characterized in that, The device comprises: A generation module is configured to obtain target perception data of an ego vehicle and surrounding vehicles, and update an initial surrounding vehicle probability model based on the target perception data to obtain a surrounding vehicle probability model, so as to obtain parameters and parameter uncertainties in a driver model and a vehicle lane-changing model, intention and intention uncertainty of the ego vehicle based on the surrounding vehicle probability model; A construction module is configured to construct an ego vehicle closed-loop strategy tree based on the parameters and parameter uncertainties, the intention and intention uncertainty of the ego vehicle, and perform search on the ego vehicle closed-loop strategy tree in combination with the ego vehicle closed-loop strategy tree and a parallel computing mode to generate a behavior sequence decision of the ego vehicle; A planning module is configured to generate a time-space trajectory of the ego vehicle based on the behavior sequence decision and the trajectory optimization model. The generation module comprises: A first generation unit is configured to perform segmentation and extraction of road map information along a route of the ego vehicle according to a destination of the ego vehicle and global path planning information, so as to generate discrete road map information of the ego vehicle according to different positions of the ego vehicle, and establish the initial surrounding vehicle probability model; An acquisition unit is configured to acquire target perception data of the ego vehicle and the surrounding vehicles according to real-time traffic twins at the beginning of each planning period; An update unit is configured to, in each planning period, retrieve the discrete road map information according to a position of the ego vehicle, and update the initial surrounding vehicle probability model in combination with the discrete road map information and parameter and intention recognition algorithms to obtain a surrounding vehicle probability model, wherein the surrounding vehicle probability model comprises at least one of surrounding vehicle intention and probability, a driver model, and a vehicle lane-changing model. The construction module comprises: A first construction unit is configured to construct an ego vehicle decision sequence comprising multiple lane-changing behaviors in a long time domain; A first screening unit is configured to screen out surrounding vehicles interacting with the ego vehicle in combination with the ego vehicle decision sequence and a confidence reachable set; A second generation unit is configured to obtain interaction information of the ego vehicle and the surrounding vehicles by using a forward deduction simulation mechanism; A third generation unit is configured to perform task allocation of the parallel computing mode based on the interaction information and the ego vehicle closed-loop strategy tree, and dynamically decompose the tasks according to the number of available cores to obtain the behavior sequence decision. The parameters and parameter uncertainties in the driver model and the vehicle lane-changing model, and the intention and intention uncertainty of the ego vehicle comprise: First-order uncertainty of parameters in the driver model. first-order uncertainty of one lane change of the preceding vehicle; uncertainty of the ego vehicle intention entropy; accumulation effect of uncertainty over time.

5. A server, characterized by comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor executing the program to implement the method of claim 1-3.

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor for implementing the method of claim 1-3.

7. A computer program product, characterised in that, The program is executed by the processor for implementing the method of claim 1-3. The program is executed by the processor for implementing the method of claim 1-3.

Images