Automobile anthropomorphic lane changing trajectory planning method and device and electronic equipment

Abstract

The embodiment of the invention discloses an automobile anthropomorphic lane changing trajectory planning method and device and electronic equipment. The method comprises the following steps: preprocessing a natural driving data set and real-time collected data, extracting target state information, determining a lane changing starting moment, transverse and longitudinal prediction offset and lane changing duration of a target vehicle through a neural network vehicle dynamics model, then establishing a polynomial lane changing trajectory planning model, and determining a lane changing trajectory of the target vehicle according to the polynomial lane changing trajectory planning model. And determining a lane changing track cluster based on the constructed constraint conditions of the lane changing starting point and the lane changing ending point, and finally, comparing with the lane changing style coefficient of the driver through a vehicle collision detection algorithm, and screening out a safe target lane changing track conforming to the lane changing style of the driver. The target lane changing track obtained by the lane changing track planning method not only pays attention to vehicle dynamics and road geometric constraints, but also makes a lane changing decision in consideration of driving habits and styles of drivers, so that the satisfaction degree of users is improved.

Description

Technical Field

[0001] One or more embodiments of the present specification relate to the field of intelligent automobile technology, and in particular, to a method, device and electronic device for planning an anthropomorphic lane-changing trajectory for an automobile. Background Art

[0002] With the rapid development of intelligent vehicle technology, trajectory planning plays a vital role in the development of autonomous driving. Through reasonable trajectory planning, the autonomous driving system can accurately predict and respond to complex traffic conditions on the road, including vehicles, pedestrians, traffic signals and obstacles, so that the driver and passengers can have a safe, efficient and comfortable driving experience. Trajectory planning can also make full use of the vehicle's power performance and handling capabilities, so that the vehicle can make optimized driving decisions in different scenarios, such as reasonable lane changes, overtaking and turning. Therefore, with the rapid development of autonomous driving technology, anthropomorphic lane change trajectory planning has become a research hotspot in the field of autonomous driving. However, in traditional lane change trajectory planning, the focus is mainly on vehicle dynamics and road geometry constraints, lacking comprehensive consideration of driver behavior and intentions. However, in actual driving, drivers often make lane change decisions based on their own driving habits, driving style and surrounding traffic environment, resulting in low user satisfaction with traditional lane change trajectory planning methods. Summary of the invention

[0003] The embodiments of this specification provide a method, device and electronic device for planning a lane-changing trajectory for an automobile, and the technical solutions are as follows:

[0004] In a first aspect, an embodiment of the present specification provides a method for planning a humanized lane change trajectory for a vehicle, the method comprising:

[0005] Determine target state information of the target vehicle based on a natural driving data set and real-time collected data, and determine lateral and longitudinal offset values ​​and a lane change start time corresponding to the target state information based on a neural network algorithm, wherein the real-time collected data is obtained based on vehicle sensors;

[0006] Determining the lane changing duration corresponding to the lateral and longitudinal offset values ​​and the lane changing start time based on the constructed vehicle dynamics model;

[0007] Determine a lane-changing trajectory cluster corresponding to the lane-changing parameters of the vehicle based on a polynomial lane-changing trajectory planning model, wherein the lane-changing parameters of the vehicle include the lane-changing start time, the lateral and longitudinal predicted offsets, and the lane-changing duration;

[0008] The lane-changing trajectory cluster is screened out based on the lane-changing style coefficient of the driver of the target vehicle to obtain a target lane-changing trajectory.

[0009] In a second aspect, a vehicle anthropomorphic lane change trajectory planning device is provided, the device comprising:

[0010] An acquisition module determines target state information of a target vehicle based on a natural driving data set and real-time collected data, and determines lateral and longitudinal offset values ​​and a lane change start time corresponding to the target state information based on a neural network algorithm, wherein the real-time collected data is obtained based on vehicle sensors;

[0011] A determination module, used to determine the lane changing duration corresponding to the lateral and longitudinal offset values ​​and the lane changing start time based on the constructed vehicle dynamics model;

[0012] A planning module, configured to determine a lane-changing trajectory cluster corresponding to a lane-changing parameter of a vehicle based on a polynomial lane-changing trajectory planning model, wherein the lane-changing parameter of the vehicle includes the lane-changing start time, the predicted lateral and longitudinal offsets, and the lane-changing duration;

[0013] The screening module is used to screen out the lane-changing trajectory cluster based on the lane-changing style coefficient of the driver of the target vehicle to obtain a target lane-changing trajectory.

[0014] In a third aspect, an electronic device is provided, including a device processor and a memory;

[0015] The device processor is connected to the memory;

[0016] The memory is used to store executable program code;

[0017] The device processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the steps of the method provided in the first aspect or any possible implementation manner of the first aspect.

[0018] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and the computer-readable storage medium stores instructions, which, when the instructions are executed on a computer or a device processor, cause the computer or the device processor to execute the method provided in the first aspect or any possible implementation of the first aspect.

[0019] The beneficial effects brought by the technical solutions provided by some embodiments of this specification include at least:

[0020] In one or more embodiments of the present specification, the natural driving data set and the real-time collected data are preprocessed to extract the target state information, and the target vehicle's lane change start time, lateral and longitudinal predicted offsets and lane change duration are determined through the neural network vehicle dynamics model. Then, a polynomial lane change trajectory planning model is established, and the lane change trajectory cluster is determined based on the constructed lane change start and end point constraints. Finally, the vehicle collision detection algorithm and the driver's lane change style coefficient are compared to select a safe target lane change trajectory that conforms to the driver's lane change style. The target lane change trajectory obtained by the above lane change trajectory planning method not only pays attention to vehicle dynamics and road geometry constraints, but also takes into account the driver's driving habits and style to make lane change decisions, thereby improving user satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 A schematic diagram of the system architecture of a vehicle anthropomorphic lane-changing trajectory planning method provided in an embodiment of this specification;

[0023] Figure 2 A flow chart of a vehicle anthropomorphic lane-changing trajectory planning method provided in an embodiment of this specification;

[0024] Figure 3 A schematic diagram of a vehicle driving heading angle change provided in an embodiment of this specification;

[0025] Figure 4 A schematic diagram of a car lane changing scenario provided in an embodiment of this specification;

[0026] Figure 5 A schematic diagram of an LSTM neural network model identification provided in an embodiment of this specification;

[0027] Figure 6 A schematic diagram of a BP neural network model identification provided in an embodiment of this specification;

[0028] Figure 7 A schematic diagram of lane changing trajectories at different speeds provided in an embodiment of this specification;

[0029] Figure 8 A schematic diagram of collision screening of surrounding vehicle trajectories provided in an embodiment of this specification;

[0030] Fig. 9A schematic diagram of the structure of a vehicle anthropomorphic lane-changing trajectory planning device provided in an embodiment of this specification;

[0031] Fig.10 A schematic diagram of the structure of an electronic device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0032] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0033] The terms "first", "second", "third", etc. in the description and claims of this specification and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0034] See also Figure 1 , Figure 1 A schematic diagram of the system architecture of a vehicle humanized lane-changing trajectory planning method provided in an embodiment of this specification is shown.

[0035] like Figure 1 As shown, the system architecture of the vehicle humanized lane-changing trajectory planning method may include at least a terminal 10 , a server 20 and a network 30 .

[0036] The terminal 10 includes but is not limited to electronic devices such as smart phones, desktop computers, tablet computers, laptops, smart speakers, digital assistants, smart wearable devices, etc., and may also be software running on the above electronic devices, such as applications, etc. Optionally, the operating system running on the electronic device may include but is not limited to Android system, IOS system, Linux, Windows, etc. Optionally, the terminal 10 provides a lane change trajectory planning service to the user, and the terminal 10 may obtain a lane change trajectory planning instruction of an application program interface and send a lane change trajectory planning request to the server 20.

[0037] The server 20 can provide background services for the terminal 10. According to the lane change trajectory planning request sent by the terminal 10, the server 20 will obtain a series of lane change trajectory planning instructions, and the server 20 transmits the lane change trajectory planning instructions to other terminals 10 through the network 30. Specifically, the server 20 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0038] The network 30 is used to provide a medium for a communication link between the terminal 10 and the server 20. The network 30 may include various connection types, such as wired or wireless communication links or optical fiber cables.

[0039] In addition, it should be noted that Figure 1 What is shown is only one system provided by the present disclosure. In practical applications, other systems may also be included, for example, more terminals may be included.

[0040] In the embodiments of the present specification, the terminal 10 and the server 20 may be directly or indirectly connected via wired or wireless communication, which is not limited in the present disclosure.

[0041] See next Figure 2 , Figure 2 The overall flow chart of a vehicle anthropomorphic lane-changing trajectory planning method provided in an embodiment of the present specification is shown. The vehicle anthropomorphic lane-changing trajectory planning method can be used in the server 20 .

[0042] like Figure 2 As shown, the vehicle anthropomorphic lane-changing trajectory planning method may at least include the following steps:

[0043] Step 201: determine target state information of the target vehicle based on the natural driving data set and the real-time collected data, and determine the lateral and longitudinal offset values ​​and the lane change start time corresponding to the target state information based on a neural network algorithm.

[0044] Wherein, the real-time collected data is obtained based on vehicle sensors.

[0045] In the embodiment of the present specification, in order to perform anthropomorphic lane change trajectory planning for the target vehicle, it is necessary to first obtain a natural driving data set to extract the historical driving trajectory information of the target vehicle and surrounding vehicles, and then combine the real-time collection data collected by the visual sensor installed on the target vehicle to determine the target state information of the target vehicle.

[0046] Among them, Figure 3 As shown in the figure, when extracting historical driving trajectory information, the heading angle θ can be calculated for the vehicle trajectory in the natural driving data set, and the points with heading angle θ>θ0 are included in the candidate peak point set. The point with the largest heading angle in the candidate peak point set is selected as the lane change peak point, and the time t corresponding to the lane change peak point is p It is defined as the lane changing peak moment, and the vehicle corresponding to the trajectory of the lane changing peak point is designated as the target vehicle.

[0047] Next, from the obtained lane-changing peak time t p Search forward and backward to obtain the time point that first meets the preset condition in the heading angle of the corresponding lane change trajectory. The preset condition is |θ|<0.01°, thereby obtaining the initial lane change time interval [t1, t2] of the target vehicle. Then, since the initial lane change time interval obtained does not meet the requirements of the subsequent modules, the compensation time Δt can be added on the basis of the initial lane change interval to obtain the compensated lane change time interval [t1-Δt, t2+Δt].

[0048] Furthermore, the position point corresponding to the target vehicle at the initial lane change start time t1 is taken as the circle center O, R is the radius threshold, and all vehicles located within the circle at this time are defined as surrounding vehicles. In addition, all trajectories of the defined target vehicle and surrounding vehicles within the lane change time interval [t1-Δt, t2+Δt] are defined as a lane change scenario. Figure 4 As shown, the driving trajectory information of the target vehicle includes the horizontal coordinate sequence X of the target vehicle in the lane change time interval. host , the vertical coordinate sequence Y host , speed sequence V host , acceleration sequence A host and the body heading angle sequence Θ host , the lateral displacement sequence X of surrounding vehicles in the lane-changing time interval nbr , longitudinal offset sequence Y nbr , speed sequence V nbr , acceleration sequence A nbr and the body heading angle sequence Θ nbr , repeat the steps until all vehicles in the natural driving dataset are traversed to form a complete lane-changing dataset scene library.

[0049] Finally, the observation interval [tt obs ,t] the state information of the target vehicle and surrounding vehicles. In the observation interval [tt obs ,t], the horizontal coordinate sequence X of the target vehicle h_obs , the horizontal axis sequence X h_obs , the vertical coordinate sequence Y h_obs , speed sequence V h_obs, acceleration sequence A h_obs and the body heading angle sequence A h_obs , the horizontal coordinate sequence of surrounding vehicles X n_obs , the horizontal axis sequence X n_obs , the vertical coordinate sequence Y n_obs , speed sequence V n_obs , acceleration sequence A n_obs and the body heading angle sequence A n_obs

[0050] Then, different neural network models are constructed through different neural network algorithms, and the acquired target state information is input into each neural network model to obtain the lateral and longitudinal offset values ​​and lane change start time corresponding to the output of each model. Among them, LSTM and BP neural network models can be constructed to calculate the lateral and longitudinal offset values ​​and lane change start time corresponding to the target state information respectively.

[0051] In one possible implementation manner, the determining the lateral and longitudinal offset values ​​and the lane change start time corresponding to the target state information based on a neural network algorithm includes:

[0052] Determine the horizontal and vertical offset values ​​corresponding to the target state information based on the trained LSTM neural network model;

[0053] The lane change start time corresponding to the target state information is determined based on the trained BP neural network model.

[0054] In the embodiments of this specification, Figure 5 As shown in Figure 1, when the lateral and longitudinal offset values ​​and the lane change start time are determined by the neural network algorithm, the obtained target state information can be input into the trained LSTM neural network to perform forget gate identification to obtain the lateral offset L of the target vehicle at the end of the lane change. lat and longitudinal offset L lon , and input the target state information into the BP neural network, such as Figure 6 As shown in the figure, the hidden layer is identified to determine whether the target vehicle starts to plan the lane change trajectory at the current moment, and the lane change start time is defined as t start , the speed of the target vehicle at this moment is defined as v start , the acceleration is defined as a start , the horizontal axis x start and the ordinate y start Among them, x start =X h_obs [t start ],y start =Y h_obs [t start ],v start =V h_obs [t start ], astart =A h_obs [t start ].

[0055] Optionally, the output result of the BP neural network is specifically a numerical value K between [-1, 1]. When K>=0, it indicates that the lane change trajectory planning starts, and when K<0, it indicates that the lane change trajectory planning is not performed.

[0056] In one possible implementation manner, before determining the lateral and longitudinal offset values ​​and the lane change start time corresponding to the target state information based on the neural network algorithm, the method further includes:

[0057] Construct the initial LSTM neural network model and the initial BP neural network model;

[0058] Determine, based on the natural driving data set, first state data corresponding to the initial LSTM neural network model and second state data corresponding to the initial BP neural network model;

[0059] Perform forget gate weight training on the initial LSTM neural network model based on the first state data to obtain a trained LSTM neural network model;

[0060] The hidden layer weights of the initial BP neural network model are trained based on the second state data to obtain a trained BP neural network model.

[0061] In the embodiments of the present specification, before determining the lateral and longitudinal offset values ​​and the lane change start time through the neural network algorithm, it is necessary to first construct an initial LSTM neural network model and an initial BP neural network model, and train them respectively through the first state data and the second state data determined by the natural driving data set to obtain the trained LSTM neural network model and the BP neural network model.

[0062] Optionally, when training the LSTM neural network model, the first state data can be used as the input sequence X of the LSTM. input :

[0063] X input =[X host ,Y host ,V host ,A host ,X nbr ,Y nbr ,V nbr ,A nbr ]

[0064] Select the state information of the target vehicle at the end of the lane change as the output sequence Y of LSTM output .

[0065] Youtput =[X host ,Y host ][t2]

[0066] The input parameter x t and the hidden state h of the previous sequence t-1 Through the memory gate, the expression is as follows:

[0067] i t =sigmoid(ω i gx t +v i gh t-1 +b i )

[0068] g t =tanh(ω g gx t +v g gh t-1 +b g )

[0069] Among them, x t For input X input A single time step value, ω i ,v i ,ω g ,v g represents the state weight coefficient of the memory gate, b i ,b g Indicates the state offset value of the memory gate.

[0070] The input parameter x t and the hidden state h at the previous moment t-1 Through the forget gate, the expression is as follows:

[0071] f t =sigmoid(ω f gx t +v f gh t-1 +b f )

[0072] Among them, ω f ,v f represents the state weight coefficient of the forget gate, b f Indicates the state offset value of the forget gate.

[0073] The output results of the memory gate and the forget gate, as well as the cell state C at the previous moment t-1 Calculate the current cell state C t , the expression is as follows:

[0074] C t =ft e C t-1 +i t eg t

[0075] Calculate the value of the output gate O t and the current hidden state h t , the expression is as follows:

[0076] o t =σ(ω o gx t +v o gh t-1 +b o )

[0077] h t =o t e tanh(C t )

[0078] Among them, ω o ,v o represents the state weight coefficient of the output gate, b o Indicates the state offset value of the output gate.

[0079] The above steps complete the weight update and output layer output of the LSTM neural network in one time step. For a set of sequence data, we can get a series of output sequences by continuously moving forward. The errors of the output sequences and the training labels are back-propagated to complete the training of the entire LSTM neural network model.

[0080] Optionally, when training the BP neural network model, the second state data may be used as an input parameter x of the BP neural network. input :

[0081] x input =[X host ,Y host ,V host ,A host ,X nbr ,Y nbr ,V nbr ,A nbr ][t x ],t x ∈[t1-△t,t1]

[0082] t x Whether the target vehicle changes lanes at this moment is used as the output y of the BP neural network output :

[0083]

[0084] The input parameter xinput After the hidden layer of the BP neural network, the expression is as follows:

[0085] h=W h ex input +b h

[0086] Among them, W h represents the connection weight from the input layer neurons to the hidden layer neurons, b h Represents the bias value of the hidden layer neuron.

[0087] The h calculated by the hidden layer is passed through an activation function, usually a sigmoid function, and the expression is as follows:

[0088] h = sigmoid(h)

[0089] Among them, the sigmoid function is defined as:

[0090]

[0091] The h obtained by the hidden layer activation function is input to the output layer to calculate the output value of the BP neural network. The expression is as follows:

[0092] y=W o e h+b o

[0093] Among them, W o represents the connection weight from the hidden layer neurons to the output layer neurons, b o Represents the bias value of the output layer neuron.

[0094] According to the output result y and training label y output The error is back propagated, and the connection weights and bias values ​​in the BP neural network are continuously updated to complete the training of the BP neural network model.

[0095] Step 202: Determine the lane changing duration corresponding to the lateral and longitudinal offset values ​​and the lane changing start time based on the constructed vehicle dynamics model.

[0096] In the embodiment of this specification, since a general vehicle needs to change lanes during driving, in order to reduce the lane change time, the vehicle will be driven at a constant speed or uniform acceleration, so a vehicle dynamics model can be first constructed, and the theoretical acceleration range of the target vehicle can be determined in turn. Then, the obtained lateral and longitudinal offset values ​​and the lane change start time are combined with the theoretical acceleration range to perform physical model calculations to obtain the corresponding lane change time.

[0097] In one possible implementation, determining the lane changing duration corresponding to the lateral and longitudinal offset values ​​and the lane changing start time based on the constructed vehicle dynamics model includes:

[0098] Determining an acceleration range of the target vehicle based on the constructed vehicle dynamics model;

[0099] Determining a lane-changing time interval of the target vehicle based on dynamic constraints, wherein the dynamic constraints include the acceleration interval, the lateral and longitudinal offset values, and the lane-changing start time;

[0100] The lane-changing duration interval is traversed to obtain the lane-changing duration of the target vehicle.

[0101] In the embodiment of this specification, a vehicle dynamics model can be established first to calculate and determine the acceleration range [0, a max ], considering the safety of lane change, drivers usually try to shorten the lane change time as much as possible. Therefore, it is only necessary to consider that the vehicle maintains a constant speed or uniform acceleration during the lane change process. Then, according to the obtained acceleration interval [0, a max ], select the acceleration a from it, when a=0 or a max According to the previously determined dynamic constraints, the lane change start time t start The corresponding target vehicle speed v start , and the determined target vehicle lateral L lat and longitudinal offset L lon , calculate the shortest lane change time t min and the maximum lane change time t max :

[0102] L lon =v start ×t min +0.5×a max ×(t min ) 2

[0103] L lon =v start ×t max

[0104] Furthermore, according to the calculated shortest lane change time t min and the maximum lane change time t max , obtain the target vehicle lane change time interval [t min ,t max ], and loop through the lane change time interval [t min ,t max ], and get the lane-changing time t of the target vehicle c .

[0105] Among them, the traversal method is, t c =(t max -tmin )×0.2+t min

[0106] Therefore, according to the target vehicle lane change starting time t start and the target vehicle lane change time t c , and obtain the complete lane-changing time interval of the target vehicle [t start ,t start +t c ].

[0107] Step 203: Determine a lane-changing trajectory cluster corresponding to the vehicle lane-changing parameters based on the polynomial lane-changing trajectory planning model.

[0108] Among them, the vehicle lane changing parameters include the lane changing start time, lateral and longitudinal predicted offsets and lane changing duration.

[0109] In the embodiments of the present specification, after obtaining the vehicle lane changing parameters of the target vehicle, i.e., the lane changing start time, the predicted lateral and longitudinal offsets, and the lane changing duration, a polynomial lane changing trajectory planning model can be constructed first, and the vehicle lane changing parameters can be input into the polynomial lane changing trajectory planning model to determine each trajectory constraint, and then solve each theoretical lane changing trajectory under the lane changing duration based on each trajectory constraint and the lane changing function, and then count all the theoretical lane changing trajectories to obtain a lane changing trajectory cluster.

[0110] In one possible implementation, the determining of a lane changing trajectory cluster corresponding to a lane changing parameter of a vehicle based on a polynomial lane changing trajectory planning model includes:

[0111] Constructing a polynomial lane-changing trajectory planning model based on a global coordinate system method, and determining a trajectory starting point constraint and a trajectory end point constraint corresponding to the vehicle lane-changing parameters;

[0112] A lane changing trajectory cluster corresponding to the trajectory starting point constraint and the trajectory ending point constraint is determined based on the polynomial lane changing trajectory planning model.

[0113] In the embodiment of the present specification, a lane-changing trajectory planning model of a quintic polynomial is first established based on the global coordinate system method.

[0114] Among them, the lane-changing trajectory of the quintic polynomial is established in the X and Y directions of the global coordinate system respectively:

[0115] x(T)=a0+a1T+a2T 2 +a3T 3 +a4T 4 +a5T 5

[0116] y(T)=b0+b1T+b2T 2 +b3T 3 +b4T 4 +b5T5

[0117] Taking the first-order derivative of the lane-changing trajectory function in the above equation, we can obtain the speed functions in the X and Y directions, namely:

[0118]

[0119]

[0120] The velocity function of the above formula is derived again to obtain the acceleration function in the X and Y directions, that is:

[0121]

[0122]

[0123] Where T is time, a i ,b i ,i=0,1,…,5 are the coefficients of the lane changing trajectory function.

[0124] Next, the constraint conditions for solving the lane-changing trajectory function are established, where the constraint conditions for the lane-changing trajectory function include the trajectory starting point constraint and the trajectory ending point constraint. The target vehicle speed v corresponding to the lane-changing start time is obtained. start , acceleration a start and the horizontal coordinate x start and the ordinate y start , as the starting point constraint of the trajectory, and the ordinate y of the target vehicle’s lane change endpoint is obtained start +L lat and the horizontal coordinate x start +L lon As the trajectory endpoint constraint, t c As the lane change duration. Generally, the lateral velocity and lateral acceleration of the vehicle are 0 at the start of the lane change, and the lateral velocity, lateral acceleration and longitudinal acceleration of the vehicle are all 0 at the end of the lane change. Further, according to the established lane change function and the determined starting and ending constraints of the lane change trajectory, the lane change trajectory under the current lane change duration is solved. Specifically, the starting and ending constraints are:

[0125]

[0126] Repeat the above steps to obtain lane-changing trajectory clusters with different lane-changing durations, such as Figure 7 As shown, since the starting point and end point coordinates of the lane changing trajectory are determined, the lane changing paths are the same. The difference is that each lane changing trajectory is planned at a different speed to reflect the lane changing style of different drivers.

[0127] Step 204: Screen out the lane-changing trajectory cluster based on the driver's lane-changing style coefficient of the target vehicle to obtain a target lane-changing trajectory.

[0128] In the embodiment of the present specification, after the lane changing trajectory cluster corresponding to the target vehicle is obtained, since each lane changing trajectory in the lane changing trajectory cluster corresponds to a lane changing style of the driver, the obtained lane changing trajectory cluster can be filtered out based on the lane changing style coefficient of the driver in the target vehicle to obtain the anthropomorphic target lane changing trajectory of the vehicle.

[0129] In one possible implementation, the filtering out the lane changing trajectory cluster based on the lane changing style coefficient of the driver of the target vehicle to obtain the target lane changing trajectory includes:

[0130] Determining predicted trajectory information of surrounding vehicles within the lane change duration based on the target state information;

[0131] Traversing the lane-changing trajectory cluster based on the surrounding vehicle predicted trajectory information to obtain each safe lane-changing trajectory;

[0132] The safe lane-changing trajectories are screened out based on the driver's lane-changing style coefficient to obtain a target lane-changing trajectory.

[0133] In the embodiment of this specification, when selectively filtering out the lane-changing trajectory cluster, the predicted trajectory information of the surrounding vehicles corresponding to the target vehicle within the lane-changing duration can be first determined through the target state information. In the same way, the LSTM neural network model can be used to predict the surrounding vehicle information in the target state information to obtain the surrounding vehicle prediction trajectory information within the lane-changing duration [t start ,t start +t c ] The predicted trajectory information of surrounding vehicles in the cluster. Then, according to the extracted predicted trajectory of surrounding vehicles, the lane-changing trajectories in all lane-changing trajectory clusters are traversed, and the lane-changing trajectories that may collide with surrounding vehicles are filtered out to obtain each safe lane-changing trajectory.

[0134] Among them, Figure 8 As shown, at the same time, when the enclosing boxes of the target vehicle and the surrounding vehicles do not overlap, it is considered that there is no collision risk between the target vehicle and the surrounding vehicles at this time, otherwise it is considered that there is a collision risk between the target vehicle and the surrounding vehicles.

[0135] Next, the lane changing style coefficient of the driver in the target vehicle is determined, and each safe lane changing trajectory is screened out according to the lane changing style coefficient of the driver, and the target lane changing trajectory closest to the lane changing style coefficient of the target driver is selected.

[0136] In one possible implementation, the filtering out each of the safe lane changing trajectories based on the driver's lane changing style coefficient to obtain a target lane changing trajectory includes:

[0137] Determining each safe lane changing style coefficient corresponding to each of the safe lane changing trajectories and a current driver lane changing style coefficient corresponding to the target vehicle;

[0138] The least square difference calculation is performed on each of the safe lane changing style coefficients and the current driver's lane changing style coefficient, and each of the safe lane changing trajectories is screened out based on the calculation results to obtain a target lane changing trajectory.

[0139] In the embodiment of this specification, the driver's lane changing style coefficient is first established to determine the safe lane changing style coefficients S corresponding to different safe lane changing trajectories:

[0140]

[0141] Among them, a ymean represents the average lateral acceleration of the target vehicle during the lane change process, v ymean represents the average lateral velocity of the target vehicle during the lane changing process, and d represents the lateral offset of the target vehicle during the lane changing process.

[0142]

[0143] d=Y host [t2]-Y host [t1]

[0144] Similarly, the lane-changing style coefficient S' of the current driver in the target vehicle corresponding to the historical driving process is calculated.

[0145] Next, the square difference of each safe lane-changing style coefficient S and the current driver's lane-changing style coefficient S' is calculated to obtain the square difference e. Further, the lane-changing trajectory corresponding to the smallest square difference e is selected according to the calculation results, that is, the anthropomorphic target lane-changing trajectory of the car is obtained.

[0146] See next Fig. 9 , Fig. 9 The structure diagram of a vehicle anthropomorphic lane change trajectory planning device provided in the embodiment of this specification is shown. It should be noted that: Fig. 9 The vehicle anthropomorphic lane change trajectory planning device shown is used to implement the present application Figure 2 For the convenience of explanation, only the part related to the embodiment of the present application is shown. For the specific technical details not disclosed, please refer to the present application. Figure 2 The embodiment shown.

[0147] like Fig. 9 As shown, the vehicle anthropomorphic lane-changing trajectory planning device may at least include:

[0148] An acquisition module 901 determines target state information of a target vehicle based on a natural driving data set and real-time collected data, and determines lateral and longitudinal offset values ​​and a lane change start time corresponding to the target state information based on a neural network algorithm, wherein the real-time collected data is obtained based on a vehicle sensor;

[0149] A determination module 902 is used to determine the lane change duration corresponding to the lateral and longitudinal offset values ​​and the lane change start time based on the constructed vehicle dynamics model;

[0150] A planning module 903, configured to determine a lane-changing trajectory cluster corresponding to a lane-changing parameter of a vehicle based on a polynomial lane-changing trajectory planning model, wherein the lane-changing parameter of the vehicle includes the lane-changing start time, the predicted lateral and longitudinal offsets, and the lane-changing duration;

[0151] The screening module 904 is used to screen out the lane-changing trajectory cluster based on the lane-changing style coefficient of the driver of the target vehicle to obtain a target lane-changing trajectory.

[0152] In one possible implementation, the acquisition module 901 is specifically used for:

[0153] Determine the horizontal and vertical offset values ​​corresponding to the target state information based on the trained LSTM neural network model;

[0154] The lane change start time corresponding to the target state information is determined based on the trained BP neural network model.

[0155] In one possible implementation, the acquisition module 901 is further specifically used for:

[0156] Construct the initial LSTM neural network model and the initial BP neural network model;

[0157] Determine, based on the natural driving data set, first state data corresponding to the initial LSTM neural network model and second state data corresponding to the initial BP neural network model;

[0158] Perform forget gate weight training on the initial LSTM neural network model based on the first state data to obtain a trained LSTM neural network model;

[0159] The hidden layer weights of the initial BP neural network model are trained based on the second state data to obtain a trained BP neural network model.

[0160] In one possible implementation, the determination module 902 is specifically configured to:

[0161] Determining an acceleration range of the target vehicle based on the constructed vehicle dynamics model;

[0162] Determining a lane-changing time interval of the target vehicle based on dynamic constraints, wherein the dynamic constraints include the acceleration interval, the lateral and longitudinal offset values, and the lane-changing start time;

[0163] The lane-changing duration interval is traversed to obtain the lane-changing duration of the target vehicle.

[0164] In one possible implementation, the planning module 903 is specifically used for:

[0165] Constructing a polynomial lane-changing trajectory planning model based on a global coordinate system method, and determining a trajectory starting point constraint and a trajectory end point constraint corresponding to the vehicle lane-changing parameters;

[0166] A lane changing trajectory cluster corresponding to the trajectory starting point constraint and the trajectory ending point constraint is determined based on the polynomial lane changing trajectory planning model.

[0167] In one embodiment, the screening module 904 is further configured to:

[0168] Determining predicted trajectory information of surrounding vehicles within the lane change duration based on the target state information;

[0169] Traversing the lane-changing trajectory cluster based on the surrounding vehicle predicted trajectory information to obtain each safe lane-changing trajectory;

[0170] The safe lane-changing trajectories are screened out based on the driver's lane-changing style coefficient to obtain a target lane-changing trajectory.

[0171] In one embodiment, the screening module 904 is further configured to:

[0172] Determining each safe lane changing style coefficient corresponding to each of the safe lane changing trajectories and a current driver lane changing style coefficient corresponding to the target vehicle;

[0173] The least square difference calculation is performed on each of the safe lane changing style coefficients and the current driver's lane changing style coefficient, and each of the safe lane changing trajectories is screened out based on the calculation results to obtain a target lane changing trajectory.

[0174] Those skilled in the art can clearly understand that the technical solutions of the embodiments of the present application can be implemented with the help of software and / or hardware. The "unit" and "module" in this specification refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a field programmable gate array (Field-Programmable Gate Array, FPGA), an integrated circuit (Integrated Circuit, IC), etc.

[0175] Each processing unit and / or module of the embodiments of the present application may be implemented by an analog circuit that implements the functions described in the embodiments of the present application, or may be implemented by software that executes the functions described in the embodiments of the present application.

[0176] See next Fig.10 , Fig.10 A schematic diagram of the structure of an electronic device provided in an embodiment of this specification is shown.

[0177] like Fig.10 As shown, the electronic device 1000 may include: at least one device processor 1001 , at least one network interface 1004 , a user interface 1003 , a memory 1005 and at least one communication bus 1002 .

[0178] The communication bus 1002 may be used to realize the connection and communication among the above-mentioned components.

[0179] The user interface 1003 may include buttons, and the optional user interface may also include a standard wired interface or a wireless interface.

[0180] The network interface 1004 may include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, etc.

[0181] Among them, the device processor 1001 may include one or more processing cores. The device processor 1001 uses various interfaces and lines to connect the various parts of the entire electronic device 1000, and executes various functions and processes data of the electronic device 1000 by running or executing instructions, programs, code sets or instruction sets stored in the memory 1005, and calling data stored in the memory 1005. Optionally, the device processor 1001 can be implemented in at least one hardware form of DSP, FPGA, and PLA. The device processor 1001 can integrate one or a combination of CPU, GPU, modem, etc. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the device processor 1001, and it can be implemented separately through a chip.

[0182] Among them, the memory 1005 may include RAM and ROM. Optionally, the memory 1005 includes a non-transitory computer-readable medium. The memory 1005 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 1005 may optionally be at least one storage device located away from the aforementioned device processor 1001. As Fig.10 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and program instructions.

[0183] Specifically, the device processor 1001 may be used to call the car anthropomorphic lane change trajectory planning application stored in the memory 1005, and specifically perform the following operations:

[0184] Determine target state information of the target vehicle based on a natural driving data set and real-time collected data, and determine lateral and longitudinal offset values ​​and a lane change start time corresponding to the target state information based on a neural network algorithm, wherein the real-time collected data is obtained based on vehicle sensors;

[0185] Determining the lane changing duration corresponding to the lateral and longitudinal offset values ​​and the lane changing start time based on the constructed vehicle dynamics model;

[0186] Determine a lane-changing trajectory cluster corresponding to the lane-changing parameters of the vehicle based on a polynomial lane-changing trajectory planning model, wherein the lane-changing parameters of the vehicle include the lane-changing start time, the lateral and longitudinal predicted offsets, and the lane-changing duration;

[0187] The lane-changing trajectory cluster is screened out based on the lane-changing style coefficient of the driver of the target vehicle to obtain a target lane-changing trajectory.

[0188] A person skilled in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by entering a program to instruct related hardware, and the program may be stored in a computer-readable memory, which may include a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0189] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A vehicle anthropomorphic lane-changing trajectory planning method, characterized in that: The method comprises: Determine target state information of the target vehicle based on a natural driving data set and real-time collected data, and determine lateral and longitudinal offset values ​​and a lane change start time corresponding to the target state information based on a neural network algorithm, wherein the real-time collected data is obtained based on vehicle sensors; Determining the lane changing duration corresponding to the lateral and longitudinal offset values ​​and the lane changing start time based on the constructed vehicle dynamics model; Determine a lane-changing trajectory cluster corresponding to the lane-changing parameters of the vehicle based on a polynomial lane-changing trajectory planning model, wherein the lane-changing parameters of the vehicle include the lane-changing start time, the lateral and longitudinal predicted offsets, and the lane-changing duration; The lane-changing trajectory cluster is screened out based on the lane-changing style coefficient of the driver of the target vehicle to obtain a target lane-changing trajectory.

2. The method according to claim 1, characterized in that The determining of the lateral and longitudinal offset values ​​and the lane change start time corresponding to the target state information based on the neural network algorithm includes: Determine the horizontal and vertical offset values ​​corresponding to the target state information based on the trained LSTM neural network model; The lane change start time corresponding to the target state information is determined based on the trained BP neural network model.

3. The method according to claim 1, characterized in that Before determining the lateral and longitudinal offset values ​​and the lane change start time corresponding to the target state information based on the neural network algorithm, the method further includes: Construct the initial LSTM neural network model and the initial BP neural network model; Determine, based on the natural driving data set, first state data corresponding to the initial LSTM neural network model and second state data corresponding to the initial BP neural network model; Perform forget gate weight training on the initial LSTM neural network model based on the first state data to obtain a trained LSTM neural network model; The hidden layer weights of the initial BP neural network model are trained based on the second state data to obtain a trained BP neural network model.

4. The method according to claim 1, characterized in that The determining, based on the constructed vehicle dynamics model, the lane changing duration corresponding to the lateral and longitudinal offset values ​​and the lane changing start time includes: Determining an acceleration range of the target vehicle based on the constructed vehicle dynamics model; Determining a lane-changing time interval of the target vehicle based on dynamic constraints, wherein the dynamic constraints include the acceleration interval, the lateral and longitudinal offset values, and the lane-changing start time; The lane-changing duration interval is traversed to obtain the lane-changing duration of the target vehicle.

5. The method according to claim 1, characterized in that The method of determining a lane-changing trajectory cluster corresponding to a lane-changing parameter of a vehicle based on a polynomial lane-changing trajectory planning model includes: Constructing a polynomial lane-changing trajectory planning model based on a global coordinate system method, and determining a trajectory starting point constraint and a trajectory end point constraint corresponding to the vehicle lane-changing parameters; A lane changing trajectory cluster corresponding to the trajectory starting point constraint and the trajectory ending point constraint is determined based on the polynomial lane changing trajectory planning model.

6. The method according to claim 1, characterized in that The filtering out the lane-changing trajectory cluster based on the lane-changing style coefficient of the driver of the target vehicle to obtain a target lane-changing trajectory includes: Determining predicted trajectory information of surrounding vehicles within the lane change duration based on the target state information; Traversing the lane-changing trajectory cluster based on the surrounding vehicle predicted trajectory information to obtain each safe lane-changing trajectory; The safe lane-changing trajectories are screened out based on the driver's lane-changing style coefficient to obtain a target lane-changing trajectory.

7. The method according to claim 6, characterized in that The step of screening out the safe lane-changing trajectories based on the driver's lane-changing style coefficient to obtain a target lane-changing trajectory includes: Determining each safe lane changing style coefficient corresponding to each of the safe lane changing trajectories and a current driver lane changing style coefficient corresponding to the target vehicle; The least square difference calculation is performed on each of the safe lane changing style coefficients and the current driver's lane changing style coefficient, and each of the safe lane changing trajectories is screened out based on the calculation results to obtain a target lane changing trajectory.

8. A vehicle anthropomorphic lane-changing trajectory planning device, characterized in that: The device comprises: An acquisition module determines target state information of a target vehicle based on a natural driving data set and real-time collected data, and determines lateral and longitudinal offset values ​​and a lane change start time corresponding to the target state information based on a neural network algorithm, wherein the real-time collected data is obtained based on vehicle sensors; A determination module, used to determine the lane changing duration corresponding to the lateral and longitudinal offset values ​​and the lane changing start time based on the constructed vehicle dynamics model; A planning module, configured to determine a lane-changing trajectory cluster corresponding to a lane-changing parameter of a vehicle based on a polynomial lane-changing trajectory planning model, wherein the lane-changing parameter of the vehicle includes the lane-changing start time, the predicted lateral and longitudinal offsets, and the lane-changing duration; The screening module is used to screen out the lane-changing trajectory cluster based on the lane-changing style coefficient of the driver of the target vehicle to obtain a target lane-changing trajectory.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, wherein the computer-readable storage medium has instructions stored therein, and when the instructions are executed on a computer or a processor, the computer or the processor executes the steps of the method according to any one of claims 1 to 7.

Images