A method, device, equipment and storage medium for autonomous driving decision making

Abstract

The present application discloses an automatic driving decision-making method, device, equipment and storage medium. The method includes: based on the vehicle information of the target self-vehicle and the vehicle information of other interactive vehicles, respectively calculating the conflict information between the target self-vehicle and each interactive vehicle with driving conflicts; determining the collision risk assessment result of the interactive vehicle to the target self-vehicle according to the conflict information; correcting the collision risk assessment result based on historical decision data, determining the output of the model constructed by multivariate Gaussian distribution to produce the future risk assessment result; generating an automatic driving decision according to the future risk assessment result. Thus, through the two-dimensional full-plane risk assessment method, the optimal joint strategy numerical solution is analyzed, and the self-vehicle driving decision is automatically and reliably made. In the case of a small amount of data parameter adjustment, the processing efficiency is improved, the interaction process between vehicles is effectively considered, and the reliability of the automatic driving decision is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, specifically to technical fields such as intelligent transportation and autonomous driving, and in particular to an autonomous driving decision-making method, device, equipment and storage medium. Background Art

[0002] With the maturity of AI and autonomous driving technology, autonomous driving vehicles will face huge challenges in the future, namely how to infer the interactive behaviors or intentions of other vehicles in the event of time and space conflicts with other traffic vehicles, and then formulate targeted driving strategies.

[0003] In the existing technology, some researchers use rule-based methods such as filtering to model the future behavior of interactive vehicles based on their historical frame information, and then specify driving strategies for the vehicle. However, the model / rule-based intention prediction method can well model the historical information of interactive vehicles, but there is no single intelligent agent in real traffic scenarios, and the behavior of vehicles will be adjusted in real time due to the behavior of other vehicles. Therefore, the effectiveness of the model will decrease as the time domain is extended.

[0004] Therefore, it is urgent to find a new autonomous driving decision-making method to solve the above technical problems. Summary of the invention

[0005] The present application provides an autonomous driving decision-making method, device, equipment and storage medium to solve the problem that the behavior of other vehicles affects the behavior of the vehicle, resulting in reduced model effectiveness, which in turn affects the accuracy of autonomous driving decisions.

[0006] The technical solution is as follows:

[0007] In a first aspect, an autonomous driving decision-making method is provided, comprising:

[0008] Obtain vehicle information of the target vehicle in the autonomous driving scenario, as well as vehicle information of other interacting vehicles;

[0009] Based on the vehicle information of the target vehicle and the vehicle information of the other interactive vehicles, respectively calculate the conflict information between the target vehicle and each interactive vehicle with which the driving conflict exists; wherein the conflict information at least includes the driving conflict zone representation information and the driving conflict representation information between the target vehicle and the interactive vehicle;

[0010] Determine a collision risk assessment result of the interactive vehicle to the target vehicle according to the collision information;

[0011] Correcting the collision risk assessment result based on historical decision data, and determining a model constructed with a multivariate Gaussian distribution to output future risk assessment results;

[0012] An autonomous driving decision is generated according to the future risk assessment result, and the autonomous driving decision is used to assist the target vehicle in avoiding collision with other interactive vehicles during the autonomous driving process.

[0013] In a possible implementation, the vehicle information of the target vehicle at least includes a planned path, and the vehicle information of the other interactive vehicles at least includes a predicted trajectory;

[0014] Based on the vehicle information of the target vehicle and the vehicle information of the other interactive vehicles, conflict information between the target vehicle and each interactive vehicle with driving conflict is calculated respectively, specifically including:

[0015] For each of the target vehicle and other interacting vehicles, a polygonal box representing the shape of the vehicle is constructed respectively, and the polygonal box of each vehicle is decomposed based on the rotation invariant primitive to construct a discrete conflict primitive;

[0016] Based on the discrete conflict primitives, the planned path of the target ego vehicle and the predicted trajectories of the other interactive vehicles, conflict information of each interactive vehicle that has a driving conflict with the target ego vehicle is calculated.

[0017] In a possible implementation, determining a collision risk assessment result of the interactive vehicle to the target vehicle according to the collision information specifically includes:

[0018] According to the following formula, the collision probability of each interactive vehicle to the target vehicle in different risk modes is calculated based on the conditional random field:

[0019]

[0020] Among them, the , D stands for danger, A stands for attention, and S stands for safety; It indicates the time when the target vehicle reaches the conflict area in normal driving; Indicates the time it takes for the interacting vehicles to reach the conflict area in normal driving; Indicates the time it takes for the interactive vehicle to slow down and drive to a safe area; represents the risk assessment value of the interactive vehicle j.

[0021] In a possible implementation, the collision risk assessment result is corrected based on historical decision data, and a model constructed by multivariate Gaussian distribution is determined to output a future risk assessment result, specifically including:

[0022] Determine the historical decision data under the current driving scenario, collect collision and non-collision information, and count the feature quantity and collision result corresponding to each decision data in the historical decision data, wherein the feature quantity corresponding to each decision data is the collision possibility of the historical interactive vehicle in the decision data with the historical target vehicle in different risk modes;

[0023] Combined with the currently calculated collision probability of the interactive vehicle with the target vehicle in different risk modes, the probability density function constructed using the multivariate Gaussian distribution is used to correct the collision risk assessment results, and the following future risk assessment results are obtained:

[0024]

[0025] in, It represents the probability that the interactive vehicle hazard assessment value distribution is x when the known data result is a collision; Indicates the probability that the data result is a collision; It represents the probability that the interactive vehicle hazard assessment value distribution is x when the data results are known.

[0026] In a possible implementation, generating an autonomous driving decision according to the future risk assessment result specifically includes:

[0027] when When the target vehicle is determined to be at risk of collision, the emergency speed planning and braking strategy are activated;

[0028] when When the target vehicle is determined to be absolutely safe, the current driving strategy will not be changed;

[0029] when When , local game is used to deduce equilibrium decision and generate autonomous driving strategy;

[0030] in, is the safety threshold, is the danger threshold.

[0031] In one possible implementation, local game theory is used to deduce equilibrium decisions and generate autonomous driving strategies, including:

[0032] Determine the decision variables and benefit functions of the target vehicle and all interactive vehicles participating in the game in the current autonomous driving scenario based on the local game algorithm; wherein the benefit function includes an interactive loss weight and an interactive loss function;

[0033] Based on the decision variables and the benefit function, a Nash equilibrium strategy is adopted to obtain an equilibrium decision quantity of the target vehicle;

[0034] A longitudinal driving planning strategy of the target vehicle is generated based on the equilibrium decision amount.

[0035] In a second aspect, an automatic driving decision-making device is provided, comprising:

[0036] The acquisition module is used to obtain the vehicle information of the target vehicle in the autonomous driving scenario and the vehicle information of other interacting vehicles;

[0037] A calculation module, configured to calculate, based on the vehicle information of the target ego vehicle and the vehicle information of the other interactive vehicles, the conflict information between the target ego vehicle and each interactive vehicle with which there is a driving conflict, respectively; wherein the conflict information at least includes driving conflict zone representation information and driving conflict representation information between the target ego vehicle and the interactive vehicle; a determination module, configured to determine a collision risk assessment result of the interactive vehicle to the target ego vehicle according to the conflict information;

[0038] A correction module, used to correct the collision risk assessment result based on historical decision data, and determine the output of the future risk assessment result by the model constructed by multivariate Gaussian distribution;

[0039] A generation module is used to generate an autonomous driving decision according to the future risk assessment result, wherein the autonomous driving decision is used to assist the target vehicle in avoiding collision with other interactive vehicles during the autonomous driving process.

[0040] In a third aspect, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the method of the above-mentioned aspect and any possible implementation manner.

[0041] In a fourth aspect, an electronic device is provided, including:

[0042] at least one processor; and

[0043] a memory communicatively connected to the at least one processor; wherein,

[0044] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any possible implementation manner and the aspects described above.

[0045] According to a fifth aspect, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the above-mentioned aspects and any possible implementation method.

[0046] In a sixth aspect, an autonomous driving vehicle is provided, comprising the electronic device as described above.

[0047] The beneficial effects of the technical solution provided by this application include at least:

[0048] It can be seen from the above technical solution that based on the vehicle information of the target self-vehicle and the vehicle information of the other interactive vehicles, the conflict information between the target self-vehicle and each interactive vehicle with driving conflicts is calculated respectively; the collision risk assessment result of the interactive vehicle to the target self-vehicle is determined according to the conflict information; the collision risk assessment result is corrected based on the historical decision data, and the model constructed with multivariate Gaussian distribution is determined to output the future risk assessment result; and the autonomous driving decision is generated according to the future risk assessment result. Thus, a general and efficient interactive space topological algorithm is used to calculate the conflict information, which is applicable to a variety of spatial conflict scenarios; at the same time, through a two-dimensional full-plane risk assessment method, a local interactive game framework is constructed based on the interactive space, and the optimal joint strategy numerical solution is analyzed to automatically and reliably make self-vehicle driving decisions. In the case of a small amount of data adjustment, the processing efficiency is improved, the interaction process between vehicles is effectively considered, and the reliability of autonomous driving decisions is improved.

[0049] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of 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 creative work.

[0051] Figure 1 This is a schematic diagram of the steps of an autonomous driving decision-making method provided in an embodiment of the present application.

[0052] Figure 2 This is a schematic diagram of constructing a vehicle boundary convex hull using a Graham scanning algorithm provided in an embodiment of the present application.

[0053] Figure 3 This is a schematic diagram of representing the boundary convex hull as a rectangular frame provided in an embodiment of the present application.

[0054] Figure 4 It is a schematic diagram of the rotationally invariant primitive decomposition provided in an embodiment of the present application.

[0055] Figure 5 It is a schematic diagram of the line segment intersection calculation provided in an embodiment of the present application.

[0056] Figure 6 These are schematic diagrams of representing two conflict graphs provided in embodiments of the present application.

[0057] Figure 7 This is a conflict diagram considering the inner wheel difference provided in an embodiment of the present application.

[0058] Figure 8 It is a schematic diagram of a 2D autonomous driving operation conflict scenario provided in an embodiment of the present application.

[0059] Figure 9-1 and 9-2 Schematic diagram of two risk assessment results provided in the embodiments of the present application.

[0060] Fig.10 It is a schematic diagram of the autonomous driving decision process provided in an embodiment of the present application.

[0061] Fig.11 This is a structural block diagram of an automatic driving decision-making device provided in yet another embodiment of the present application.

[0062] Fig.12 This is a block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0063] The following is a description of exemplary embodiments of the present application in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0064] Obviously, the described embodiments are only part of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0065] It should be noted that the terminal devices involved in the embodiments of the present application may include but are not limited to mobile phones, personal digital assistants (PDAs), wireless handheld devices, tablet computers and other smart devices; display devices may include but are not limited to personal computers, televisions and other devices with display functions.

[0066] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0067] In the field of autonomous driving, driving decisions play a pivotal role. Behavior prediction methods based on machine learning effectively consider interactive features such as road environment / vehicles, but in the foreseeable future, when interactive data samples are difficult to collect effectively, the reliability of prediction results is difficult to guarantee. Model / rule-based intention prediction methods can well model the historical information of interactive vehicles, but there is no single intelligence in real traffic scenarios, and vehicle behavior will be adjusted in real time due to the behavior of other vehicles. Therefore, the effectiveness of the model will decrease as the time domain is extended.

[0068] To this end, the present application proposes a new autonomous driving decision-making method to circumvent the problem of poor reliability of autonomous driving decisions caused by the above-mentioned problems. The main inventive concept is: based on the vehicle information of the target self-vehicle and the vehicle information of the other interactive vehicles, respectively calculate the conflict information between the target self-vehicle and each interactive vehicle with driving conflicts; determine the collision risk assessment result of the interactive vehicle to the target self-vehicle according to the conflict information; correct the collision risk assessment result based on historical decision data, determine the output of the model constructed by multivariate Gaussian distribution to produce the future risk assessment result; generate autonomous driving decisions according to the future risk assessment result. Thus, a general and efficient interactive space topological algorithm is used to calculate conflict information, which is applicable to a variety of spatial conflict scenarios; at the same time, through a two-dimensional full-plane risk assessment method, a local interactive game framework is constructed based on the interactive space, and the optimal joint strategy numerical solution is analyzed to automatically and reliably make self-vehicle driving decisions. In the case of a small amount of data parameter adjustment, the processing efficiency is improved, the interaction process between vehicles is effectively considered, and the reliability of autonomous driving decisions is improved.

[0069] The present application scheme is described in detail below through specific embodiments.

[0070] Reference Figure 1 As shown, it is a schematic diagram of the steps of an autonomous driving decision-making method provided by an embodiment of the present application. The execution subject of the decision-making method may be an autonomous driving decision-making device. In the present application, the autonomous driving decision-making device may be a software module with computing and data processing capabilities, or an electronic device integrated with a similar software module. Optionally, the execution subject of the autonomous driving decision-making device in the present application may be an autonomous driving system server, an autonomous driving vehicle, or a related processing module integrated in an autonomous driving vehicle.

[0071] like Figure 1 As shown, the automatic driving decision-making method may include the following steps:

[0072] Step 102: Obtain vehicle information of the target vehicle in the autonomous driving scenario and vehicle information of other interacting vehicles.

[0073] In the present application scheme, the autonomous driving scene in which the target vehicle is located can be roughly determined first. Then, based on the determined autonomous driving scene, the vehicle information of the target vehicle and the vehicle information of other interacting vehicles are obtained. Among them, the vehicle information of the target vehicle may include vehicle shape, vehicle speed, vehicle posture, etc., as well as planned paths, etc. The vehicle interacting with the target vehicle may be one or more vehicles, therefore, the vehicle information of other interacting vehicles may be the vehicle shape, vehicle speed, vehicle posture, etc., as well as the predicted trajectory of one or more other interacting vehicles.

[0074] Step 104 calculates conflict information between the target self-vehicle and each interactive vehicle with which a driving conflict exists based on the vehicle information of the target self-vehicle and the vehicle information of the other interactive vehicles; wherein the conflict information at least includes driving conflict zone representation information and driving conflict representation information between the target self-vehicle and the interactive vehicle.

[0075] Optionally, the vehicle information of the target ego vehicle at least includes a planned path, and the vehicle information of the other interactive vehicles at least includes a predicted trajectory; then step 104, when respectively calculating the conflict information between the target ego vehicle and each interactive vehicle having a driving conflict based on the vehicle information of the target ego vehicle and the vehicle information of the other interactive vehicles, specifically includes the following steps:

[0076] In the first step, a polygonal box representing the shape of the vehicle is constructed for each of the target vehicle and other interacting vehicles, and the polygonal box of each vehicle is decomposed based on the rotation invariant primitive to construct a discrete conflict primitive.

[0077] Specifically, refer to Figure 2 As shown in Figure 1, the Graham scanning algorithm is used to construct the vehicle boundary convex hull. The boundary convex hull is represented as a rectangular box, as shown in Figure 1. Figure 3 As shown. In fact, the boundary convex hull can be represented by any polygon, here we take a rectangular box as an example. Figure 2 and Figure 3 The target vehicle and other interacting vehicles can be represented as rectangular boxes by the operation, thereby obtaining multiple rectangular boxes.

[0078] When a vehicle is represented by a rectangular box for collision detection, the direction of the vehicle must be clearly stated. However, if the vehicle shape is approximated to a rotationally invariant disk shape, it has rotational invariance and can greatly simplify the collision calculation process. Therefore, the rectangular box can be simply covered by n circles with a radius of r. The decomposition process is as follows: Figure 4 As shown, a rectangular frame with a length of l and a width of w is covered and decomposed by three circles with a radius of r, and the distance between the centers of adjacent circles is d.

[0079] The calculation formula is:

[0080]

[0081]

[0082] Among them, r is the radius of the decomposition circle, n is the number of decompositions, l is the length of the vehicle, w is the width of the vehicle, and d is the distance between the centers of adjacent decomposition circles.

[0083] In the second step, based on the discrete conflict primitives, the planned path of the target ego vehicle and the predicted trajectories of the other interactive vehicles, conflict information of each interactive vehicle that has a driving conflict with the target ego vehicle is calculated.

[0084] Each primitive is equally spaced along the planned path / predicted trajectory , emit the decomposed circle radius left and right along the normal direction of the curve to obtain the boundary points, connect all the boundary points to obtain the left and right driving boundaries, and represent them in the form of polyline, refer to Figure 5 The schematic diagram of line segment intersection calculation is shown in Figure 1. Specifically, the corner points of two line segments are solved based on the geometric method as follows:

[0085] Depend on ,have:

[0086]

[0087] From the parallelogram With parallelogram Same base, height and ,have:

[0088]

[0089] Right now

[0090]

[0091] Then, the conflict area is represented as follows:

[0092]

[0093] The conflict graph is represented as Figure 6 As shown, the left figure is a schematic diagram of the sampling of two vehicle trajectory points; the right figure is a schematic diagram of the trajectory sampling after the vehicle box (rectangular box) of each trajectory point is decomposed into discrete conflict primitives to obtain the boundary points of three circles based on the left figure. For special cases, considering the inner wheel difference of vehicle steering, the front and rear rotation invariant primitive paths should be different, and the inner wheel difference is calculated:

[0094]

[0095] At this time, the conflict is manifested as Figure 7 As shown, this is a schematic diagram of the conflict considering the inner wheel difference. Figure 7 The pink car in the middle represents the ego car, the green irregular polygon on the right represents the obstacle polygon given by the perception module, the right one of the two discrete curves represents the predicted driving trajectory of the obstacle car head, and the left one represents the inferred driving trajectory of the obstacle car tail. The middle area between the two curves is the potential conflict area, that is, the conflict area where scratches may occur.

[0096] Step 106: Determine a collision risk assessment result of the interactive vehicle to the target vehicle according to the conflict information.

[0097] Optionally, when determining the collision risk assessment result of the interactive vehicle to the target ego vehicle according to the conflict information, the collision probability of each interactive vehicle to the target ego vehicle in different risk modes may be calculated based on the conditional random field according to the following formula:

[0098]

[0099] Among them, the , D stands for danger, A stands for attention, and S stands for safety; It indicates the time when the target vehicle reaches the conflict area in normal driving; Indicates the time it takes for the interacting vehicles to reach the conflict area in normal driving; Indicates the time it takes for the interactive vehicle to slow down and drive to a safe area; represents the risk assessment value of the interactive vehicle j.

[0100] In addition, the collision risk assessment result of the interactive vehicle to the target vehicle may be determined based on the conflict information according to other existing risk assessment methods and formulas, which are not listed here one by one.

[0101] In specific implementation, a risk assessment method based on a Gaussian probability model can be used to perform multimodal risk assessment on 2D full-plane obstacles. Figure 8 As shown, 2D plane autonomous driving operation conflict scenarios can generally be divided into two categories: meeting / merging conflicts at intersections, and merging / diverging / cut-in conflicts on motor vehicle lanes. For these two types of scenarios, this application proposes a general solution architecture: modeling risk assessment as a safety indicator function.

[0102] In step 106, the time to zone (TTZ) of the autonomous driving vehicle to reach the conflict zone can be used to directly represent the interaction risk between vehicles in the 2D scene:

[0103]

[0104] in, is the boundary of the near-end conflict zone, is the speed at time t, and the conflict area is calculated by the general conflict operator. In addition, the time it takes for the target vehicle to decelerate to a safe state (TTS, Time to Stop) is used to indirectly represent the interaction risk between vehicles:

[0105]

[0106] in, is the speed of the autonomous vehicle, Deceleration is issued for planning.

[0107] Usually, the braking deceleration command is planned and issued, and the actual braking effect is affected by the road environment and vehicle execution. Therefore, it can be increased Coefficient (the empirical value here can be 1.1):

[0108]

[0109] The safety of the scene is inversely proportional to the amount of braking issued by the target vehicle. Therefore, three deceleration thresholds can be used to describe the braking requirements of the target vehicle. The deceleration thresholds are introduced and defined as: a. Vehicle emergency braking ; b. Vehicle moderate braking ; c. Vehicle gentle braking ; The deceleration threshold is described as follows:

[0110]

[0111] st

[0112] Therefore, there are:

[0113]

[0114]

[0115] Defining safe time : , according to the autonomous driving braking threshold, the interactive vehicle The interaction risk levels are divided into three categories: Dangerous (D), Attentive (A), Safe (S), and Interactive Vehicles. The risk assessment is defined as:

[0116]

[0117] Among them, when When the state between the target vehicle and the interactive vehicle should be defined as absolutely dangerous,

[0118]

[0119]

[0120] In other cases,

[0121]

[0122] Estimate the likelihood of different risk modes of interacting agents based on conditional random fields (CRF):

[0123]

[0124] in, The risk assessment results are shown as follows: Figure 9-1 and Figure 9-2 shown. Figure 9-1 and Figure 9-2 The pink car represents the ego car, the green car represents the obstacle car, and the red dot on the green car indicates that the car is within the risk monitoring range and is being monitored. Figure 9-1 The monitoring data of the obstacle vehicle are: the probability of danger is 0.598288, the probability of being between danger and safety and requiring attention is 0.409792, and the probability of safety is 0.000000. Figure 9-2 The monitoring data of the obstacle vehicle are: the probability of danger is 1.000000, the probability of being between danger and safety is 0.000000, and the probability of safety is 0.000000. Assuming that the threshold of the probability of danger is 0.900000, then, Figure 9-1 The danger probability of the obstacle car is significantly lower than the threshold, so it does not interfere with the ego car and the ego car does not slow down for a long time; Figure 9-2 The danger probability of the obstacle vehicle is greater than the threshold, so it will interfere with the ego vehicle and the ego vehicle will slow down to give way.

[0125] Step 108: The collision risk assessment result is modified based on the historical decision data, and a model constructed with multivariate Gaussian distribution is determined to output future risk assessment results.

[0126] Optionally, when the collision risk assessment result is corrected based on the historical decision data and the model constructed with the multivariate Gaussian distribution is determined to output the future risk assessment result, the historical decision data under the current driving scenario can be specifically determined, collision and non-collision information can be collected, and the characteristic quantity and collision result corresponding to each decision data in the historical decision data can be counted, wherein the characteristic quantity corresponding to each decision data is the collision probability of the historical interactive vehicle with the historical target vehicle in the decision data in different risk modes; combined with the currently calculated collision probability of the interactive vehicle with the target vehicle in different risk modes, the collision risk assessment result is corrected using the probability density function constructed with the multivariate Gaussian distribution to obtain the following future risk assessment result:

[0127]

[0128] in, It represents the probability that the interactive vehicle hazard assessment value distribution is x when the known data result is a collision; Indicates the probability that the data result is a collision; It represents the probability that the interactive vehicle hazard assessment value distribution is x when the data results are known.

[0129] It should be understood that in the present application, in addition to using the Bayesian method of multivariate Gaussian distribution to correct the collision risk assessment result of step 106, other existing correction methods can also be used, or relevant information in historical decision data can be directly integrated with the collision risk assessment result to obtain future risk assessment results.

[0130] The judgment of danger (collision in the future) needs to be carefully judged. If it is determined that there is a risk of collision in the short term in the future, the emergency speed planning will be immediately connected, so as to significantly slow down to ensure safety. However, when the danger identification is inaccurate, both safety and efficiency will be lost. For example, sudden braking leads to rear-end collision or excessive yielding. In addition, for different systems (different versions), due to the differences in design solutions, there is also the possibility that there are alternative solutions to ensure safety in the future if emergency braking is not done at present. Therefore, the historical operation data of the running version is introduced to collect collision / no collision information, and the possibility of collision in the future based on the current system and status is statistically analyzed.

[0131] Collect the past operation data of the current system, and count the characteristic quantities and collision results, such as:

[0132]

[0133]

[0134]

[0135] Construct a multivariate Gaussian distribution model with a mean of:

[0136]

[0137]

[0138]

[0139]

[0140] The covariance matrix is:

[0141]

[0142]

[0143]

[0144]

[0145]

[0146] The probability density function of the multivariate Gaussian distribution is:

[0147]

[0148] According to Bayes' theorem, the future risk assessment results can be obtained:

[0149] Step 110: Generate an autonomous driving decision based on the future risk assessment result, where the autonomous driving decision is used to assist the target vehicle in avoiding collision with other interactive vehicles during the autonomous driving process.

[0150] Based on the above future risk assessment results, when generating autonomous driving decisions:

[0151] when When the target vehicle is determined to be at risk of collision, the emergency speed planning and braking strategy are activated;

[0152] when When the target vehicle is determined to be absolutely safe, the current driving strategy will not be changed;

[0153] when When , the local game is used to deduce the equilibrium decision and generate the autonomous driving strategy; is the safety threshold, is the danger threshold.

[0154] Further, when When the local game is used to deduce the equilibrium decision, and when the autonomous driving strategy is generated, the decision variables and the profit function of the target ego vehicle and all interactive vehicles participating in the game in the current autonomous driving scenario are determined based on the local game algorithm; wherein the profit function includes the interaction loss weight and the interaction loss function; based on the decision variables and the profit function, the Nash equilibrium strategy is adopted to obtain the equilibrium decision quantity of the target ego vehicle; based on the equilibrium decision quantity, the longitudinal driving planning strategy of the target ego vehicle is generated.

[0155] In the present application, the cost is considered in two parts: efficiency and safety. Efficiency is defined by the time it takes for a game participant to pass through the boundary of the conflict area. It is foreseeable that this time can reflect the average speed of the vehicle through the entire conflict, so it is defined as above. Safety considers global safety, which is characterized by the time it takes for all traffic participants to reach the conflict area. It is also foreseeable that as long as the vehicles do not appear in the conflict area at the same time, they are safe and time-space non-conflicting. In order to solve the interactive game problem of N vehicles in the conflict scenario, the set of game participants is defined as , each participant (the target vehicle and each vehicle in the interaction vehicle) The decision variables are , then the decision set after all participants select is , at this time, the corresponding income of each participant is , assuming that the actions of the game participants are carried out simultaneously, during the game Decision variables for the car Defined as the longitudinal acceleration along the current path toward the conflict area, the benefit function is defined as follows:

[0156] in, and Respectively represent participants Time to reach the front / rear border of the conflict zone, represents the efficiency loss weight, represents the interaction loss weight, Represents a controlled agent The efficiency loss function is Indicates participants and The interaction loss function between , intuitively, if and , the larger the value, the greater the efficiency loss; if and Or vice versa, one party enters the conflict zone to complete the interaction and obtains the corresponding reward, which represents security, and the reward is proportional to this value; if and , there is a risk of collision in the interaction process, which will lead to huge collision losses.

[0157] Furthermore, the interaction loss weight The definition is as follows:

[0158]

[0159] Assume that there is a set of decision variables To make the global decision optimal, the participants The profit function is the optimal decision Taylor expansion, there are

[0160]

[0161] Game Nash Equilibrium Strategy It should be ensured that for any participant ,have

[0162]

[0163] Furthermore, for any participant , that is, satisfy

[0164]

[0165]

[0166] To solve this problem, all participants must meet the above conditions. Assuming there are N participants, i=0, 1, ..., N, construct the following simultaneous equations

[0167]

[0168] In order to quickly solve the problem, a numerical method is used, that is, sampling the set of participant actions [ u 0 , u 1 ,..., u N ] , substituting into the above equation constraints, we finally get the equilibrium decision X * = [ u 0 * , u 1 * ,..., u N * ] .

[0169] For the decision of the ego car, assuming that we always regard car 0 as the ego car in the above situation, we only need to obtain the equilibrium decision quantity of the ego car The subsequent longitudinal planning quantities s, v, a, and t of the vehicle are as follows:

[0170]

[0171]

[0172]

[0173]

[0174] Where T represents the planned horizon length.

[0175] Thus, the longitudinal planning factors, speed, acceleration, etc. of the ego vehicle can be obtained, so that the target ego vehicle can make driving decisions based on these planning factors. The general method mainly refers to the game method between two participants. This type of method does not consider the interaction process between other obstacles, resulting in distortion of the game process. The method and model proposed in this application effectively consider the cross-influence of other plane traffic participants while analyzing the game benefits of a certain participant, and restore the game process in the actual driving environment. Therefore, it can ensure that the determined driving decisions are more accurate and reliable, and ensure the safety of autonomous driving.

[0176] It should be noted that part or all of the execution subject of steps 102 to 110 may be an application located in the local terminal, or may also be a functional unit such as a plug-in or software development kit (SDK) set in the application located in the local terminal, or may also be a processing engine located in a network-side server, or may also be a distributed system located on the network side, for example, a processing engine or distributed system in an autonomous driving platform on the network side, etc. This embodiment does not specifically limit this.

[0177] It is understandable that the application may be a local program (nativeApp) installed on the local terminal, or may also be a webpage program (webApp) of a browser on the local terminal, which is not limited in this embodiment.

[0178] Reference Fig.10 The figure is a flow chart of the automatic driving decision-making method in the embodiment of the present application.

[0179] First, the planned path of the target ego vehicle and the predicted trajectory of the interactive vehicle are obtained. In fact, not only the planned path of the target ego vehicle is obtained, but also other vehicle information of the target ego vehicle, such as speed, posture, etc. Similarly, not only the predicted trajectory of the interactive vehicle is obtained, but also other vehicle information of the interactive vehicle, such as shape, speed, posture, etc.

[0180] Then, the vehicle geometric primitives are decomposed. According to the vehicle information of the target vehicle and the interactive vehicle, the decomposition method in the above step 104 is used for processing to obtain multiple conflict primitives. After that, multiple conflict areas are calculated based on the discrete conflict primitives. The threat level and collision risk of the interactive vehicle are calculated based on the conflict information to obtain a 2D risk map; and considering that the calculated collision risk is not highly adaptable, the system operation history data is combined to construct a multivariate Gaussian analysis system future risk. When evaluating the future risk of the system, for safe situations, the obstacle is ignored and no processing is required; for dangerous situations, emergency deceleration is planned; for other situations, local games are constructed, interactive benefits are analyzed, discrete control actions are taken, and local game problems are numerically solved; finally, decision control quantities are output according to the analysis results of each situation, such as brakes / accelerators, etc.

[0181] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0182] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0183] Fig.11 A structural block diagram of an automatic driving decision-making device provided by an embodiment of the present application is shown as follows: Fig.11The automatic driving decision-making device 1100 of this embodiment may include an acquisition module 1101 , a calculation module 1102 , a determination module 1103 , a correction module 1104 and a generation module 1105 . Among them, the acquisition module 1101 is used to obtain the vehicle information of the target self-vehicle in the automatic driving scene, and the vehicle information of other interactive vehicles; the calculation module 1102 is used to calculate the conflict information between the target self-vehicle and each interactive vehicle with driving conflict based on the vehicle information of the target self-vehicle and the vehicle information of the other interactive vehicles; wherein the conflict information at least includes the driving conflict zone representation information and the driving conflict representation information of the target self-vehicle and the interactive vehicle; the determination module 1103 is used to determine the collision risk assessment result of the interactive vehicle to the target self-vehicle according to the conflict information; the correction module 1104 is used to correct the collision risk assessment result based on the historical decision data, and determine the future risk assessment result output by the model constructed with the multivariate Gaussian distribution; the generation module 1105 is used to generate an automatic driving decision according to the future risk assessment result, and the automatic driving decision is used to assist the target self-vehicle to avoid collision with other interactive vehicles during the automatic driving process.

[0184] It should be noted that part or all of the autonomous driving decision-making device of this embodiment may be an application located in the local terminal, or may also be a functional unit such as a plug-in or software development kit (SDK) set in the application located in the local terminal, or may also be a processing engine located in a network-side server, or may also be a distributed system located on the network side, for example, a processing engine or distributed system in an autonomous driving platform on the network side, etc. This embodiment does not specifically limit this.

[0185] It is understandable that the application may be a local program (nativeApp) installed on the local terminal, or may also be a webpage program (webApp) of a browser on the local terminal, which is not limited in this embodiment.

[0186] Optionally, in a possible implementation of this embodiment, the vehicle information of the target vehicle at least includes a planned path, and the vehicle information of the other interactive vehicles at least includes a predicted trajectory;

[0187] The calculation module 1102, when calculating the conflict information between the target self-vehicle and each interactive vehicle having a driving conflict based on the vehicle information of the target self-vehicle and the vehicle information of the other interactive vehicles, is specifically used to: construct a polygonal box representing the shape of the vehicle for each of the target self-vehicle and the other interactive vehicles, and decompose the polygonal box of each vehicle based on the rotation invariant primitive to construct a discrete conflict primitive; based on the discrete conflict primitives, as well as the planned path of the target self-vehicle and the predicted trajectory of the other interactive vehicles, calculate the conflict information of each interactive vehicle having a driving conflict with the target self-vehicle.

[0188] Optionally, in a possible implementation of this embodiment, when the determination module 1103 calculates the collision risk assessment result of the interactive vehicle to the target ego vehicle according to the conflict information, it is specifically used to: calculate the collision probability of each interactive vehicle to the target ego vehicle in different risk modes based on the conditional random field according to the following formula:

[0189] Among them, the , D stands for danger, A stands for attention, and S stands for safety; It indicates the time when the target vehicle reaches the conflict area in normal driving; Indicates the time it takes for the interacting vehicles to reach the conflict area in normal driving; Indicates the time it takes for the interactive vehicle to slow down and drive to a safe area; represents the risk assessment value of the interactive vehicle j.

[0190] Optionally, in a possible implementation of this embodiment, when the correction module 1104 corrects the collision risk assessment result based on the historical decision data and determines that the model constructed with the multivariate Gaussian distribution outputs the future risk assessment result, it is specifically used to: determine the historical decision data under the current driving scenario, collect collision and non-collision information, and count the feature quantities and collision results corresponding to each decision data, wherein the feature quantity corresponding to each decision data is the collision probability of the historical interactive vehicle with the historical target vehicle in the decision data in different risk modes; in combination with the currently calculated collision probability of the interactive vehicle with the target vehicle in different risk modes, the probability density function constructed with the multivariate Gaussian distribution is used to correct the collision risk assessment result to obtain the following future risk assessment result:

[0191] in, It represents the probability that the interactive vehicle hazard assessment value distribution is x when the known data result is a collision; Indicates the probability that the data result is a collision; It represents the probability that the interactive vehicle hazard assessment value distribution is x when the data results are known.

[0192] Optionally, in a possible implementation of this embodiment, when the generation module 1105 generates an autonomous driving decision according to the future risk assessment result, it is specifically used to: When the target vehicle is determined to be at risk of collision, the emergency speed planning braking strategy is activated; when When the target vehicle is determined to be absolutely safe, the current driving strategy will not be changed; when When , the local game is used to deduce the equilibrium decision and generate the autonomous driving strategy; is the safety threshold, is the danger threshold.

[0193] Optionally, in a possible implementation of this embodiment, when the generation module 1105 adopts local game to deduce equilibrium decision and generate autonomous driving strategy, it is specifically used to: determine the decision variables and benefit functions of the target self-vehicle and all interactive vehicles participating in the game in the current autonomous driving scenario based on the local game algorithm; wherein the benefit function includes interaction loss weight and interaction loss function; based on the decision variables and benefit function, adopt the Nash equilibrium strategy to obtain the equilibrium decision quantity of the target self-vehicle; and generate the longitudinal driving planning strategy of the target self-vehicle based on the equilibrium decision quantity.

[0194] In this embodiment, based on the vehicle information of the target self-vehicle and the vehicle information of the other interactive vehicles, the conflict information between the target self-vehicle and each interactive vehicle with driving conflicts can be calculated respectively; the collision risk assessment result of the interactive vehicle to the target self-vehicle is determined according to the conflict information; the collision risk assessment result is corrected based on the historical decision data, and the model constructed with the multivariate Gaussian distribution is determined to output the future risk assessment result; and the autonomous driving decision is generated according to the future risk assessment result. Thus, a general and efficient interactive space topological algorithm is used to calculate the conflict information, which is applicable to a variety of spatial conflict scenarios; at the same time, through a two-dimensional full-plane risk assessment method, a local interactive game framework is constructed based on the interactive space, and the optimal joint strategy numerical solution is analyzed to automatically and reliably make the self-vehicle driving decision. In the case of a small amount of data parameter adjustment, the processing efficiency is improved, the interaction process between vehicles is effectively considered, and the reliability of autonomous driving decisions is improved.

[0195] An embodiment of the present application provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is loaded and executed by a processor to implement the autonomous driving decision-making method as described above.

[0196] An embodiment of the present application provides an electronic device, which includes a processor and a memory, wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the autonomous driving decision-making method as described above.

[0197] An embodiment of the present application provides a computer program product, including a computer program, which implements the autonomous driving decision-making method as described above when executed by a processor.

[0198] An embodiment of the present application provides an autonomous driving vehicle, including the electronic device as described above. Specifically, the autonomous driving vehicle can be a vehicle of level L2 or above.

[0199] In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the relevant laws and regulations and do not violate public order and good morals.

[0200] Fig.12 A schematic block diagram of an example electronic device 1200 that can be used to implement an embodiment of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0201] like Fig.12 As shown, the electronic device 1200 includes a computing unit 1201, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1202 or a computer program loaded from a storage unit 1208 to a random access memory (RAM) 1203. In the RAM 1203, various programs and data required for the operation of the electronic device 1200 can also be stored. The computing unit 1201, the ROM 1202, and the RAM 1203 are connected to each other via a bus 1204. An input / output (I / O) interface 1205 is also connected to the bus 1204.

[0202] Multiple components in the electronic device 1200 are connected to the I / O interface 1205, including: an input unit 1206, such as a keyboard, a mouse, etc.; an output unit 1207, such as various types of displays, speakers, etc.; a storage unit 1208, such as a disk, an optical disk, etc.; and a communication unit 1209, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1209 allows the electronic device 1200 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0203] The computing unit 1201 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 1201 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 1201 performs the various methods and processes described above, such as the autonomous driving decision method. For example, in some embodiments, the autonomous driving decision method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 1208. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 1200 via the ROM 1202 and / or the communication unit 1209. When the computer program is loaded into the RAM 1203 and executed by the computing unit 1201, one or more steps of the autonomous driving decision method described above may be performed. Alternatively, in other embodiments, the computing unit 1201 may be configured to perform the autonomous driving decision method in any other appropriate manner (e.g., by means of firmware).

[0204] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0205] The program code for implementing the method of the present application can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, implements the functions / operations specified in the flow chart and / or block diagram. The program code can be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0206] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0207] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0208] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0209] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0210] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps disclosed in this application can be performed in parallel, sequentially or in different orders, as long as the desired results of the technical solution disclosed in this application can be achieved, and this document does not limit this.

[0211] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. An automatic driving decision-making method, characterized in that: include: Obtain vehicle information of the target vehicle in the autonomous driving scenario, as well as vehicle information of other interacting vehicles; Based on the vehicle information of the target vehicle and the vehicle information of the other interactive vehicles, respectively calculate the conflict information between the target vehicle and each interactive vehicle with which the driving conflict exists; wherein the conflict information at least includes the driving conflict zone representation information and the driving conflict representation information between the target vehicle and the interactive vehicle; Determine a collision risk assessment result of the interactive vehicle to the target vehicle according to the collision information; The collision risk assessment result is corrected based on the historical decision data, and the future risk assessment result is determined to be output by the model constructed in a Bayesian manner using a multivariate Gaussian distribution, specifically including: determining the historical decision data under the current driving scenario, collecting collision and non-collision information, and counting the feature quantity and collision result corresponding to each decision data in the historical decision data, wherein the feature quantity corresponding to each decision data is the collision probability of the historical interactive vehicle against the historical target vehicle in the decision data in different risk modes; combining the currently calculated collision probability of the interactive vehicle against the target vehicle in different risk modes, the collision risk assessment result is corrected using a probability density function constructed using a multivariate Gaussian distribution, and the following future risk assessment result is obtained: in, It represents the probability that the interactive vehicle hazard assessment value distribution is x when the known data result is a collision; Indicates the probability that the data result is a collision; It represents the probability that the interactive vehicle hazard assessment value distribution is x when the data results are known; An autonomous driving decision is generated according to the future risk assessment result, and the autonomous driving decision is used to assist the target vehicle in avoiding collision with other interactive vehicles during the autonomous driving process.

2. The method according to claim 1, characterized in that The vehicle information of the target vehicle at least includes a planned path, and the vehicle information of other interactive vehicles at least includes a predicted trajectory; Based on the vehicle information of the target vehicle and the vehicle information of the other interactive vehicles, conflict information between the target vehicle and each interactive vehicle with driving conflict is calculated respectively, specifically including: For each of the target vehicle and other interacting vehicles, a polygonal box representing the shape of the vehicle is constructed respectively, and the polygonal box of each vehicle is decomposed based on the rotation invariant primitive to construct a discrete conflict primitive; Based on the discrete conflict primitives, the planned path of the target ego vehicle and the predicted trajectories of the other interactive vehicles, conflict information of each interactive vehicle that has a driving conflict with the target ego vehicle is calculated.

3. The method according to claim 1, characterized in that Determining a collision risk assessment result of the interactive vehicle to the target vehicle according to the conflict information specifically includes: According to the following formula, the collision probability of each interactive vehicle to the target vehicle in different risk modes is calculated based on the conditional random field: Among them, the , D stands for danger, A stands for attention, and S stands for safety; It indicates the time when the target vehicle reaches the conflict area in normal driving; Indicates the time it takes for the interacting vehicles to reach the conflict area in normal driving; Indicates the time it takes for the interactive vehicle to slow down and drive to a safe area; represents the risk assessment value of the interactive vehicle j.

4. The method according to claim 1, characterized in that Generating an autonomous driving decision based on the future risk assessment result, specifically including: when When the target vehicle is determined to be at risk of collision, the emergency speed planning and braking strategy are activated; when When the target vehicle is determined to be absolutely safe, the current driving strategy will not be changed; when When , local game is used to deduce equilibrium decision and generate autonomous driving strategy; in, is the safety threshold, is the danger threshold.

5. The method according to claim 4, characterized in that The local game is used to deduce equilibrium decisions and generate autonomous driving strategies, including: Determine the decision variables and benefit functions of the target vehicle and all interactive vehicles participating in the game in the current autonomous driving scenario based on the local game algorithm; wherein the benefit function includes an interactive loss weight and an interactive loss function; Based on the decision variables and the benefit function, a Nash equilibrium strategy is adopted to obtain an equilibrium decision quantity of the target vehicle; A longitudinal driving planning strategy of the target vehicle is generated based on the equilibrium decision amount.

6. An automatic driving decision-making device, characterized in that: include: The acquisition module is used to obtain the vehicle information of the target vehicle in the autonomous driving scenario and the vehicle information of other interacting vehicles; A calculation module, configured to calculate, based on the vehicle information of the target ego vehicle and the vehicle information of the other interactive vehicles, conflict information between the target ego vehicle and each interactive vehicle with which a driving conflict exists, respectively; wherein the conflict information at least includes driving conflict zone representation information and driving conflict representation information between the target ego vehicle and the interactive vehicle; A determination module, configured to determine a collision risk assessment result of the interactive vehicle to the target vehicle according to the collision information; The correction module is used to determine the historical decision data under the current driving scenario, collect collision and non-collision information, and count the feature quantity and collision result corresponding to each decision data in the historical decision data, wherein the feature quantity corresponding to each decision data is the collision probability of the historical interactive vehicle against the historical target vehicle in the decision data in different risk modes; combined with the currently calculated collision probability of the interactive vehicle against the target vehicle in different risk modes, the probability density function constructed by the multivariate Gaussian distribution is used to correct the collision risk assessment result, and the following future risk assessment result is obtained: in, It represents the probability that the interactive vehicle hazard assessment value distribution is x when the known data result is a collision; Indicates the probability that the data result is a collision; It represents the probability that the interactive vehicle hazard assessment value distribution is x when the data results are known; A generation module is used to generate an autonomous driving decision according to the future risk assessment result, wherein the autonomous driving decision is used to assist the target vehicle in avoiding collision with other interactive vehicles during the autonomous driving process.

7. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-5.

9. A computer program product, characterized in that The invention comprises a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 5.

10. An autonomous driving vehicle, characterized in that: Comprising the electronic device as claimed in claim 7.

Images