Automatic driving method and device, intelligent driving automobile and driving control system

Abstract

The invention discloses an automatic driving method and device, an intelligent driving automobile and a driving control system, and the method comprises the steps: obtaining the real-time driving data of the automobile, and collecting the real-time environment data around the automobile; wherein the real-time driving data comprises the position of the current road section where the vehicle is located; the real-time environment data comprises real-time motion behavior data of a target group in a motion state and traffic indication data; downloading a basic driving decision corresponding to the current road section from a cloud server; wherein the basic driving decision is obtained by learning historical driving decisions adopted by all vehicles through the current road section by the cloud server; predicting according to the real-time motion behavior data of the target group in the motion state to obtain an expected behavior of the target group in the motion state; according to the expected behavior of the target group in the motion state, the traffic indication data and the real-time driving data, predicting to obtain an expected driving behavior of the vehicle; and coupling the expected driving behavior of the vehicle by using the basic driving decision to obtain a final driving decision for controlling driving, and automatically driving based on the final driving decision.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving technology, and in particular to an automatic driving method, device, intelligent driving vehicle and driving control system. Background Art

[0002] With the development of intelligent and electrified cars, the automatic driving function has become a standard feature of cars. Automatic driving collects surrounding environment data through various sensors, and then uses complex algorithms to analyze and process it, and then automatically controls the car's power, steering, braking and other systems, enabling the car to make autonomous driving decisions, laying a solid foundation for the construction of the future intelligent transportation system.

[0003] However, traditional car autonomous driving mainly relies on sensor data inside the car and environmental perception of the actual vehicle conditions to make driving decisions, but it cannot predict expected changes in vehicle conditions. Therefore, it has limitations when dealing with complex and changeable traffic scenarios, resulting in low stability of autonomous driving and even traffic accidents. Summary of the invention

[0004] In order to solve or partially solve the technical problem that the autonomous driving technology of the prior art has limitations when facing complex and changeable traffic scenes, the present invention provides an autonomous driving method, device, intelligent driving car and driving control system, which predicts the expected driving behavior of the vehicle by referring to the actual traffic environment, and corrects it in combination with the basic driving decision in the cloud server. The final driving strategy obtained can well cope with the real-time and changeable complex traffic environment, can perform intelligent driving control of the car with high precision, and enhance the stability of autonomous driving.

[0005] In order to solve the above technical problems, the first aspect of the present invention discloses an automatic driving method, the method comprising:

[0006] Acquire the real-time driving data of the vehicle and collect the real-time environmental data around the vehicle; wherein the real-time driving data includes the location of the current road section where the vehicle is located; the real-time environmental data includes: the real-time motion behavior data of the target group in motion, and traffic indication data;

[0007] Downloading the basic driving decision corresponding to the current road section from the cloud server; wherein the basic driving decision is obtained by learning the historical driving decisions taken by all vehicles passing through the current road section by the cloud server;

[0008] Predicting the expected behavior of the target group in motion according to the real-time motion behavior data of the target group in motion;

[0009] Predicting the expected driving behavior of the vehicle according to the expected behavior of the target group in motion, the traffic instruction data and the real-time driving data;

[0010] The basic driving decision is used to couple the expected driving behavior of the vehicle to obtain a final driving decision for controlling driving, and the vehicle is driven automatically based on the final driving decision.

[0011] Optionally, the target group in motion is pedestrians or non-motorized vehicles, and the predicting of the expected behavior of the target group in motion according to the real-time motion behavior data of the target group in motion specifically includes:

[0012] Analyze the real-time motion behavior data of the pedestrian or non-motor vehicle to obtain the current behavior intention of the pedestrian or non-motor vehicle;

[0013] Find the historical behavior intention of the pedestrian or non-motor vehicle on the current road section;

[0014] The historical behavior intention is used to couple the current behavior intention to obtain the expected behavior of the pedestrian or non-motor vehicle and its execution probability.

[0015] Optionally, the target group in motion is cars located around the vehicle; and the step of predicting the expected behavior of the target group in motion based on the real-time motion behavior data of the target group in motion specifically includes:

[0016] Extracting key car navigation data from the real-time motion behavior data of the car; wherein the key car navigation data at least includes: navigation destination, travel purpose, lane where the car is located, and driver gender;

[0017] Predicting a number of vehicle driving behavior intentions based on the lane in which the vehicle is located;

[0018] The execution probability of the plurality of automobile driving behavior intentions is predicted with reference to the navigation destination, the travel purpose, and the driver's gender.

[0019] Optionally, the predicting of the expected driving behavior of the vehicle according to the expected behavior of the target group in motion, the traffic instruction data and the real-time driving data specifically includes:

[0020] The expected driving behavior of the vehicle is predicted based on the expected behavior of the pedestrian or non-motor vehicle and its execution probability, the plurality of vehicle driving behavior intentions and their execution probability, the traffic indication data and the real-time driving data.

[0021] Optionally, the predicting of the expected driving behavior of the vehicle according to the expected behavior of the pedestrian or non-motor vehicle and its execution probability, the execution probability of the plurality of vehicle driving behavior intentions, the traffic indication data and the real-time driving data specifically includes:

[0022] Extracting the lane where the vehicle is located from the real-time driving data, and predicting a number of driving behavior intentions of the vehicle according to the lane where the vehicle is located;

[0023] Determining a first influence coefficient on the driving of the vehicle according to the expected behavior of the pedestrian or non-motor vehicle and its execution probability;

[0024] Determining, according to the plurality of vehicle driving behavior intentions and their execution probabilities, respective second influence coefficients on the driving of the vehicle;

[0025] Conditionally coupling the vehicle driving data with reference to the traffic instruction data and / or the vehicle driving data to determine a third influence coefficient affecting the vehicle driving;

[0026] The target execution probability of each of the plurality of driving behavior intentions of the vehicle is determined according to the first influencing factor, the second influencing coefficient, and the third influencing coefficient.

[0027] Optionally, the utilizing the basic driving decision to couple the expected driving behavior of the vehicle to obtain a final driving decision for controlling driving specifically includes:

[0028] Extracting basic execution probabilities for each of the plurality of own-vehicle driving behavior intentions from the basic driving decision;

[0029] The basic execution probabilities and the target execution probabilities of the plurality of driving behavior intentions of the vehicle are integrated to obtain the final execution probabilities of the plurality of driving behavior intentions of the vehicle;

[0030] The behavior intention with the highest probability is selected from the final execution probabilities of the plurality of driving behavior intentions of the vehicle as the final driving decision.

[0031] Optionally, after the automatic driving based on the final driving decision, the method further includes:

[0032] Verifying the driving behavior corresponding to the final driving decision, and assigning a reward or penalty value of a reward or penalty mechanism to the final driving decision according to the verification result;

[0033] When the penalty value of the final driving decision reaches a set threshold, the final driving decision is uploaded to the cloud server for learning to update the basic driving decision.

[0034] A second aspect of the present invention discloses an automatic driving device, comprising:

[0035] An acquisition module is used to acquire real-time driving data of the vehicle and collect real-time environmental data around the vehicle; wherein the real-time driving data includes the location of the current road section where the vehicle is located; the real-time environmental data includes: real-time motion behavior data of the target group in motion, and traffic indication data;

[0036] A communication module, used for downloading the basic driving decision corresponding to the current road section from a cloud server; wherein the basic driving decision is obtained by learning the historical driving decisions taken by all vehicles passing through the current road section by the cloud server;

[0037] A first prediction module is used to predict the expected behavior of the target group in motion according to the real-time motion behavior data of the target group in motion;

[0038] A second prediction module is used to predict the expected driving behavior of the vehicle according to the expected behavior of the target group in motion, the traffic instruction data and the real-time driving data;

[0039] A coupling module is used to couple the expected driving behavior of the vehicle with the basic driving decision to obtain a final driving decision for controlling driving, and automatically drive based on the final driving decision.

[0040] Optionally, the target group in motion is pedestrians or non-motorized vehicles, and the first prediction module is specifically used to:

[0041] Analyze the real-time motion behavior data of the pedestrian or non-motor vehicle to obtain the current behavior intention of the pedestrian or non-motor vehicle;

[0042] Find the historical behavior intention of the pedestrian or non-motor vehicle on the current road section;

[0043] The historical behavior intention is used to couple the current behavior intention to obtain the expected behavior of the pedestrian or non-motor vehicle and its execution probability.

[0044] Optionally, the target group in motion is cars located around the vehicle; the first prediction module is specifically used to:

[0045] Extracting key car navigation data from the real-time motion behavior data of the car; wherein the key car navigation data at least includes: navigation destination, travel purpose, lane where the car is located, and driver gender;

[0046] Predicting a number of vehicle driving behavior intentions based on the lane in which the vehicle is located;

[0047] The execution probability of the plurality of automobile driving behavior intentions is predicted with reference to the navigation destination, the travel purpose, and the driver's gender.

[0048] Optionally, the second anticipation module is specifically used to predict the expected driving behavior of the vehicle based on the expected behavior of the pedestrian or non-motor vehicle and its execution probability, the several automobile driving behavior intentions and their execution probabilities, the traffic indication data and the real-time driving data.

[0049] Optionally, the second expected module is specifically used to:

[0050] Extracting the lane where the vehicle is located from the real-time driving data, and predicting a number of driving behavior intentions of the vehicle according to the lane where the vehicle is located;

[0051] Determining a first influence coefficient on the driving of the vehicle according to the expected behavior of the pedestrian or non-motor vehicle and its execution probability;

[0052] Determining a second influence coefficient of each of the plurality of vehicle driving behavior intentions and their execution probabilities on the driving of the vehicle;

[0053] Conditionally coupling the vehicle driving data with reference to the traffic instruction data and / or the vehicle driving data to determine a third influence coefficient affecting the vehicle driving;

[0054] The target execution probability of each of the plurality of driving behavior intentions of the vehicle is determined according to the first influencing factor, the second influencing coefficient, and the third influencing coefficient.

[0055] Optionally, the coupling module is specifically used for:

[0056] Extracting basic execution probabilities for each of the plurality of own-vehicle driving behavior intentions from the basic driving decision;

[0057] The basic execution probabilities and the target execution probabilities of the plurality of driving behavior intentions of the vehicle are integrated to obtain the final execution probabilities of the plurality of driving behavior intentions of the vehicle;

[0058] The behavior intention with the highest probability is selected from the final execution probabilities of the plurality of driving behavior intentions of the vehicle as the final driving decision.

[0059] Optionally, the device further comprises a feedback module, configured to:

[0060] Verifying the driving behavior corresponding to the final driving decision, and assigning a reward or penalty value of a reward or penalty mechanism to the final driving decision according to the verification result;

[0061] When the penalty value of the final driving decision reaches a set threshold, the final driving decision is uploaded to the cloud server for learning to update the basic driving decision.

[0062] According to a third aspect of the present invention, an intelligent driving car is disclosed, comprising a memory, an intelligent driving domain controller, and a computer program stored in the memory and executable on the intelligent driving domain controller, wherein the intelligent driving domain controller implements the steps of the method described in the first aspect when executing the program.

[0063] A fourth aspect of the present invention discloses a driving control system, which includes: a cloud server, such as the smart driving car described in the third aspect.

[0064] Through one or more technical solutions of the present invention, the present invention has the following beneficial effects or advantages:

[0065] The present invention provides an automatic driving method, device, intelligent driving car and driving control system, which predict the expected driving behaviors of surrounding pedestrians, non-motor vehicles and cars according to the surrounding real-time environmental data, and obtain the expected driving behaviors of the car that can cope with the actual traffic environment in combination with the real-time driving data of the car; in addition, by combining the intelligent driving domain controller and the cloud server, the basic driving decision is downloaded from the cloud server. Since the basic driving decision is obtained by learning the historical driving decisions taken by all vehicles on the current road section by the cloud server, the benchmark accuracy of the basic driving decision for the current road section can be guaranteed, and the basic driving decision is then used to couple the expected driving behavior of the car. The final driving strategy obtained can cope well with the real-time and changing complex traffic environment, can perform intelligent driving control of the car with high precision, and enhance the stability of automatic driving.

[0066] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0068] Figure 1 A schematic diagram of the structure of a driving control system according to an embodiment of the present invention is shown;

[0069] Figure 2A schematic diagram of a process flow of an automatic driving method according to an embodiment of the present invention is shown;

[0070] Figure 3 A flowchart for predicting the expected driving behavior of the vehicle according to one embodiment of the present invention is shown;

[0071] Figure 4 A schematic diagram of the interaction logic between the vehicle and the cloud server according to an embodiment of the present invention is shown;

[0072] Figure 5 A schematic diagram of an intelligent driving domain controller according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0073] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0074] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0075] In order to solve the technical problem that the autonomous driving in the prior art has limitations when facing complex and changeable traffic scenarios, the present invention provides an autonomous driving method, device, intelligent driving car and driving control system.

[0076] First, in order to facilitate the description and understanding of the present invention, the driving control system of the present invention is first introduced below. Figure 1 As shown, the driving control system includes: a cloud server and an intelligent driving car.

[0077] The cloud server is loaded with a large cloud AI (Artificial Intelligence) model.

[0078] Smart driving cars include: collector, TBOX (Telematics Box, vehicle communication terminal), and smart driving domain controller. Smart driving cars refer to all cars with autonomous driving functions, and there is no limit on the number.

[0079] Intelligent driving cars use collectors (such as sensors, cameras, radars, etc.) to collect road and vehicle condition information at a certain intersection.

[0080] Intelligent driving cars communicate with cloud servers through TBOX. Specifically, intelligent driving cars upload the road and vehicle condition information of a certain intersection and the driving decisions that have been made at the intersection to the cloud server through TBOX for training and learning of the cloud AI big model to obtain basic driving decisions about the intersection.

[0081] The cloud server sends basic driving decisions to the intelligent car through TBOX to correct the intelligent driving decisions of the intelligent car. If the intelligent car faces a new road, it will first drive according to the basic driving decisions sent by the cloud server and make corrections based on the actual situation.

[0082] Inside the smart car, a local AI big model is installed to predict and analyze the road and vehicle condition information collected at the intersection to infer the expected behavior of pedestrians, non-motor vehicles, and cars around the car, and further predict the intelligent driving behavior of the car. In addition, the collected road and vehicle condition information, the expected behavior of pedestrians, non-motor vehicles, and cars around the car, and the intelligent driving behavior of the car are transmitted to the cloud server, and trained and learned by the cloud AI big model in the cloud server. The cloud AI big model contains multiple behavior prediction models, such as pedestrian behavior prediction models, car behavior prediction models, etc., which can learn more accurate and universal basic driving decisions, improve the accuracy of traffic prediction for the intersection, and optimize the intelligent driving control of each road section. The control logic of the car's intelligent driving refers to the automatic driving method disclosed in the second aspect below, which will not be repeated here.

[0083] In the second aspect, an embodiment of the present invention provides an autonomous driving method, which is mainly applied to an intelligent driving domain controller.

[0084] See also Figure 2 , the method comprises at least the following steps:

[0085] S201, obtaining real-time driving data of the vehicle, and collecting real-time environmental data around the vehicle.

[0086] The real-time driving data of the vehicle includes but is not limited to: the current speed of the vehicle, the current time, the location of the current road section of the vehicle, the lane of the vehicle, the destination of the vehicle, the working hours, the type of the destination of the vehicle (such as the company, home, shopping mall, hospital, etc.), and the remaining mileage of the destination of the vehicle. The real-time driving data of the vehicle is collected by various sensors installed on the vehicle.

[0087] The real-time environmental data includes: real-time motion behavior data of the target group in motion, and traffic indication data; the target group in motion is: pedestrians, or non-motorized vehicles, and of course, it can also be cars located around the vehicle.

[0088] Taking pedestrians as an example, the real-time motion behavior data of pedestrians includes: pedestrian static data (appearance, gender, clothing, company location, distance between pedestrians and the company, etc.) and pedestrian dynamic data (walking posture, walking speed, walking direction, etc.).

[0089] Taking non-motorized vehicles as an example, the real-time motion behavior data of non-motorized vehicles includes: static data of non-motorized vehicles (appearance, gender, clothing, license plate number, company location, distance from the company, etc.) and dynamic data of non-motorized vehicles (driving speed, driving direction, etc.).

[0090] Of course, the real-time motion behavior data of all pedestrians around the vehicle (especially in front of the vehicle) and the real-time motion behavior data of non-motor vehicles need to be collected. Among them, the real-time motion behavior data of the target group in motion is collected by the camera, radar, communication terminal and other collection equipment installed on the vehicle. For example, the vehicle uses the camera to collect data such as the appearance, gender, and clothing of pedestrians, and searches for hidden data such as the location of the company corresponding to the pedestrian, the distance between the pedestrian and the company, etc. from the local database or upload network. For another example, the vehicle uses radar to collect and analyze the dynamic data (driving speed, direction, etc.) of non-motor vehicles.

[0091] The real-time motion behavior data of the car at least includes the key data of car navigation and the car condition data. Among them, the key data of car navigation at least include: navigation destination, travel purpose (such as company, home, shopping mall, hospital, etc.), car use (work, medical treatment, meeting, online car-hailing, travel, home, etc.), car driver gender, car lane. The car condition data at least includes: the current speed of the car. Of course, the real-time motion behavior driving data of all cars around this car need to be collected. Among them, the real-time motion behavior data of the car is collected by collecting equipment such as cameras, radars, and communication terminals installed on this car. For example, the current destination type, purpose, etc. of the car are remotely obtained using the on-board communication terminal.

[0092] The traffic indication data at least includes: the current road speed limit, the distance from the vehicle to the traffic light intersection, and the switching time of the traffic light intersection. The traffic indication data is collected by collecting equipment such as cameras, radars, and communication terminals installed on the vehicle.

[0093] S202, downloading the basic driving decision corresponding to the current road section from the cloud server.

[0094] Among them, the basic driving decision is learned by the cloud server based on the historical driving decisions taken by the vehicle or all vehicles including the vehicle when traveling on the current road section.

[0095] Specifically, the cloud server collects the historical environmental information faced by the vehicle when passing through the current section of road and the historical driving decisions taken for learning, and obtains the basic driving decision for the current section of road. For example, the current section of road is the section that the vehicle must pass through to go to work, and the working hours and routes are relatively fixed. The cloud server learns based on the historical environmental information and historical driving decisions around the vehicle, and the basic driving decisions obtained are more in line with the driving behavior habits of the vehicle. In order to make the basic driving decisions more applicable, the cloud server learns based on the historical environmental information faced by all vehicles (including this vehicle) and the historical driving decisions taken, and the basic driving decisions obtained are more adaptable.

[0096] For this vehicle, the location of the current road section of this vehicle is uploaded to the cloud server through TBOX, and the corresponding basic driving decision is downloaded from the cloud server. Of course, the lane where this vehicle is located can also be merged and uploaded, so that the downloaded basic driving decision can accurately guide the intelligent driving of the lane where this vehicle is located. Among them, the basic driving decision includes the basic execution probabilities of several driving behavior intentions for the lane where this vehicle is located. Taking the 3-lane as an example, assuming that the lane where this vehicle is located is the middle lane, then its driving behavior intentions are: straight at a constant speed, straight acceleration, straight deceleration, change lanes to the left, and change lanes to the right. The basic driving decision contains the corresponding basic execution probability. For example: straight at a constant speed of 0.5, straight acceleration of 0.2, straight deceleration of 0.2, change lanes to the left of 0.1, and change lanes to the right of 0.1. The basic driving decision changes according to the change of position.

[0097] S203, predicting the expected behavior of the target group in motion according to the real-time motion behavior data of the target group in motion.

[0098] In the process of predicting the expected behavior of target groups in motion, different target groups are analyzed with different strategies.

[0099] If the target group in motion is a pedestrian or a non-motor vehicle, the real-time motion behavior data of the pedestrian or non-motor vehicle is analyzed to obtain the current behavior intention of the pedestrian or non-motor vehicle; specifically, the real-time motion behavior data of the target group in motion includes static data and dynamic data. The current behavior intention is obtained by combining the static data and dynamic data analysis. Taking pedestrian A as an example, the current behavior intention and its execution probability are obtained based on the pedestrian's appearance, gender, clothing, company location, distance from the pedestrian to the company, walking posture, walking speed, and walking direction. In this embodiment, the built-in local AI large model of the vehicle can be used for behavior prediction, or the mapping relationship obtained by long-term learning can be used for behavior prediction. For example, the current behavior intention of pedestrian A is to cross the road ahead, and its execution probability varies according to the purpose of pedestrian A. Assuming that the purpose of crossing the road ahead is work, its execution probability is A1; assuming that the purpose of crossing the road ahead is work, its execution probability is A2; assuming that the purpose of crossing the road ahead is work, its execution probability is A3; first appearance (unknown): execution probability A4.

[0100] Furthermore, the historical behavior intentions of pedestrians or non-motor vehicles on the current road section are searched. For example, the historical behavior intention of pedestrian A on the current road section is to run a red light, and the execution probability is A5. The current behavior intention is coupled with the historical behavior intention to obtain the expected behavior of pedestrians or non-motor vehicles and their execution probability. In the coupling process, the average of the historical behavior intention and the current behavior intention is calculated to obtain the expected behavior of the target group in motion and its execution probability. For example, the current behavior intention of the pedestrian is to cross the road ahead, and its execution probability is A3, then the execution probability of the pedestrian's expected behavior is: A = (A1 + A3) / 2.

[0101] See Table 1 below, which shows the expected behaviors of the target groups (pedestrians, non-motorized vehicles) in motion and their execution probabilities.

[0102] Table 1

[0103] Expected behavior Execution probability Pedestrian A Crossing the road ahead 65％ Pedestrian B Crossing the road ahead 55％ Non-motorized vehicle C Crossing the road ahead 55％ Non-motorized vehicle D Crossing the road ahead 50％

[0104] In Table 1, the expected behaviors of pedestrian A, pedestrian B, non-motorized vehicle C, and non-motorized vehicle D will all affect the driving of the vehicle. If the pedestrian goes straight along the road and does not affect the driving of the vehicle, it can be ignored. Further, it is determined whether the execution probability of the expected behavior of the pedestrian or non-motorized vehicle is greater than the threshold. If it is greater than the threshold, such as greater than 50%, it will affect the driving of the vehicle and be included in the calculation of the influence coefficient. If it is less than the threshold, it will be ignored.

[0105] If the target group in motion is the cars located around the vehicle, the key data of car navigation is extracted from the real-time motion behavior data of the car. Among them, the key data of car navigation includes at least: navigation destination, travel purpose, lane where the car is located, and gender of the driver. According to the lane where the car is located, predict several car driving behavior intentions. In this embodiment, the built-in local AI large model of the vehicle can be used for behavior prediction, or the behavior prediction can be performed based on the mapping relationship obtained by long-term learning. Among them, several car driving behavior intentions specifically include: straight acceleration, straight deceleration, straight uniform speed, left lane change, right lane change. With reference to the navigation destination, travel purpose, and driver's gender, predict the execution probability of several car driving behavior intentions. For example, the probability of straight acceleration is E1, the probability of straight deceleration is E2, the probability of straight uniform speed is E3, the probability of left lane change is E4, and the probability of right lane change is E5. Among them, all cars around the vehicle refer to the above execution logic, such as the car in front of the vehicle, the vehicle behind the vehicle, the car in front of the left of the vehicle, the car in front of the right of the vehicle, the car behind the left of the vehicle, and the car behind the right of the vehicle, and obtain their respective expected behaviors and their execution probabilities.

[0106] S204, predicting the expected driving behavior of the vehicle according to the expected behavior of the target group in motion, traffic instruction data and real-time driving data.

[0107] Among them, the expected driving behavior of the vehicle is predicted based on the expected behavior of pedestrians or non-motor vehicles and their execution probability, several car driving behavior intentions and their execution probability, traffic instruction data and real-time driving data.

[0108] See also Figure 3 , is a flowchart for predicting the expected driving behavior of the vehicle, specifically performing the following steps:

[0109] S2041, extracting the lane where the vehicle is located from the real-time driving data, and predicting several driving behavior intentions of the vehicle based on the lane where the vehicle is located.

[0110] Specifically, the number and type of driving behavior intentions corresponding to the blocked lanes are different. For example, if the vehicle is in the middle lane, its driving behavior intentions include: going straight at a constant speed, going straight and accelerating, going straight and decelerating, changing lanes to the left, and changing lanes to the right.

[0111] S2042: Determine a first influence coefficient on the driving of the vehicle according to the expected behavior of the pedestrian or non-motor vehicle and its execution probability.

[0112] Among them, each driving behavior intention of the vehicle has its own first influence coefficient.

[0113] Specifically, in the process of determining the first influence coefficient, the first influence relationship of the expected behavior of pedestrians or non-motor vehicles on the driving behavior intention of each vehicle is determined; wherein the first influence relationship is a positive influence relationship or a negative influence relationship. The expected behavior of pedestrians or non-motor vehicles has different first influence relationships on different driving behavior intentions of the vehicle. According to the execution probability of the expected behavior of pedestrians or non-motor vehicles, the first influence degree value for each driving behavior intention of the vehicle is determined. Among them, the execution probability of the expected behavior of pedestrians or non-motor vehicles and the first influence degree value of the driving behavior intention of the vehicle have a mapping relationship, and the first influence degree value can be obtained by searching according to the mapping relationship. According to the first influence relationship and the first influence degree value of each driving behavior intention of the vehicle, the first influence coefficient of each driving behavior intention of the vehicle is obtained.

[0114] Taking pedestrian A's expected behavior: crossing the road ahead as an example, if the driving intention of the vehicle is to accelerate straight ahead, then pedestrian A's expected behavior will have a negative impact on the acceleration of the vehicle, and its first impact coefficient on the driving of the vehicle is negative, and the degree of impact is reflected in the specific value of the first impact coefficient. If the driving intention of the vehicle is to decelerate straight ahead, then pedestrian A's expected behavior has a positive impact on the deceleration of the vehicle, and its first impact coefficient on the driving of the vehicle is positive. If the driving intention of the vehicle is to change lanes to the left, then pedestrian A's expected behavior will not affect the left lane change of the vehicle, and its first impact coefficient on the driving of the vehicle is 0.

[0115] Refer to Table 2 below, which shows the first influence coefficient of the target group in motion (taking pedestrians and non-motorized vehicles as examples) on the driving of the vehicle.

[0116] Table 2

[0117] The vehicle is accelerating straight ahead The vehicle is moving straight at a constant speed The vehicle is driving straight and slowing down The vehicle is changing lane left The vehicle is changing lane right Pedestrian A -0.2 0.1 0.1 0 0 Pedestrian B -0.1 0.1 0.1 0 0 Non-motorized vehicle C -0.3 0.1 0.1 0 0 Non-motorized vehicle D -0.2 0.1 0.1 0 0

[0118] S2043: Determine a second influence coefficient of each of the plurality of vehicle driving behavior intentions and their execution probabilities on the driving of the vehicle.

[0119] Among them, each driving behavior intention of the vehicle has its own second influence coefficient.

[0120] The second influence coefficient is determined by the location of the vehicle, a number of driving behavior intentions corresponding to the vehicle and their execution probabilities, and a number of driving behavior intentions of the vehicle itself.

[0121] Specifically, in the process of determining the second influence coefficient, the second influence relationship of each car driving behavior intention on each car driving behavior intention is determined according to the position of the car relative to the car and each car driving behavior intention; wherein, cars in different positions and different car driving behavior intentions will have different second influence relationships on each car driving behavior intention, and the second influence relationship includes a positive influence relationship and a negative influence relationship. Different car driving behavior intentions have different second influence relationships on different car driving behavior intentions. According to the execution probability of each car driving behavior intention, the second influence degree value of each car driving behavior intention is determined. Among them, the execution probability of each car driving behavior intention and the second influence degree value of the car driving behavior intention have a mapping relationship, and the second influence degree value can be obtained by searching according to the mapping relationship. According to the second influence relationship and the second influence degree value of each car driving behavior intention, the second influence coefficient of each car driving behavior intention is obtained.

[0122] Taking the car in front of the vehicle as an example, if the driving behavior intention of the vehicle is to accelerate straight ahead, and the driving behavior intention of the car in front to accelerate straight ahead, change lanes left, and change lanes right has a positive impact on the acceleration of the vehicle in straight ahead, then its second impact coefficient on the driving of the vehicle is positive, and the degree of impact is reflected in the specific value of the second impact coefficient. The driving behavior intention of the car in front to decelerate straight ahead has a negative impact on the acceleration of the vehicle in straight ahead, then its second impact coefficient on the driving of the vehicle is negative, and the degree of impact is reflected in the specific value of the second impact coefficient. The driving behavior intention of the car in front to drive straight ahead at a constant speed has no impact on the acceleration of the vehicle in straight ahead, then its second impact coefficient on the driving of the vehicle is 0. For another example, if the driving behavior intention of the vehicle is to change lanes to the left, and the driving behavior intention of the car in front to accelerate straight ahead and change lanes right has a positive impact on the left lane change of the vehicle, then its second impact coefficient on the driving of the vehicle is positive, and the degree of impact is reflected in the specific value of the second impact coefficient. If the driving behavior intention of the car ahead to slow down and change lanes left has a negative impact on the left lane change of the vehicle ahead, then its second influence coefficient on the driving of the vehicle ahead is negative, and the degree of influence is reflected in the specific value of the second influence coefficient. If the driving behavior intention of the car ahead to go straight at a constant speed has no impact on the left lane change of the vehicle ahead, then its second influence coefficient on the driving of the vehicle ahead is 0.

[0123] See Table 3 below for specific examples of the values ​​of the second influence coefficients of different vehicle driving behavior intentions on the vehicle driving.

[0124] Table 3

[0125]

[0126]

[0127] S2044, it is worth noting that the size of the second influence coefficient is also affected by the size of the execution probability corresponding to the vehicle driving behavior intention.

[0128] Conditionally coupling the vehicle driving data with reference to the traffic indication data and / or the vehicle driving data to determine a third influence coefficient affecting the vehicle driving;

[0129] Traffic indication data Traffic indication data at least includes: current road speed limit, distance from the vehicle to the traffic light intersection, and switching time of the traffic light intersection. The vehicle driving data includes but is not limited to: the current speed of the vehicle, the current time, the lane the vehicle is in, the destination of the vehicle, the type of the destination of the vehicle (such as the company, home, shopping mall, hospital, etc.), working hours, and the remaining mileage of the destination of the vehicle. The car driving data includes but is not limited to: the current speed of the car.

[0130] Based on the aforementioned public data, at least the following conditions are coupled:

[0131] (1) Conditional coupling is performed based on the vehicle's minimum speed limit and the vehicle's current speed, and the third influence coefficient is determined based on the coupling result. The vehicle's minimum speed limit is obtained by dividing the remaining mileage to the vehicle's destination by the time difference between the current time and the working time.

[0132] (2) Conditional coupling is performed based on the minimum speed limit of the vehicle and the current road speed limit, and the third influence coefficient is determined based on the coupling result.

[0133] (3) Conditional coupling is performed based on the time it takes for the vehicle to reach the traffic light intersection and the time it takes to switch between the traffic light intersections, and the third influence coefficient is determined based on the coupling result. The time it takes for the vehicle to reach the traffic light intersection is obtained by the ratio of the distance from the vehicle to the traffic light intersection to the current speed of the vehicle.

[0134] (4) The driving data of the vehicle is coupled according to whether there are cars around the vehicle, and the fourth influence coefficient is determined according to the coupling result.

[0135] (5) Conditionally couple the current speed of the vehicle according to the current speed of the vehicle, and / or conditionally couple the current speed difference between the vehicle and surrounding vehicles and the first ratio of the switching time at the traffic light intersection with the second ratio of the current speed of the vehicle and the switching time at the traffic light intersection, and determine the fourth influence coefficient by combining the two coupling results.

[0136] Different driving behavior intentions of the vehicle have different feedbacks on the coupling result, which is specifically reflected in the third influence coefficient.

[0137] S2045, determining the target execution probability of each of a plurality of driving behavior intentions of the vehicle according to the first influencing factor, the second influencing coefficient, and the third influencing coefficient.

[0138] Specifically, for each driving behavior intention of the vehicle, the first influencing factor, the second influencing coefficient, and the third influencing coefficient are added together to obtain the corresponding target execution probability.

[0139] It is worth noting that since the driving data of the vehicle and the real-time environmental data around the vehicle are being processed and changed at any time, the target execution probability of each driving behavior intention of the vehicle can be calculated once every set time (for example, 10ms).

[0140] S205, using the basic driving decision to couple the expected driving behavior of the vehicle to obtain a final driving decision for controlling driving, and automatically driving based on the final driving decision.

[0141] In the coupling process, the basic execution probabilities of several driving behavior intentions of the vehicle are extracted from the basic driving decision; the basic execution probabilities of several driving behavior intentions of the vehicle are fused with their target execution probabilities to obtain the final execution probabilities of several driving behavior intentions of the vehicle. Among them, the basic execution probability and the target execution probability of each behavior intention can be fused by adding them to obtain the final execution probability of each driving behavior intention of the vehicle. The behavior intention with the highest probability is selected from the final execution probabilities of several driving behavior intentions of the vehicle as the final driving decision.

[0142] It is worth noting that since the target execution probability is calculated once every set time (for example, 10ms), the basic execution probability and the target execution probability of several driving behavior intentions of the vehicle are fused every set time (for example, 10ms) to obtain the final execution probability of several driving behavior intentions of the vehicle. Furthermore, the behavior intention that meets the set number of times and the set probability threshold is selected from the final execution probability as the final driving decision. For example, the final execution probability is calculated once every 10ms. If the final execution probability of straight acceleration reaches more than 0.5 for 100 consecutive times, straight acceleration is used as the final execution decision to control the driving of the vehicle.

[0143] After automatic driving based on the final driving decision, the driving behavior corresponding to the final driving decision is verified, and the final driving decision is assigned a penalty value of the reward mechanism based on the verification result; when the penalty value of the final driving decision reaches the set threshold, the final driving decision is uploaded to the cloud server for learning to update the basic driving decision. For example, the driving behavior corresponding to the final driving decision is verified based on the actual driving results. If the driving is completed successfully, a positive reward is given, and the reward value is +1. If the driving is not completed successfully, such as being stopped by a vehicle or pedestrian, the reward value is -1. And the corresponding impact coefficient is adaptively changed according to the cause of failure. When the number of rewards reaches 50, the judgment is recognized as a valid judgment, and the final driving decision is uploaded to the cloud server for learning by the cloud AI large model, so as to further improve the intelligent control of the road section.

[0144] See also Figure 4 , is a logical diagram of the interaction between the vehicle and the cloud server. In the vehicle, the expected driving behaviors of pedestrians, non-motorized vehicles, and cars are predicted based on the surrounding real-time environmental data, and the expected driving behaviors of the vehicle that can cope with the actual traffic environment are obtained by combining the real-time driving data of the vehicle; in addition, the basic driving decision is downloaded from the cloud server through the combination of the intelligent driving domain controller and the cloud server. Since the basic driving decision is obtained by learning the historical driving decisions taken by all vehicles on the current road section by the cloud server, the benchmark accuracy of the basic driving decision for the current road section can be guaranteed. Then, the basic driving decision is used to couple the expected driving behavior of the vehicle. The final driving strategy can cope well with the real-time and changing complex traffic environment, and can perform intelligent driving control of the vehicle with high precision, which enhances the stability of automatic driving.

[0145] Through the technical solution of the present invention, massive amounts of data and computing power are delivered to the cloud server, which not only reduces the intelligent driving cost of a single vehicle, but also greatly improves the learning effect and iteration ability of the intelligent driving large model because the learning data of the cloud server is relatively concentrated, thereby enhancing the intelligent control of each road section.

[0146] In a third aspect, based on the same inventive concept as the autonomous driving method provided in the first aspect embodiment, an embodiment of the present invention further provides an autonomous driving device, which is loaded in a smart driving domain controller chip, see Figure 5 ,include:

[0147] The acquisition module 501 is used to acquire the real-time driving data of the vehicle and collect the real-time environmental data around the vehicle; wherein the real-time driving data includes the location of the current road section where the vehicle is located; the real-time environmental data includes: the real-time motion behavior data of the target group in motion, and the traffic indication data;

[0148] The communication module 502 is used to download the basic driving decision corresponding to the current road section from the cloud server; wherein the basic driving decision is obtained by the cloud server learning the historical driving decisions taken by all vehicles passing through the current road section;

[0149] A first prediction module 503 is used to predict the expected behavior of the target group in motion according to the real-time motion behavior data of the target group in motion;

[0150] A second prediction module 504 is used to predict the expected driving behavior of the vehicle according to the expected behavior of the target group in motion, the traffic indication data and the real-time driving data;

[0151] The coupling module 505 is used to couple the expected driving behavior of the vehicle with the basic driving decision to obtain a final driving decision for controlling driving, and automatically drive based on the final driving decision.

[0152] It should be noted that the intelligent driving domain controller provided in the embodiment of the present invention, wherein the specific manner in which each module performs operations has been described in detail in the method embodiment provided in the above-mentioned first aspect. The specific implementation process can refer to the method embodiment provided in the above-mentioned first aspect, and will not be elaborated here.

[0153] In the fourth aspect, based on the same inventive concept as the method provided in the embodiment of the first aspect mentioned above, an embodiment of the present invention also discloses an intelligent driving car, comprising a memory, an intelligent driving domain controller, and a computer program stored in the memory and executable on the intelligent driving domain controller, characterized in that the intelligent driving domain controller implements the steps of any of the methods described above when executing the program.

[0154] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0155] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. An automatic driving method, characterized in that: The method comprises: Acquire the real-time driving data of the vehicle and collect the real-time environmental data around the vehicle; wherein the real-time driving data includes the location of the current road section where the vehicle is located; the real-time environmental data includes: the real-time motion behavior data of the target group in motion, and traffic indication data; Downloading the basic driving decision corresponding to the current road section from the cloud server; wherein the basic driving decision is obtained by learning the historical driving decisions taken by all vehicles passing through the current road section by the cloud server; Predicting the expected behavior of the target group in motion according to the real-time motion behavior data of the target group in motion; Predicting the expected driving behavior of the vehicle according to the expected behavior of the target group in motion, the traffic instruction data and the real-time driving data; The basic driving decision is used to couple the expected driving behavior of the vehicle to obtain a final driving decision for controlling driving, and the vehicle is driven automatically based on the final driving decision.

2. The method according to claim 1, characterized in that The target group in motion is pedestrians or non-motorized vehicles, and the predicting of the expected behavior of the target group in motion according to the real-time motion behavior data of the target group in motion specifically includes: Analyze the real-time motion behavior data of the pedestrian or non-motor vehicle to obtain the current behavior intention of the pedestrian or non-motor vehicle; Find the historical behavior intention of the pedestrian or non-motor vehicle on the current road section; The historical behavior intention is used to couple the current behavior intention to obtain the expected behavior of the pedestrian or non-motor vehicle and its execution probability.

3. The method according to claim 2, characterized in that The target group in motion is the cars located around the vehicle; the predicted expected behavior of the target group in motion based on the real-time motion behavior data of the target group in motion specifically includes: Extracting key car navigation data from the real-time motion behavior data of the car; wherein the key car navigation data at least includes: navigation destination, travel purpose, lane where the car is located, and driver gender; Predicting a number of vehicle driving behavior intentions based on the lane in which the vehicle is located; The execution probability of the plurality of automobile driving behavior intentions is predicted with reference to the navigation destination, the travel purpose, and the driver's gender.

4. The method according to claim 3, characterized in that The predicting of the expected driving behavior of the vehicle according to the expected behavior of the target group in motion, the traffic instruction data and the real-time driving data specifically includes: The expected driving behavior of the vehicle is predicted based on the expected behavior of the pedestrian or non-motor vehicle and its execution probability, the plurality of vehicle driving behavior intentions and their execution probability, the traffic indication data and the real-time driving data.

5. The method according to claim 4, characterized in that The predicting of the expected driving behavior of the vehicle according to the expected behavior of the pedestrian or non-motor vehicle and its execution probability, the execution probability of the plurality of vehicle driving behavior intentions, the traffic instruction data and the real-time driving data specifically includes: Extracting the lane where the vehicle is located from the real-time driving data, and predicting a number of driving behavior intentions of the vehicle according to the lane where the vehicle is located; Determining a first influence coefficient on the driving of the vehicle according to the expected behavior of the pedestrian or non-motor vehicle and its execution probability; Determining a second influence coefficient of each of the plurality of vehicle driving behavior intentions and their execution probabilities on the driving of the vehicle; Conditionally coupling the vehicle driving data with reference to the traffic instruction data and / or the vehicle driving data to determine a third influence coefficient affecting the vehicle driving; The target execution probability of each of the plurality of driving behavior intentions of the vehicle is determined according to the first influencing factor, the second influencing coefficient, and the third influencing coefficient.

6. The method according to claim 5, characterized in that The method of coupling the expected driving behavior of the vehicle with the basic driving decision to obtain a final driving decision for controlling driving specifically includes: Extracting basic execution probabilities for each of the plurality of own-vehicle driving behavior intentions from the basic driving decision; The basic execution probabilities and the target execution probabilities of the plurality of driving behavior intentions of the vehicle are integrated to obtain the final execution probabilities of the plurality of driving behavior intentions of the vehicle; The behavior intention with the highest probability is selected from the final execution probabilities of the plurality of driving behavior intentions of the vehicle as the final driving decision.

7. The method according to claim 1, characterized in that After the automatic driving based on the final driving decision, the method further includes: Verifying the driving behavior corresponding to the final driving decision, and assigning a reward or penalty value of a reward or penalty mechanism to the final driving decision according to the verification result; When the penalty value of the final driving decision reaches a set threshold, the final driving decision is uploaded to the cloud server for learning to update the basic driving decision.

8. An automatic driving device, characterized in that: include: An acquisition module is used to acquire real-time driving data of the vehicle and collect real-time environmental data around the vehicle; wherein the real-time driving data includes the location of the current road section where the vehicle is located; the real-time environmental data includes: real-time motion behavior data of the target group in motion, and traffic indication data; A communication module, used for downloading the basic driving decision corresponding to the current road section from a cloud server; wherein the basic driving decision is obtained by learning the historical driving decisions taken by all vehicles passing through the current road section by the cloud server; A first prediction module is used to predict the expected behavior of the target group in motion according to the real-time motion behavior data of the target group in motion; A second prediction module is used to predict the expected driving behavior of the vehicle according to the expected behavior of the target group in motion, the traffic instruction data and the real-time driving data; A coupling module is used to couple the expected driving behavior of the vehicle with the basic driving decision to obtain a final driving decision for controlling driving, and automatically drive based on the final driving decision.

9. An intelligent driving car, comprising a memory, an intelligent driving domain controller, and a computer program stored in the memory and executable on the intelligent driving domain controller, characterized in that: When the intelligent driving domain controller executes the program, the steps of the method described in any one of claims 1-7 are implemented.

10. A driving control system, characterized in that: The system includes: a cloud server and the intelligent driving car as described in claim 9.

Images