An automatic driving method, device and electronic equipment

Abstract

The application discloses an automatic driving method, device and electronic equipment. The method comprises the following steps: obtaining a target planning route of a current vehicle; determining reference control information of the current vehicle when driving on the target planning route according to driving data of each non-automatic driving vehicle when driving on the target planning route; and controlling the current vehicle to drive on the target planning route according to the reference control information. According to the method, the reference control information is determined according to the driving data of each non-automatic driving vehicle when driving on the target planning route, and then the automatic driving is realized according to the reference control information, so that the automatic driving technology can learn the driving experience of excellent drivers, and the safety of the automatic driving is improved.

Description

Technical Field

[0001] This application relates to the field of vehicle intelligent control technology, and in particular to an autonomous driving method, device and electronic device. Background Technology

[0002] With the advancement of artificial intelligence technology, autonomous driving technology has developed rapidly. Autonomous driving technology refers to the technology that relies on computers and artificial intelligence to assist or replace drivers on the road without human intervention, performing a series of operations such as following other vehicles, braking, and changing lanes. Therefore, autonomous driving technology can be used to plan the driving routes for autonomous vehicles.

[0003] Currently, autonomous driving technology mainly divides the autonomous driving system into subsystems such as lane line recognition, vehicle recognition, pedestrian detection, and vehicle control. The lane line recognition subsystem outputs lane line perception information, the vehicle recognition subsystem outputs vehicle perception information, the pedestrian detection subsystem outputs pedestrian perception information, and the vehicle control subsystem uses this perception information to generate a local high-precision map of the autonomous vehicle centered on its current location. Based on the local high-precision map, the system generates a planned route for the vehicle in the next time period, thereby determining the control information for the autonomous vehicle to drive on the planned route.

[0004] Since the aforementioned control information relies solely on the planned route, which is generated based on environmental perception information, and this environmental perception information is highly dependent on the recognition accuracy of the onboard sensors, misidentification by the sensors will affect the accuracy of the planned route, thereby affecting the accuracy of the control information and jeopardizing the driving safety of autonomous driving. Summary of the Invention

[0005] This application provides an autonomous driving method, device, and electronic device that can determine reference control information based on the driving data of each non-autonomous vehicle traveling on a target planned route, and then realize autonomous driving based on the reference control information, so that the autonomous driving technology can learn from the driving experience of excellent drivers and improve the safety of autonomous driving.

[0006] In a first aspect, this application provides an autonomous driving method, the method comprising:

[0007] Obtain the target route for the current vehicle;

[0008] Based on the driving data of each non-autonomous vehicle while driving on the target planned route, the reference control information of the current vehicle while driving on the target planned route is determined;

[0009] Based on the reference control information, the current vehicle is controlled to travel on the target planned route.

[0010] The above-mentioned autonomous driving method determines reference control information based on the driving data of each non-autonomous vehicle on the target planned route, and then realizes autonomous driving based on the reference control information. This allows the autonomous driving technology to learn from the driving experience of excellent drivers and improve the safety of autonomous driving.

[0011] In one possible scenario, obtaining the target planned route for the current vehicle includes:

[0012] Extract the first feature data from the toroidal image and generate the first feature map corresponding to the first feature data, wherein the first feature map is the front view corresponding to the toroidal image;

[0013] The first feature map is visually corrected to obtain a second feature map, wherein the second feature map is a top view corresponding to the first feature map;

[0014] The target planned route is determined in the second feature map.

[0015] By using the above method to convert the surround view image into a second feature map, we can obtain environmental information of the current vehicle with less distortion and higher realism.

[0016] In one possible scenario, determining the target planned route in the second feature map includes:

[0017] The historical driving trajectory of the current vehicle and the predicted driving trajectories of the surrounding vehicles in the next time period are determined based on the second feature map.

[0018] Based on the historical driving trajectory and the predicted driving trajectory, the various planned routes that the current vehicle can travel are determined.

[0019] Based on the obstacle information in the second feature map, the target planned route is determined among the various planned routes.

[0020] Using the above method, based on the current vehicle's historical driving trajectory and the predicted driving trajectories of surrounding vehicles, various planned routes that the current vehicle can travel are determined, and then a more accurate target planned route can be determined from among these planned routes.

[0021] In one possible scenario, determining the various planned routes that the current vehicle can travel based on the historical driving trajectory and the predicted driving trajectory includes:

[0022] Obtain each first trajectory point corresponding to the historical driving trajectory, and each second trajectory point corresponding to the predicted driving trajectory;

[0023] Determine the vehicle control data corresponding to each of the first trajectory points and each of the second trajectory points;

[0024] Based on the control data of each vehicle, the driving probability of the current vehicle at each drivable trajectory point in the next time period is calculated.

[0025] Based on the drivable trajectory points corresponding to the target drivability in each drivability probability, the planned routes for the current vehicle in the next time period are obtained.

[0026] Using the above method, each planned route can be determined based on the probability of the current vehicle traveling to each drivable trajectory point in the next time period.

[0027] In one possible scenario, determining the reference control information of the current vehicle while traveling on the target planned route includes:

[0028] Acquire driving data of each non-autonomous vehicle while it travels along the target planned route;

[0029] Based on the various driving data, the data range for safe driving of each non-autonomous vehicle is determined;

[0030] Within the data range, the reference control information corresponding to the current vehicle is determined.

[0031] By using the above method, reference control information can be determined based on the driving data of each non-autonomous vehicle traveling on the target planned route, which can improve the safety of autonomous driving.

[0032] In one possible scenario, controlling the current vehicle to travel on the target planned route based on the reference control information includes:

[0033] Determine the target control information corresponding to the target planned route;

[0034] The target control information is compared with the reference control information;

[0035] If the comparison is inconsistent, the target control information will be adjusted according to the reference control information;

[0036] The current vehicle is controlled to travel on the target planned route according to the adjusted target control information.

[0037] By using the above method, the target control information corresponding to the target planned route is adjusted according to the reference control information, so that the autonomous driving technology can learn from the experience of excellent drivers and improve the safety of autonomous driving.

[0038] Secondly, this application provides an autonomous driving device, the device comprising:

[0039] The acquisition module is used to obtain the target planned route for the current vehicle;

[0040] The determination module is used to determine the reference control information of the current vehicle when it is traveling on the target planned route, based on the driving data of each non-autonomous vehicle traveling on the target planned route.

[0041] The control module is used to control the current vehicle to travel on the target planned route based on the reference control information.

[0042] In one possible scenario, the acquisition module includes:

[0043] An extraction unit is used to extract first feature data from a toroidal image and generate a first feature map corresponding to the first feature data, wherein the first feature map is a front view corresponding to the toroidal image;

[0044] The correction unit is used to perform visual correction on the first feature map to obtain a second feature map, wherein the second feature map is a top view corresponding to the first feature map;

[0045] A determining unit is used to determine the target planned route in the second feature map.

[0046] In one possible scenario, the determining unit is specifically used for:

[0047] Secondly, based on the second feature map, the historical driving trajectory of the current vehicle and the predicted driving trajectories of the surrounding vehicles of the current vehicle in the next time period are determined.

[0048] Based on the historical driving trajectory and the predicted driving trajectory, the various planned routes that the current vehicle can travel are determined.

[0049] Based on the obstacle information in the second feature map, the target planned route is determined among the various planned routes.

[0050] In one possible scenario, the determining unit is further configured to:

[0051] Obtain each first trajectory point corresponding to the historical driving trajectory, and each second trajectory point corresponding to the predicted driving trajectory;

[0052] Determine the vehicle control data corresponding to each of the first trajectory points and each of the second trajectory points;

[0053] Based on the control data of each vehicle, the driving probability of the current vehicle at each drivable trajectory point in the next time period is calculated.

[0054] Based on the drivable trajectory points corresponding to the target drivability in each drivability probability, the planned routes for the current vehicle in the next time period are obtained.

[0055] In one possible scenario, the determining module is specifically used for:

[0056] Acquire driving data of each non-autonomous vehicle while it travels along the target planned route;

[0057] Based on the various driving data, the data range for safe driving of each non-autonomous vehicle is determined;

[0058] Within the data range, the reference control information corresponding to the current vehicle is determined.

[0059] In one possible scenario, the control module is specifically used for:

[0060] Determine the target control information corresponding to the target planned route;

[0061] The target control information is compared with the reference control information;

[0062] If the comparison is inconsistent, the target control information will be adjusted according to the reference control information;

[0063] The current vehicle is controlled to travel on the target planned route according to the adjusted target control information.

[0064] Thirdly, this application provides an electronic device, comprising:

[0065] Memory, used to store computer programs;

[0066] When the processor executes the computer program stored in the memory, it implements the above-described steps of the autonomous driving method.

[0067] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described autonomous driving method steps.

[0068] Based on the autonomous driving method provided in this application, reference control information is determined according to the driving data of each non-autonomous vehicle when driving on the target planned route, and then autonomous driving is realized according to the reference control information, so that the autonomous driving technology can learn from the driving experience of excellent drivers and improve the safety of autonomous driving.

[0069] The technical effects of each of the second to fourth aspects mentioned above, as well as the technical effects that each aspect may achieve, are described above with reference to the technical effects that can be achieved for the first aspect or the various possible solutions in the first aspect, and will not be repeated here. Attached Figure Description

[0070] Figure 1 A flowchart of an autonomous driving method provided in this application;

[0071] Figure 2 A schematic diagram for determining a vehicle trajectory provided in this application;

[0072] Figure 3 A flowchart for determining a driving route is provided in this application;

[0073] Figure 4 This application provides an illustration of an application scenario for an autonomous driving device.

[0074] Figure 5 A schematic diagram of the structure of an autonomous driving device provided in this application;

[0075] Figure 6 This is a schematic diagram of the structure of an electronic device provided in this application. Detailed Implementation

[0076] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The specific operational methods in the method embodiments can also be applied to the device embodiments or system embodiments. It should be noted that in the description of this application, "multiple" is understood as "at least two". "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing together, or B existing alone. A connected to B can represent: A and B directly connected, or A and B connected through C. Furthermore, in the description of this application, terms such as "first" and "second" are used only for distinguishing the purpose of description and should not be construed as indicating or implying relative importance or order.

[0077] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0078] Currently, the driving routes of autonomous vehicles are generated based on environmental perception information. This environmental perception information is highly dependent on the recognition accuracy of onboard sensors. When sensors misidentify, it will affect the accuracy of the driving route generation. Since the vehicle control information during autonomous driving is determined based on the driving route, including vehicle speed, steering angle, etc., if the accuracy of the driving route is not high, it will affect the realization of autonomous driving.

[0079] To address the aforementioned problems, this application provides an autonomous driving method that determines reference control information based on the driving data of each non-autonomous vehicle traveling on a target planned route. Autonomous driving is then implemented based on this reference control information, allowing the autonomous driving technology to learn from the driving experience of experienced drivers and improving its safety. The methods and apparatus described in this application are based on the same technical concept. Since the principles by which the methods and apparatus solve the problems are similar, embodiments of the apparatus and methods can be referred to interchangeably, and repeated details will not be elaborated further.

[0080] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application is provided in conjunction with the accompanying drawings.

[0081] like Figure 1 The diagram shown is a flowchart of an autonomous driving method provided in this application, which specifically includes the following steps:

[0082] S11, Obtain the target planned route for the current vehicle;

[0083] S12, based on the driving data of each non-autonomous vehicle while driving on the target planned route, determine the reference control information of the current vehicle while driving on the target planned route;

[0084] S13, based on the reference control information, control the current vehicle to travel on the target planned route.

[0085] In this embodiment, the vehicle has autonomous driving capabilities and is equipped with multiple onboard surround-view cameras, such as fisheye cameras. After acquiring surround-view images of the vehicle's surroundings through these cameras, the vehicle's position and size information need to be calibrated. Furthermore, the images need preprocessing, including removing interference from useless features such as leaves on the road or clouds in the sky. Preferably, salient features in the surround-view images also need data augmentation. These salient features, such as outlines and lane lines, reflect the vehicle's position and size information. After the calibration, preprocessing, and data augmentation operations, a feature extraction network is used to extract features from the surround-view images, resulting in first feature data containing the vehicle's position and size information.

[0086] Furthermore, since the first feature data is obtained from various surround-view cameras, it contains multiple perspectives of the same object. In this embodiment, a multi-layer perceptron (MLP) can be used to fuse the feature data corresponding to the same object in the first feature data to obtain multi-view data corresponding to each object. Based on the multi-view data, a first feature map corresponding to the first feature data is generated, where the first feature map is the front view corresponding to the surround-view image. Compared to the surround-view feature map, each object in the first feature map can be represented by multi-view data, and the distortion is smaller and the realism is higher.

[0087] After obtaining the first feature map, the camera extrinsic parameters corresponding to each surround-view camera are acquired. These parameters determine the camera's position and orientation in a certain space, such as rotation and translation parameters. Then, the first feature map is visually corrected using these camera extrinsic parameters to obtain the second feature map. The second feature map is a top-down view corresponding to the first feature map, such as a bird's-eye view (BEV). BEV refers to a technique that uses neural networks to transform visual information from image space to BEV space.

[0088] Using the above method, the panoramic image can be converted into a second feature map with less distortion and higher realism.

[0089] After obtaining the second feature map, the target driving route and the target planned route are further determined from the second feature map. The specific determination method can be as follows:

[0090] First, the historical driving trajectory of the current vehicle and the predicted driving trajectories of surrounding vehicles in the next time period are determined based on the second feature map. Specifically, the second feature map is input into a spatial-temporal decoder for decoding to obtain the historical driving trajectory of the current vehicle and the historical driving trajectories of surrounding vehicles. The predicted driving trajectories of surrounding vehicles can then be obtained from their historical driving trajectories. For example... Figure 2 As shown, the current vehicle is vehicle A, and the surrounding vehicles are vehicle B. The second feature map is input into the spatial-temporal decoder for decoding to obtain the historical driving trajectory of vehicle A as L1 and the historical driving trajectory of vehicle B as L2. Furthermore, based on L2, the predicted driving trajectory of vehicle B can be obtained as L3.

[0091] Then, the historical driving trajectory of the current vehicle, the historical driving trajectory of surrounding vehicles, and the predicted driving trajectory of surrounding vehicles are extracted from the second feature map through the feature extraction network. The first fully connected graph corresponding to each driving trajectory is generated by the graph neural network. The graph neural network (GNN) is an algorithm that uses neural networks to extract and discover the driving trajectory of each vehicle and generate a fully connected graph. The first fully connected graph represents the relationship graph corresponding to each driving trajectory.

[0092] Furthermore, different instances, such as vehicles and drivable lanes, are added to the first fully connected graph. This enables information exchange and the establishment of relationships between different driving trajectories and instances, such as the positional relationships between vehicles and between vehicles and lanes. Simultaneously, the historical driving trajectory of the current vehicle and a standard-definition map (e.g., Gaode Maps) are added to the first fully connected graph. The historical driving trajectory of the current vehicle is matched to the standard-definition map, resulting in a second fully connected graph. This allows for the fusion of information from the historical driving trajectory of the current vehicle, the historical driving trajectories of surrounding vehicles, the predicted driving trajectories of surrounding vehicles, the relationships between instances, and the standard-definition map.

[0093] After obtaining the second fully connected graph, the various drivable routes of the current vehicle are determined in each instance relationship. The historical driving trajectory of the current vehicle, the predicted driving trajectory of surrounding vehicles, and each drivable route are encoded by a gated cyclic encoder to obtain each first trajectory point corresponding to the historical driving trajectory of the current vehicle, each second trajectory point corresponding to the predicted driving trajectory of surrounding vehicles, and each third trajectory point corresponding to each drivable route. The distance from each trajectory point to the current vehicle is further determined, and each target trajectory point whose distance to the current vehicle is less than a preset distance value is selected. Data processing resources are then allocated to each target trajectory point.

[0094] Furthermore, the trajectory point with the smallest distance to the current vehicle among all target trajectory points is determined as the initial point. Using a graph traversal discretization strategy, starting from the initial point, the search is performed according to a preset search method, such as searching in descending order of importance. All trajectory points in the graph are visited along the graph edges, and each trajectory point is visited only once. Based on the vehicle control data corresponding to each of the searched trajectory points, the driving probability of the current vehicle for each drivable trajectory point in the next time period is calculated.

[0095] Next, the driving probabilities of each drivable trajectory point are filtered through a multi-head attention layer and MLP. The driving probabilities that are greater than the preset probability value are determined as the target driving probabilities. Based on the drivable trajectory points corresponding to each target driving probability, the planned routes for the current vehicle in the next time period are obtained.

[0096] Finally, based on the obstacle information in the second feature map, the target planned route is determined from each planned route. Specifically, the second feature map is input into a multi-head attention-based translation decoder to highlight salient features. Further, it passes through a multilayer perceptron to determine the information corresponding to the salient features, such as pedestrians, vehicles, or safety cones. Based on the salient feature information, obstacle information around the current vehicle is obtained, including both dynamic and static obstacle information. The acquired obstacle information is then input into the second fully connected graph, and the target planned route is determined from each planned route based on the obstacle information. Figure 3 .

[0097] Using the above method, based on the current vehicle's historical driving trajectory and the predicted driving trajectories of surrounding vehicles, various planned routes that the current vehicle can travel are determined, and then a more accurate target planned route can be determined from among these planned routes.

[0098] After obtaining the target planned route, the reference control information for the current vehicle is determined based on the driving data of each non-autonomous vehicle on the target planned route. Specifically, the database stores a large amount of driving data of excellent drivers on various driving routes, such as vehicle speed and turning angle. By searching the database, the driving data of each non-autonomous vehicle on the target planned route can be determined. Furthermore, the data range for safe driving of each non-autonomous vehicle is determined from the driving data, and the corresponding reference control information for the current vehicle is determined within the data range.

[0099] For example, non-autonomous vehicle 1 travels at a speed of 31 on the target planned route, non-autonomous vehicle 2 travels at a speed of 40 on the target planned route, non-autonomous vehicle 3 travels at a speed of 25 on the target planned route, non-autonomous vehicle 4 travels at a speed of 33 on the target planned route, and non-autonomous vehicle 5 travels at a speed of 35 on the target planned route. The safe driving speed range for each non-autonomous vehicle is determined to be (30~35). Speeds 1, 4, and 5 are all within this speed range. Furthermore, by calculating the average of speeds 1, 4, and 5, the reference control information corresponding to the current vehicle is determined, i.e., the reference speed is 33.

[0100] By using the above method, reference control information can be determined based on the driving data of each non-autonomous vehicle traveling on the target planned route, which can improve the safety of autonomous driving.

[0101] After obtaining the reference control information, the system controls the vehicle to travel on the target planned route. Specifically, based on the target planned route and the vehicle's driving data, target control information corresponding to the vehicle's travel on the target planned route is generated, such as the vehicle's speed and steering angle. This target control information is then input into a Generative Adversarial Network (GAN), a deep learning model that includes at least two modules: a generative model and a discriminative model. The reference control information is then processed through a multilayer perceptron to extract data corresponding to the target control information, which is then used as ground truth input into the GAN's discriminative module. The discriminative module compares the target control information with the reference control information. If the comparison is inconsistent, the generative module adjusts the target control information according to the reference control information, ultimately generating vehicle braking, steering wheel angle, wheel speed, and other behaviors that conform to human driver behavior norms. Upon receiving these behaviors, the control module controls the vehicle to travel on the target planned route according to these behaviors.

[0102] The above-mentioned autonomous driving method determines reference control information based on the driving data of each non-autonomous vehicle on the target planned route, and then realizes autonomous driving based on the reference control information. This allows the autonomous driving technology to learn from the driving experience of excellent drivers and improve the safety of autonomous driving.

[0103] To further illustrate the autonomous driving method provided in this application, the following explanation uses specific application scenarios. For example... Figure 4 As shown, in this application scenario, the acquired surround-view image data is used to directly output the current vehicle control information through an end-to-end network, thereby achieving autonomous driving of the vehicle on the target planned route. The end-to-end network is trained using driving data from various skilled drivers navigating the target planned route, and can be divided into four parts. Specifically:

[0104] In the first part of the network model, the input is the toroidal image data.

[0105] (1) The panoramic image is preprocessed, input into the feature extraction network, and then passed through a multilayer perceptron to generate a perspective feature map, which is equivalent to the first feature map in the above scheme.

[0106] (2) Convert the perspective feature map into a BEV feature map using camera extrinsic parameters to complete the view transformation;

[0107] (3) The generated BEV feature map is decoded by the decoder to obtain local semantic fragments.

[0108] In the second part of the network model, the input is the BEV feature map from the first part.

[0109] (1) Input the BEV feature map into the transformer decoder, which contains multiple multi-head attention and MLP layers to obtain feature vectors, which are equivalent to the salient features in the above scheme;

[0110] (2) The feature vector is passed through a multilayer perceptron to obtain obstacle information and complete the extraction of dynamic and static obstacles around the current vehicle.

[0111] In the third part of the network model, the inputs are the BEV feature map and generated local semantic fragments from the first part of the network, the obstacle information from the second part, and the standard definition map.

[0112] (1) Input the BEV feature map into the space-time decoder to obtain the predicted driving trajectory of the surrounding vehicles;

[0113] (2) Extract each driving trajectory from the local semantic segment, and then build a fully connected graph corresponding to each driving trajectory through a graph neural network;

[0114] (3) Add each instance to the graph neural network so that each instance can exchange information and establish relationships between each instance;

[0115] (4) Input the standard definition map and the current vehicle location into the graph neural network, and match the historical driving trajectory of the current vehicle onto the standard definition map;

[0116] (5) Input the historical driving trajectory of the current vehicle into the graph neural network, and encode the predicted driving trajectory of the surrounding vehicles, the historical driving trajectory of the current vehicle, and the relationship between each instance through the gated cyclic encoder to obtain motion code, surrogate code and initial node code, which are equivalent to each first trajectory point, each second trajectory point and each third trajectory point in the above scheme.

[0117] (6) The obtained motion code, proxy code and initial node code are input into the attention mechanism of the node, and the graph traversal discrete strategy is used to obtain the driving probabilities of the current vehicle.

[0118] (7) The various driving probabilities are filtered through a multi-head attention layer and a multi-layer perceptron to obtain the planned routes with the highest probabilities;

[0119] (8) The dynamic and static obstacles are filtered by the neural network of the input graph of each planned route, and the target planned route is obtained in the fully connected graph.

[0120] In the fourth part of the network model, the input is the target planned route of the third part of the network, as well as the driving data of the vehicles when each excellent driver drives the vehicle on the target planned route.

[0121] (1) Generate the absolute speed and turning angle of the current vehicle from the final motion trajectory;

[0122] (2) The driver’s driving data is fed into the recognizer of the generative adversarial network as the network truth value after passing through the multilayer perceptron. At the same time, the generator in the generative adversarial network receives the absolute speed and the angle of rotation. Then, the recognizer identifies the behavior generated by the generator and outputs the current vehicle control information.

[0123] Based on the same inventive concept, this application also provides an autonomous driving device, such as... Figure 4 The diagram shown is a structural schematic of an autonomous driving device according to this application. The device includes:

[0124] Module 51 is used to obtain the target planned route of the current vehicle;

[0125] The determining module 52 is used to determine the reference control information of the current vehicle when it is traveling on the target planned route based on the driving data of each non-autonomous vehicle when it is traveling on the target planned route.

[0126] The control module 53 is used to control the current vehicle to travel on the target planned route according to the reference control information.

[0127] In one possible scenario, the acquisition module 51 includes:

[0128] An extraction unit is used to extract first feature data from a toroidal image and generate a first feature map corresponding to the first feature data, wherein the first feature map is a front view corresponding to the toroidal image;

[0129] The correction unit is used to perform visual correction on the first feature map to obtain a second feature map, wherein the second feature map is a top view corresponding to the first feature map;

[0130] A determining unit is used to determine the target planned route in the second feature map.

[0131] In one possible scenario, the determining unit is specifically used for:

[0132] Secondly, based on the second feature map, the historical driving trajectory of the current vehicle and the predicted driving trajectories of the surrounding vehicles of the current vehicle in the next time period are determined.

[0133] Based on the historical driving trajectory and the predicted driving trajectory, the various planned routes that the current vehicle can travel are determined.

[0134] Based on the obstacle information in the second feature map, the target planned route is determined among the various planned routes.

[0135] In one possible scenario, the determining unit is further configured to:

[0136] Obtain each first trajectory point corresponding to the historical driving trajectory, and each second trajectory point corresponding to the predicted driving trajectory;

[0137] Determine the vehicle control data corresponding to each of the first trajectory points and each of the second trajectory points;

[0138] Based on the control data of each vehicle, the driving probability of the current vehicle at each drivable trajectory point in the next time period is calculated.

[0139] Based on the drivable trajectory points corresponding to the target drivability in each drivability probability, the planned routes for the current vehicle in the next time period are obtained.

[0140] In one possible scenario, the determining module 52 is specifically used for:

[0141] Acquire driving data of each non-autonomous vehicle while it travels along the target planned route;

[0142] Based on the various driving data, the data range for safe driving of each non-autonomous vehicle is determined;

[0143] Within the data range, the reference control information corresponding to the current vehicle is determined.

[0144] In one possible scenario, the control module 53 is specifically used for:

[0145] Determine the target control information corresponding to the target planned route;

[0146] The target control information is compared with the reference control information;

[0147] If the comparison is inconsistent, the target control information will be adjusted according to the reference control information;

[0148] The current vehicle is controlled to travel on the target planned route according to the adjusted target control information.

[0149] The aforementioned autonomous driving device determines reference control information based on the driving data of each non-autonomous vehicle traveling on the target planned route. Then, autonomous driving is achieved based on the reference control information, enabling the autonomous driving technology to learn from the driving experience of excellent drivers and improve the safety of autonomous driving.

[0150] Based on the same inventive concept, this application also provides an electronic device that can realize the functions of the aforementioned autonomous driving device. (Refer to...) Figure 6 The electronic device includes:

[0151] At least one processor 61 and a memory 62 connected to at least one processor 61. In this embodiment, the specific connection medium between the processor 61 and the memory 62 is not limited. Figure 6 The example shown is the connection between processor 61 and memory 62 via bus 60. Bus 60 is... Figure 6 The connections between other components are indicated by thick lines and are for illustrative purposes only, not as limiting information. Bus 60 can be divided into address bus, data bus, control bus, etc., for ease of representation. Figure 6 The term 61 is represented by a single thick line, but this does not imply that there is only one bus or one type of bus. Alternatively, the processor 61 can also be called a controller; there is no restriction on the name.

[0152] In this embodiment, memory 62 stores instructions executable by at least one processor 61. By executing the instructions stored in memory 62, at least one processor 61 can perform the autonomous driving method described above. Processor 61 can implement... Figure 5 The functions of each module in the device shown.

[0153] The processor 61 is the control center of the device. It can connect to various parts of the control device through various interfaces and lines. By running or executing instructions stored in memory 62 and calling data stored in memory 62, the processor can perform various functions and process data, thereby monitoring the device as a whole.

[0154] In one possible design, processor 61 may include one or more processing units. Processor 61 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into processor 61. In some embodiments, processor 61 and memory 62 may be implemented on the same chip; in some embodiments, they may also be implemented on separate chips.

[0155] Processor 61 can be a general-purpose processor, such as a central processing unit (CPU), digital signal processor, application-specific integrated circuit, field-programmable gate array or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the autonomous driving method disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0156] Memory 62, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. Memory 62 may include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type memory, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic storage, magnetic disk, optical disk, etc. Memory 62 can be any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited thereto. In the embodiments of this application, memory 62 may also be a circuit or any other device capable of implementing storage functions for storing program instructions and / or data.

[0157] By designing and programming the processor 61, the code corresponding to the autonomous driving method described in the foregoing embodiments can be embedded into the chip, enabling the chip to execute the code during runtime. Figure 1 The steps of the autonomous driving method shown in the embodiment are described. How to design and program the processor 61 is a technique well-known to those skilled in the art and will not be elaborated here.

[0158] Based on the same inventive concept, embodiments of this application also provide a storage medium storing computer instructions that, when executed on a computer, cause the computer to perform the aforementioned autonomous driving method.

[0159] In some possible implementations, various aspects of the autonomous driving method provided in this application may also be implemented in the form of a program product, which includes program code that, when the program product is run on a device, causes the control device to perform the steps in the autonomous driving method according to the various exemplary embodiments of this application described above.

[0160] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0161] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0162] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0163] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0164] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. An autonomous driving method, characterized in that, The method includes: The target planned route of the current vehicle is determined from the various planned routes that the current vehicle can drive; the various planned routes that the current vehicle can drive are determined based on the historical driving trajectory and the predicted driving trajectory; the historical driving trajectory and the predicted driving trajectory are determined based on the second feature map; Based on the driving data of each non-autonomous vehicle while driving on the target planned route, the reference control information of the current vehicle while driving on the target planned route is determined; Based on the reference control information, the current vehicle is controlled to travel on the target planned route; The determination of the reference control information for the current vehicle while traveling on the target planned route includes: Acquire driving data of each non-autonomous vehicle while it travels along the target planned route; Based on the various driving data, the data range for safe driving of each non-autonomous vehicle is determined; Within the data range, the reference control information corresponding to the current vehicle is determined; The planned routes that the current vehicle can travel are determined in the following way: Obtain each first trajectory point corresponding to the historical driving trajectory, and each second trajectory point corresponding to the predicted driving trajectory; Determine the vehicle control data corresponding to each of the first trajectory points and each of the second trajectory points; Based on the control data of each vehicle, the driving probability of the current vehicle at each drivable trajectory point in the next time period is calculated. Based on the drivable trajectory points corresponding to the target drivability in each drivability probability, the planned routes for the current vehicle in the next time period are obtained.

2. The method as described in claim 1, characterized in that, The second feature map is determined in the following way: Extract the first feature data from the toroidal image and generate the first feature map corresponding to the first feature data, wherein the first feature map is the front view corresponding to the toroidal image; The first feature map is visually corrected to obtain a second feature map, wherein the second feature map is a top view corresponding to the first feature map.

3. The method as described in claim 2, characterized in that, Determining the target planned route for the current vehicle from the various planned routes that the current vehicle can travel on includes: The historical driving trajectory of the current vehicle and the predicted driving trajectories of the surrounding vehicles in the next time period are determined based on the second feature map. Based on the historical driving trajectory and the predicted driving trajectory, the various planned routes that the current vehicle can travel are determined. Based on the obstacle information in the second feature map, the target planned route is determined among the various planned routes.

4. The method as described in claim 1, characterized in that, The step of controlling the current vehicle to travel on the target planned route based on the reference control information includes: Determine the target control information corresponding to the target planned route; The target control information is compared with the reference control information; If the comparison is inconsistent, the target control information will be adjusted according to the reference control information; The current vehicle is controlled to travel on the target planned route according to the adjusted target control information.

5. An autonomous driving device, characterized in that, The device includes: The acquisition module is used to determine the target planned route of the current vehicle from the various planned routes that the current vehicle can drive; the various planned routes that the current vehicle can drive are determined based on historical driving trajectories and predicted driving trajectories; the historical driving trajectories and predicted driving trajectories are determined based on a second feature map; The determination module is used to determine the reference control information of the current vehicle when it is traveling on the target planned route, based on the driving data of each non-autonomous vehicle traveling on the target planned route. The control module is used to control the current vehicle to travel on the target planned route according to the reference control information; The determining module is specifically used for: Acquire driving data of each non-autonomous vehicle while it travels along the target planned route; Based on the various driving data, the data range for safe driving of each non-autonomous vehicle is determined; Within the data range, the reference control information corresponding to the current vehicle is determined; The determining module is also used for: Obtain each first trajectory point corresponding to the historical driving trajectory, and each second trajectory point corresponding to the predicted driving trajectory; Determine the vehicle control data corresponding to each of the first trajectory points and each of the second trajectory points; Based on the control data of each vehicle, the driving probability of the current vehicle at each drivable trajectory point in the next time period is calculated. Based on the drivable trajectory points corresponding to the target drivability in each drivability probability, the planned routes for the current vehicle in the next time period are obtained.

6. The apparatus as claimed in claim 5, characterized in that, The acquisition module includes: An extraction unit is used to extract first feature data from a toroidal image and generate a first feature map corresponding to the first feature data, wherein the first feature map is a front view corresponding to the toroidal image; The correction unit is used to perform visual correction on the first feature map to obtain a second feature map, wherein the second feature map is a top view corresponding to the first feature map.

7. The apparatus as claimed in claim 6, characterized in that, The determining module is specifically used for: The historical driving trajectory of the current vehicle and the predicted driving trajectories of the surrounding vehicles in the next time period are determined based on the second feature map. Based on the historical driving trajectory and the predicted driving trajectory, the various planned routes that the current vehicle can travel are determined. Based on the obstacle information in the second feature map, the target planned route is determined among the various planned routes.

8. The apparatus as claimed in claim 5, characterized in that, The control module is specifically used for: Determine the target control information corresponding to the target planned route; The target control information is compared with the reference control information; If the comparison is inconsistent, the target control information will be adjusted according to the reference control information; The current vehicle is controlled to travel on the target planned route according to the adjusted target control information.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, when executing a computer program stored in the memory, implements the method steps of any one of claims 1-4.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method steps of any one of claims 1-4.

Images