An intelligent control system and method based on driverless vehicles

Abstract

This application provides an intelligent control system and method for driverless vehicles. The present invention solves the problem that existing driverless vehicles do not fully consider the current road traffic conditions and the driving state of the vehicle in front when performing lane-changing operations, resulting in a reduction in driving efficiency. The present invention adopts a multi-modal information fusion technology, comprehensively considers information such as the driving speed and vehicle type of the vehicle in front, to determine whether to actively perform a lane-changing operation, thereby improving the efficiency and safety of driverless driving. In addition, the present invention also optimizes the selection of the lane-changing direction, predicts the driving efficiency of the candidate lane by evaluating the number of vehicles on the candidate lane and the lane stability, so as to avoid ineffective or unnecessary lane-changing behaviors.

Description

Technical Field

[0001] The present invention relates to the technical field of autonomous driving, and in particular to an intelligent vehicle control system and method for driverless vehicles during driving. Background Art

[0002] With the rapid development of machine learning, artificial intelligence, sensor technology, and computer vision, driverless vehicles have progressed from the theoretical stage to practical applications. The core technologies of such vehicles mainly involve environmental perception, driving route planning, and automatic execution of driving operations. Especially in complex traffic environments, the lane-changing function has become an important part of driverless vehicles, which requires the vehicle to autonomously make lane-changing decisions and effectively execute them based on the traffic conditions ahead, driving goals, and safety standards.

[0003] In current driverless vehicle technologies, lane changes are mostly caused by external factors, such as encountering obstacles, road construction, vehicle failures, or following a preset navigation route for lane changes. These traditional lane-changing strategies are mainly to avoid emergencies or comply with navigation instructions, rather than making decisions actively based on real-time traffic flow changes.

[0004] The prior art has not effectively addressed the scenario where active lane changes are required due to the slow speed of the vehicle ahead in the absence of obstacles. On multi-lane sections such as highways or urban expressways, the low speed of the vehicle ahead may slow down the speed of the following vehicle and reduce driving efficiency. The current technology lacks a mechanism that enables driverless vehicles to actively switch lanes according to the traffic flow conditions of each lane, find a smoother or more efficient driving route, thereby reducing driving time and improving fuel or battery efficiency.

[0005] Therefore, the present invention proposes an intelligent control method and system for driverless vehicles based on a relationship module, aiming to overcome the problem of ignoring real-time lane traffic conditions in the prior art, and thus improving the overall driving performance and user experience of driverless vehicles. Summary of the Invention

[0006] In view of this, the present invention provides an intelligent control method for driverless vehicles, which includes the following steps:

[0007] S1: When receiving an active lane change request, determine the candidate lane change direction through lidar devices on the side and rear sides.

[0008] S2: When the candidate lane change direction is on one side, use the candidate lane change direction determined in S1 as the final lane change direction. When the candidate lane change direction is on both sides, turn on the side front cameras on both rearview mirrors to collect side front video images, and determine the final lane change direction based on the analysis results of the side front video.

[0009] The determination of the final lane - changing direction based on the side - front video analysis results specifically includes:

[0010] Using the relationship module to obtain the association scores between targets:

[0011] ;

[0012] Let \(\alpha\) be the balance coefficient used to balance the detection confidence and the associated targets in the previous and next frames, and respectively represent the confidence of the \(i\) - th target in the \((k - 1)\) - th frame and the confidence of the \(j\) - th target in the \(k\) - th frame, where is the \(i\) - th detection result in the primary ROI proposal box in the \((k - 1)\) - th frame, is the \(j\) - th detection result in the primary ROI proposal box in the \(k\) - th frame. Iteratively calculate the association matrix for all frames in the video except the first frame to obtain the video association matrix , where \(T\) is the number of video frames;

[0013] Based on the video association matrix \(M\), screen out multiple best association connections, and calculate the stable score of each target through the best association connections: ;

[0014] where, represents the probability that the detection result of the best association connection \(t_i\) in the \(k\) - th frame belongs to target \(i\). The number of best association connections \(n\) is equal to the number of targets detected in the last frame, represents the association score between target \(i\) and target \(j\) in the \(k\) - th frame in the association connection manner \(t\).

[0015] The present invention provides an intelligent control system for driverless vehicles, and the system includes:

[0016] A candidate lane - changing direction judgment module. When the candidate lane - changing direction judgment module receives an active lane - changing request, it determines the candidate lane - changing direction through the lidar devices on the side and the rear - side;

[0017] A candidate lane - changing direction determination module: When the candidate lane - changing direction determination module receives an active lane - changing request, it determines the candidate lane - changing direction through the lidar devices on the side and the rear - side;

[0018] Final lane change direction determination module: The final lane change direction determination module is used to, when the candidate lane change direction is on one side, determine the candidate lane change direction determined in the candidate lane change direction determination module as the final lane change direction; when the candidate lane change direction is on both sides, activate the side front cameras on both side rearview mirrors to collect side front video images, and determine the final lane change direction based on the side front video analysis result;

[0019] The determination of the final lane change direction based on the side front video analysis result specifically includes:

[0020] Obtain the association scores between targets using the relationship module:

[0021] ;

[0022] is a balance coefficient used to balance the detection confidence and the associated targets in the previous and subsequent frames, and respectively represent the confidence of the i-th target in the (k - 1)-th frame and the confidence of the j-th target in the k-th frame, where is the i-th detection result in the primary ROI proposal box in the (k - 1)-th frame , is the j-th detection result in the primary ROI proposal box in the k-th frame Iteratively calculate the association matrices for all frames in the video except the first frame to obtain the video association matrix , is the number of video frames;

[0023] Based on the video association matrix M, filter out multiple best association connections, and calculate the stable score for each target through the best association connections:

[0024] ; where, represents the probability that the detection result of the best association connection ti in the k-th frame belongs to target i. The number of best association connections n is equal to the number of targets detected in the last frame, represents the association score between target i and target j in the k-th frame in the association connection manner t.

[0025] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned intelligent control method for driverless vehicles is implemented.

[0026] The present invention also provides a computer-readable storage medium storing a computer program, which when executed by a processor, implements the above-described intelligent control method for driverless vehicles.

[0027] Compared with the prior art, the present invention solves the problem that existing driverless vehicles do not fully consider the current road traffic conditions and the driving state of the vehicle in front during lane-changing operations, resulting in a reduction in driving efficiency. The present invention adopts a multi-modal information fusion technology, comprehensively considering information such as the driving speed and vehicle type of the vehicle in front to determine whether to actively perform a lane-changing operation, thereby improving the efficiency and safety of driverless driving. In addition, the present invention also optimizes the selection of the lane-changing direction by evaluating the number of vehicles on the candidate lane and the lane stability, predicting the driving efficiency of the candidate lane, so as to avoid ineffective or unnecessary lane-changing behaviors. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0029] Figure 1 It is a structural diagram of a specific target recognition network in the present application;

[0030] Figure 2 It is a schematic diagram of the setting of the camera device and the lidar device in the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] The following will describe the embodiments of the present application in detail with reference to the drawings.

[0032] The following specific examples illustrate the implementation manners of the present application. Those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The present application can also be implemented or applied through other different specific implementation manners. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, without conflict, the following embodiments and the features in the embodiments can be combined with each other. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts belong to the scope of protection of the present application.

[0033] Note that the following description relates to various aspects of embodiments within the scope of the appended claims. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on this application, those skilled in the art should understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number and aspects described herein can be used to implement a device and / or practice a method. Additionally, this device and / or method can be implemented using other structures and / or functionality in addition to one or more of the aspects described herein.

[0034] In addition, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the examples can be practiced without these specific details.

[0035] An embodiment of this specification proposes an intelligent control method based on an autonomous vehicle: The method includes the following steps:

[0036] S1: When an active lane change request is received, determine the candidate lane change direction through lidar devices on the side and rear side.

[0037] S2: When the candidate lane change direction is to one side, use the candidate lane change direction determined in S1 as the final lane change direction. When the candidate lane change direction is to both sides, activate the front side cameras on both side mirrors to collect front side video images, and determine the final lane change direction based on the front side video analysis results;

[0038] Determining the final lane change direction based on the front side video analysis results specifically includes:

[0039] Obtain the association scores between targets using a relationship module:

[0040] ;

[0041] is a balance coefficient used to balance the detection confidence and associated targets in the previous and current frames, and respectively represent the confidence of the i-th target in the (k - 1)-th frame and the confidence of the j-th target in the k-th frame, where is the i-th detection result in the primary ROI proposal box in the (k - 1)-th frame, is the j-th detection result in the primary ROI proposal box in the k-th frame. Iteratively calculate the association matrix for all frames except the first frame in the video to obtain the video association matrix , is the number of video frames;

[0042] Based on the video correlation matrix M, multiple optimal correlation connections are selected, and each target stability score is calculated through the optimal correlation connections: ;

[0043] where, represents the probability that the detection result of the optimal correlation connection ti in the k-th frame belongs to the target i. The number n of the optimal correlation connections is equal to the number of targets detected in the last frame. represents the correlation score between the target i and the target j in the k-th frame in the correlation connection mode t.

[0044] where, there is the maximum relationship score between the front and rear nodes of the optimal correlation connection. After each optimal correlation connection is selected, the element corresponding to the upper node of the optimal correlation connection in the video correlation matrix M is set to zero, and the updated video correlation matrix is used to continue selecting the optimal correlation connection to obtain the optimal correlation connection set T.

[0045] After calculating each target stability score through the optimal correlation connection combination, it further includes: according to the number of rearview mirrors detected in the images of the left front and the right front , and their respective target stability scores , judge the final lane change direction;

[0046] ;

[0047] ;

[0048] ;

[0049] where, represents the initial judgment score, and are the decay scores of the number of rearview mirrors and the decay score of the stability score.

[0050] Before determining the candidate lane change direction through the lidar devices on the side and the rear side, it also includes whether to perform an active lane change judgment, and the active lane change judgment includes:

[0051] S3: Obtain the image information collected by the front camera of the driverless vehicle;

[0052] S4: Perform specific target recognition on the image information to obtain the leading vehicle category information, analyze the driving state of the leading vehicle based on the position information and the speed information, and judge whether to perform an active lane change according to the leading vehicle category information and the analysis result of the driving state of the leading vehicle.

[0053] As Figure 2 shown, this application sets multiple cameras and lidar devices on the driverless vehicle. Among them, the front camera is set at the front of the driverless vehicle, and its specific position can be set above or below the vehicle logo of the driverless vehicle. The front camera is mainly used to capture the image in front of the vehicle, and the shooting angle can be adjusted;

[0054] Among them, the front-side cameras are set on the two side mirrors of the driverless vehicle. The front-side cameras are used to capture the images on the front sides of the vehicle, and the shooting angles can be adjusted;

[0055] Among them, the front lidar is set at the front of the driverless vehicle, and its specific position can be set above or below the vehicle logo of the driverless vehicle. The front lidar is mainly used to measure the real-time distance between the current driverless vehicle and the vehicle in front.

[0056] Among them, the side lidars are set on both sides of the driverless vehicle, and their specific positions can be set above the front wheels of the vehicle. The side lidars are mainly used to judge whether there is a vehicle driving on the side of the current driverless vehicle and the real-time distance.

[0057] Among them, the rear-side lidars are set on both sides of the driverless vehicle, and their specific positions can be set above the rear wheels of the vehicle or near the two rear taillights. The rear-side lidars are mainly used to judge whether there is a vehicle driving on the rear side of the current driverless vehicle and the real-time distance.

[0058] The driverless vehicle of this application is equipped with a GNSS positioning system, and the positioning system is used to obtain the position information of the current driverless vehicle.

[0059] This application uses the front camera to collect the image information in front of the driverless vehicle. The acquisition frequency of the front camera is preset, and the front camera is used to capture the vehicle information directly in front of the driving direction of the current driverless vehicle.

[0060] This application uses the front lidar to collect the distance information between the current driverless vehicle and the vehicle in front.

[0061] The specific process of obtaining the vehicle category information of the vehicle in front by performing specific target recognition on the image information includes;

[0062] S41: Construct a specific target recognition network;

[0063] S42: Pre-collect road vehicle data, and the road vehicle data is the rear image of the vehicle during road driving;

[0064] S43: Annotate the collected vehicle data. The annotation information includes position annotation of vehicles, license plates, and vehicle identifiers in the image, and category annotation of license plates and vehicle identifiers. Among them, the position annotation is to frame the overall rear of the vehicle, the license plate range, and the vehicle identifier range;

[0065] The license plate category annotation is: blue license plate and non - blue license plate, and the vehicle identifier category annotation is: novice identifier and non - novice identifier;

[0066] S44: Divide the annotated vehicle data to obtain a training set, and train a specific target recognition network;

[0067] S45: Input the image collected by the front - mounted camera of the current driverless vehicle into the specific target recognition network, and make an active lane - change judgment according to the leading - vehicle category information output by the specific target recognition network.

[0068] The analysis of the leading - vehicle driving state based on the position information and speed information specifically includes:

[0069] S46: When receiving a request for analyzing the driving state of the leading vehicle, start the front - mounted lidar detector, and calculate the distance S_1 between the current driverless vehicle and the leading vehicle at the first moment according to the time delay difference between the transmitted laser and the received laser;

[0070] S47: Repeat step S46 to calculate the distance S_2 between the current driverless vehicle and the leading vehicle at the second moment, and obtain the leading - vehicle speed information V_f according to the distance S_1 at the first moment, the distance S_2 at the second moment, the time interval t, and the vehicle speed information V_c of the current driverless vehicle;

[0071] S48: Start the GNSS system to obtain the position information of the current driverless vehicle, match the position information of the current driverless vehicle with the electronic map, and obtain the road information where the current driverless vehicle is traveling. The road information includes the maximum speed limit information V_max of the current road and the congestion condition of the current road;

[0072] S49: Preset a standard low - speed percentage threshold, and dynamically lower the standard low - speed percentage threshold according to the congestion condition of the current road to obtain the low - speed percentage threshold N% of the current road at the current moment. When ..., request an active lane - change, otherwise return to step S3 for image acquisition at the next moment.

[0073] The leading - vehicle speed information V_f=(S_2 - S_1) / t+V_c;

[0074] The congestion condition of the current road includes normal traffic, mild congestion, moderate congestion, and severe congestion.

[0075] In addition to the type factor of the leading vehicle, the present application also analyzes the driving state of the leading vehicle. The present application measures the speed of the leading vehicle through radar laser, and judges the driving state of the leading vehicle according to the speed of the leading vehicle and the speed limit information of the current road section. If the speed of the leading vehicle is too low, the driver of the leading vehicle may have bad driving behaviors such as driving fatigue and drowsiness, answering calls, playing with mobile phones, etc. At the same time, following a leading vehicle with a low speed is also not conducive to the passing efficiency of the driverless vehicle. Therefore, when the speed of the leading vehicle is too low, the present invention will also request an active lane change to ensure the driving safety and efficiency of the driverless vehicle.

[0076] When receiving an active lane change request, determining the candidate lane change direction through the lidar devices on the side and the rear side specifically includes:

[0077] S11: Preset the dangerous lane change area of the current driverless vehicle, and the dangerous lane change area is two sector areas constructed with the geometric center of the current driverless vehicle as the vertex;

[0078] S12: Turn on the lidar devices on the side and the rear side, and calculate whether there are vehicles driving on the side and the rear side of the current driverless vehicle and the distance of the vehicles according to the time delay difference between the transmitted laser and the received laser of the lidar devices.

[0079] S13: If there are no vehicles driving on both sides of the current driverless vehicle, determine the candidate lane change direction as both sides;

[0080] If there are vehicles driving on both sides of the current driverless vehicle, judge whether the vehicles driving on both sides are in the dangerous lane change area. If the vehicles driving on both sides are in the dangerous lane change area, cancel the lane change request; if the vehicles driving on both sides are not in the dangerous lane change area, determine the candidate lane change direction as both sides; if only the vehicle driving on one side is in the dangerous lane change area, determine the candidate lane change direction as the other side;

[0081] If there is only one vehicle driving on one side of the current driverless vehicle, judge whether the vehicle driving on this side is in the dangerous lane change area. If it is not in the dangerous area, determine the candidate lane change direction as both sides; if it is in the dangerous area, take the other side as the candidate lane change direction.

[0082] The present invention fully considers this basic element of the type of the vehicle in front in road traffic. First, a specific target recognition network is established and trained. The specific recognition network can recognize specific targets. In the actual road environment, when the vehicle in front is a vehicle such as a large truck, a large bus, or a school bus, due to its large vehicle volume and serious line of sight obstruction, the driving risk of the following vehicle will increase. Similarly, if the vehicle in front is a training vehicle or has a novice label on it, since the driver of the vehicle in front is in the training or internship period and the driving skills are not yet mature, dangerous driving behaviors such as sudden stops and sharp turns may be made. Therefore, the following vehicle behind a training vehicle or a vehicle with a novice label also has a relatively high driving risk. The driving risk of a driverless vehicle following a vehicle of the above type on the road is higher than that of other types of vehicles. Therefore, the present invention identifies the type of the vehicle in front based on the common features of large vehicles and training vehicles, namely, the yellow license plate and the circular label of the novice vehicle identification. When it is recognized that the type of the vehicle in front belongs to the vehicle that may cause an increase in the driving risk of the driverless vehicle, an active lane change request is started. The lane change request of the present invention is not a passive lane change due to a real traffic accident occurring in front or due to a driving plan, but an active lane change based on the driving risk estimated according to the type of the vehicle in front. This active lane change method according to the type of the vehicle in front can reduce the driving danger of the driverless vehicle and improve the safety of passengers during the ride.

[0083] The specific target recognition network includes a feature extraction module, a multi-scale feature fusion module, a multi-scale proposal box generation module, and a target output module;

[0084] The feature extraction layer includes four convolutional modules. An attention module is added between two adjacent convolutional modules. The outputs of the second, third, and fourth convolutional modules are respectively connected to the multi-scale feature fusion module and the multi-scale proposal box generation module;

[0085] The multi-scale proposal box generation module contains convolutional kernels of multiple scales. The convolutional kernels of the multiple scales are respectively applied to the feature maps output by the second, third, and fourth convolutional modules to obtain the position information of multiple-scale proposal boxes;

[0086] The input of the multi-scale fusion module is the feature maps output by the upsampling of the second convolutional module, the third convolutional module, and the fourth convolutional module, and the output is a fused feature map;

[0087] The position and category information of the vehicle, license plate, and vehicle identification are obtained by using the position information of the multiple-scale proposal boxes to guide the fused feature map to obtain multiple proposal boxes and inputting the fused feature map with multiple proposal boxes into the target output module.

[0088] The present invention needs to perform overall recognition of the rear of the vehicle in the captured image to determine whether there is a vehicle in front of the current driverless vehicle, and at the same time, it also needs to recognize the license plate and vehicle identification in the captured image. Obviously, the sizes of the vehicle as a whole and the license plate or identification are extremely different. If a single convolutional recognition network is used, it cannot achieve a high recognition accuracy for all-scale targets. Therefore, in this application, multi-scale convolutional kernels are used to perform convolutional processing on the feature maps at different levels in the feature extraction network, and multi-scale convolutional kernels are adopted to adapt to the target features of different scales to be recognized in this application. Performing separate multi-scale convolutional operations on the feature maps at different levels can recognize the same target by fusing the feature details of multiple feature maps, improving the target recognition accuracy and avoiding the problem of information loss of small target features due to excessive convolutional operations.

[0089] The present invention provides side front cameras at the positions of the left and right rearview mirrors of the vehicle. In the prior art, the front cameras are mostly arranged above the vehicle logo or the windshield. Although the above cameras can capture and observe a certain range in front of the vehicle during driving, due to the limited height of the vehicle and the line of sight obstruction of the vehicle in front, in fact, the front camera can only capture a very limited range in the front and side front. In the prior art, when judging the driving condition of the vehicle in front, it is often necessary to move the vehicle horizontally, that is, drive the vehicle close to one side lane line to obtain a wider front view. However, the above driving behavior of driving the vehicle close to one side lane line (straddling the lane line) increases the driving risk and is also likely to bring panic to passengers. Therefore, the present invention provides side front cameras at the left and right rearview mirrors of the vehicle to obtain a wide side front view and avoid the risky behavior of horizontal movement of the vehicle.

[0090] The present invention also optimizes the lane change selection. When it is possible to change lanes to the left and right at the same time, the present invention uses the side front camera to capture the side front image, analyzes the number of rearview mirrors in the image to judge the number of vehicles in front of the vehicle in both lanes on both sides, and at the same time uses the stable scores of the vehicles in front in both lanes to predict the safety and necessity after the lane change. When the vehicles in one lane often change lanes, actively changing lanes to that lane may not be very stable, increasing the driving risk. The present invention uses the number of vehicles in front in both lanes and the stable scores in both lanes to select the lane with lower traffic density and better stability, ensuring driving safety and driving efficiency and avoiding ineffective lane changes.

[0091] The present invention provides an intelligent control system based on a driverless vehicle, and the system includes:

[0092] A candidate lane change direction judgment module, when the candidate lane change direction judgment module receives an active lane change request, it determines the candidate lane change direction through the lidar devices on the side and the rear side;

[0093] Final Lane Change Direction Determination Module: The final lane change direction determination module is used to, when the candidate lane change direction is on one side, determine the candidate lane change direction determined in the candidate lane change direction determination module as the final lane change direction; when the candidate lane change direction is on both sides, activate the side front cameras on both side mirrors to collect side front video images, and determine the final lane change direction based on the side front video analysis results;

[0094] The determination of the final lane change direction based on the side front video analysis results specifically includes:

[0095] Obtain the association scores between targets using the relationship module:

[0096] ;

[0097] is a balance coefficient used to balance the detection confidence and the associated targets in the previous and subsequent frames, and respectively represent the confidence of the i-th target in the (k - 1)-th frame and the confidence of the j-th target in the k-th frame, where is the i-th detection result in the primary ROI proposal box in the (k - 1)-th frame , is the j-th detection result in the primary ROI proposal box in the k-th frame Iteratively calculate the association matrices for all frames in the video except the first frame to obtain the video association matrix , is the number of video frames;

[0098] Based on the video association matrix M, filter out multiple best association connections, and calculate the stable score for each target through the best association connections:

[0099] ;

[0100] wherein, represents the probability that the detection result of the best association connection ti in the k-th frame belongs to target i. The number of best association connections n is equal to the number of targets detected in the last frame, represents the association score between target i and target j in the k-th frame in the association connection mode t.

[0101] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned intelligent control method for driverless vehicles.

[0102] The present invention also provides a computer-readable storage medium storing a computer program, which when executed by a processor, implements the above-mentioned intelligent control method based on driverless vehicles.

[0103] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned method embodiments. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0104] In this specification, the same or similar parts among the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the embodiments described later, the description is relatively simple, and the relevant parts can be referred to the partial description of the foregoing embodiments.

[0105] As described above, the above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An intelligent control method for driverless vehicles, characterized in that, It includes the following steps: S1: When a proactive lane change request is received, determine the candidate lane change direction through the lidar devices on the side and rear side, including: Preset the dangerous lane change area of the current driverless vehicle in advance. The dangerous lane change area is two fan-shaped areas on both sides constructed with the geometric center of the current driverless vehicle as the vertex; turn on the lidar devices on the side and rear side, and calculate whether there are vehicles driving and the vehicle distances on the side and rear side; if there are no vehicles driving on both sides of the vehicle, determine the candidate lane change direction as both sides; if there are vehicles driving on both sides, judge whether the vehicles driving on both sides are within the dangerous lane change area. If both are within the dangerous lane change area, cancel the lane change request; if both are not within the dangerous lane change area, determine the candidate lane change direction as both sides; if only one side is within the dangerous lane change area, determine the candidate lane change direction as the other side; when there is only a vehicle driving on one side of the current driverless vehicle, judge whether the vehicle driving on this side is within the dangerous lane change area. If it is not within the dangerous area, determine the candidate lane change direction as both sides. If it is within the dangerous area, use the other side as the candidate lane change direction; S2: When the candidate lane change direction is one side, use the candidate lane change direction determined in S1 as the final lane change direction; when the candidate lane change direction is both sides, turn on the front side cameras on both rearview mirrors to collect the front side video images, and determine the final lane change direction based on the front side video analysis result, including using the front side cameras to capture the front side images, and selecting the lane by analyzing the number of vehicles in front of the vehicle in the image in the rearview mirrors; Before determining the candidate lane change direction, it also includes whether to judge a proactive lane change, including: S3: Obtain the image information collected by the front camera of the driverless vehicle; S4: Perform specific target recognition on the image information to obtain the leading vehicle category information, analyze the driving state of the leading vehicle based on the position information and speed information, and judge whether to perform a proactive lane change according to the leading vehicle category information and the leading vehicle driving state analysis result; specifically, performing specific target recognition on the image information to obtain the leading vehicle category information includes: constructing a specific target recognition network; pre-collecting road vehicle data, which is the rear image of the vehicle during road driving; annotating the collected data, including performing position annotation on the vehicle, license plate, and vehicle logo in the image and performing category annotation on the license plate and vehicle logo. Among them, the position annotation is to frame the positions of the whole rear of the vehicle, the license plate range, and the vehicle logo range; the license plate category annotation is: blue license plate and non-blue license plate, and the vehicle logo category annotation is: novice logo and non-novice logo; The analysis of the driving state of the vehicle ahead based on position information and speed information specifically includes: calculating the speed information V_f of the vehicle ahead; obtaining the position information of the current driverless vehicle, performing position matching on the position information and the electronic map to obtain the road information on which the current driverless vehicle is traveling, including the maximum speed limit information V_max of the current road and the congestion condition of the current road; presetting a standard low-speed percentage threshold, and dynamically reducing the standard low-speed percentage threshold according to the congestion condition of the current road to obtain the low-speed percentage threshold N% at the current moment on the current road, When, request an active lane change, otherwise return to step S3 to perform image acquisition at the next moment.

2. The intelligent control method for a driverless vehicle according to claim 1, wherein The specific target recognition network includes a feature extraction module, a multi-scale feature fusion module, a multi-scale proposal box generation module, and a target output module; The feature extraction layer includes four convolutional modules, an attention module is added between adjacent two convolutional modules, and the outputs of the second, third, and fourth convolutional modules are respectively connected to the multi-scale feature fusion module and the multi-scale proposal box generation module; The multi-scale proposal box generation module includes convolutional kernels of multiple scales, and applies the convolutional kernels of the multiple scales to the feature maps output by the second, third, and fourth convolutional modules respectively to obtain the position information of the proposal boxes of multiple scales; The input of the multi-scale feature fusion module is the feature maps output by the upsampling of the second convolutional module, the third convolutional module, and the fourth convolutional module, and the output is the fused feature map; The position information of the multiple scales of proposal boxes is used to guide the fused feature map to obtain multiple proposal boxes, and the fused feature map with multiple proposal boxes is input into the target output module to obtain the position and category information of the vehicle, license plate, and vehicle identification.

3. An intelligent control system for driverless vehicles, based on the intelligent control method for driverless vehicles according to any one of claims 1-2, characterized in that The system includes: A candidate lane change direction judgment module, which determines the candidate lane change direction through the lidar devices on the side and the rear side when receiving an active lane change request; A candidate lane change direction determination module: The candidate lane change direction determination module determines the candidate lane change direction through the lidar devices on the side and the rear side when receiving an active lane change request; A final lane change direction determination module: The final lane change direction determination module is used to use the candidate lane change direction determined in the candidate lane change direction determination module as the final lane change direction when the candidate lane change direction is on one side, and when the candidate lane change direction is on both sides, turn on the front side cameras on both side mirrors to collect the front side video images, and determine the final lane change direction based on the front side video analysis results.

4. An electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements a method for intelligent control of an autonomous vehicle as described in any one of claims 1-2 when executing the computer program.

5. A computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program implements a method for intelligent control of an autonomous vehicle as described in any one of claims 1 to 2 when executed by a processor.

Images