CN116985793B - Automatic driving safety control system and method based on deep learning algorithm - Google Patents
Automatic driving safety control system and method based on deep learning algorithm Download PDFInfo
- Publication number
- CN116985793B CN116985793B CN202311248489.XA CN202311248489A CN116985793B CN 116985793 B CN116985793 B CN 116985793B CN 202311248489 A CN202311248489 A CN 202311248489A CN 116985793 B CN116985793 B CN 116985793B
- Authority
- CN
- China
- Prior art keywords
- time sequence
- vehicle speed
- distance
- vehicle
- feature vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000013135 deep learning Methods 0.000 title claims abstract description 50
- 238000000034 method Methods 0.000 title claims abstract description 43
- 239000013598 vector Substances 0.000 claims abstract description 133
- 230000003993 interaction Effects 0.000 claims abstract description 77
- 238000012549 training Methods 0.000 claims description 85
- 239000011159 matrix material Substances 0.000 claims description 24
- 230000007246 mechanism Effects 0.000 claims description 19
- 238000005070 sampling Methods 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 11
- 238000000605 extraction Methods 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 6
- 238000012790 confirmation Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 2
- 230000008859 change Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000006399 behavior Effects 0.000 description 2
- 239000003795 chemical substances by application Substances 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000009941 weaving Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W30/00—Purposes of road vehicle drive control systems not related to the control of a particular sub-unit, e.g. of systems using conjoint control of vehicle sub-units
- B60W30/08—Active safety systems predicting or avoiding probable or impending collision or attempting to minimise its consequences
- B60W30/095—Predicting travel path or likelihood of collision
- B60W30/0953—Predicting travel path or likelihood of collision the prediction being responsive to vehicle dynamic parameters
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
- B60W50/08—Interaction between the driver and the control system
- B60W50/14—Means for informing the driver, warning the driver or prompting a driver intervention
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W60/00—Drive control systems specially adapted for autonomous road vehicles
- B60W60/001—Planning or execution of driving tasks
- B60W60/0015—Planning or execution of driving tasks specially adapted for safety
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2520/00—Input parameters relating to overall vehicle dynamics
- B60W2520/10—Longitudinal speed
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2554/00—Input parameters relating to objects
- B60W2554/80—Spatial relation or speed relative to objects
- B60W2554/802—Longitudinal distance
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Human Computer Interaction (AREA)
- Traffic Control Systems (AREA)
Abstract
An automatic driving safety control system and method based on a deep learning algorithm are disclosed. Firstly, acquiring vehicle distance values of a plurality of preset time points in a preset time period acquired by a laser range finder arranged on a vehicle, then acquiring vehicle speed values of the plurality of preset time points acquired by a speed sensor arranged on the vehicle, then carrying out joint analysis on the vehicle distance values of the plurality of preset time points and the vehicle speed values of the plurality of preset time points to obtain a vehicle distance-vehicle speed time sequence interaction feature vector, and finally, determining whether collision early warning prompt is generated or not based on the vehicle distance-vehicle speed time sequence interaction feature vector. Therefore, the method can be combined with a deep learning algorithm to automatically judge whether collision early warning prompt is generated or not so as to avoid dangerous situations.
Description
Technical Field
The present application relates to the field of autopilot, and more particularly, to an autopilot safety control system and method based on a deep learning algorithm.
Background
Autopilot refers to a technology that enables a vehicle to navigate and drive without human intervention using computer and sensor technology. The goal of autopilot technology is to achieve a safer, efficient and convenient transportation system. The collision early warning system can help the automatic driving vehicle to timely identify potential collision risks and take corresponding measures to avoid collision. Through timely early warning and reaction, traffic accidents can be greatly reduced, and the safety of road traffic is improved.
However, the existing sensing technology still has the problems of misjudgment and missed judgment, which may lead to inaccuracy of collision early warning. Thus, an optimized solution is desired.
Disclosure of Invention
The present application has been made in order to solve the above technical problems. The embodiment of the application provides an automatic driving safety control system and method based on a deep learning algorithm. The method can be combined with a deep learning algorithm to automatically judge whether collision early warning prompt is generated or not so as to avoid dangerous situations.
According to one aspect of the present application, there is provided an automatic driving safety control method based on a deep learning algorithm, including: acquiring vehicle distance values of a plurality of preset time points in a preset time period acquired by a laser range finder arranged on a vehicle; acquiring vehicle speed values at the plurality of predetermined time points acquired by speed sensors disposed at the vehicle; performing joint analysis on the vehicle distance values of the plurality of preset time points and the vehicle speed values of the plurality of preset time points to obtain a vehicle distance-vehicle speed time sequence interaction feature vector; and determining whether collision early warning prompt is generated or not based on the vehicle distance-vehicle speed time sequence interaction feature vector.
According to another aspect of the present application, there is provided an automatic driving safety control system based on a deep learning algorithm, including: the vehicle distance value acquisition module is used for acquiring vehicle distance values of a plurality of preset time points in a preset time period acquired by a laser range finder arranged on a vehicle; a speed value acquisition module for acquiring vehicle speed values at the plurality of predetermined time points acquired by a speed sensor disposed at the vehicle; the joint analysis module is used for performing joint analysis on the vehicle distance values of the plurality of preset time points and the vehicle speed values of the plurality of preset time points to obtain a vehicle distance-vehicle speed time sequence interaction feature vector; and the prompt confirmation module is used for determining whether collision early warning prompts are generated or not based on the vehicle distance-vehicle speed time sequence interaction feature vector.
Compared with the prior art, the automatic driving safety control system and the method based on the deep learning algorithm firstly acquire the vehicle distance values of a plurality of preset time points in a preset time period acquired by a laser range finder arranged on a vehicle, then acquire the vehicle speed values of the preset time points acquired by a speed sensor arranged on the vehicle, then perform joint analysis on the vehicle distance values of the preset time points and the vehicle speed values of the preset time points to obtain a vehicle distance-vehicle speed time sequence interaction feature vector, and finally determine whether collision early warning prompt is generated or not based on the vehicle distance-vehicle speed time sequence interaction feature vector. Therefore, the method can be combined with a deep learning algorithm to automatically judge whether collision early warning prompt is generated or not so as to avoid dangerous situations.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly introduced below, which are not intended to be drawn to scale in terms of actual dimensions, with emphasis on illustrating the gist of the present application.
Fig. 1 is a flowchart of an automatic driving safety control method based on a deep learning algorithm according to an embodiment of the present application.
Fig. 2 is a schematic architecture diagram of an autopilot safety control method based on a deep learning algorithm according to an embodiment of the present application.
Fig. 3 is a flowchart of substep S130 of the deep learning algorithm-based autopilot safety control method according to an embodiment of the present application.
Fig. 4 is a block diagram of an autopilot safety control system based on a deep learning algorithm in accordance with an embodiment of the present application.
Fig. 5 is an application scenario diagram of an automatic driving safety control method based on a deep learning algorithm according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some, but not all embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present application without making any inventive effort, are also within the scope of the present application.
As used in this application and in the claims, the terms "a," "an," "the," and/or "the" are not specific to the singular, but may include the plural, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.
Although the present application makes various references to certain modules in a system according to embodiments of the present application, any number of different modules may be used and run on a user terminal and/or server. The modules are merely illustrative, and different aspects of the systems and methods may use different modules.
Flowcharts are used in this application to describe the operations performed by systems according to embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in order precisely. Rather, the various steps may be processed in reverse order or simultaneously, as desired. Also, other operations may be added to or removed from these processes.
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application and not all of the embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Aiming at the technical problems, the technical conception of the method is that the multi-sensor acquisition technology is utilized to sense the distance information and the speed information of the current vehicle and the front vehicle, and the deep learning algorithm is combined to automatically judge whether collision early warning prompt is generated or not so as to avoid dangerous situations.
Based on this, fig. 1 is a flowchart of an automatic driving safety control method based on a deep learning algorithm according to an embodiment of the present application. Fig. 2 is a schematic architecture diagram of an autopilot safety control method based on a deep learning algorithm according to an embodiment of the present application. As shown in fig. 1 and 2, the automatic driving safety control method based on the deep learning algorithm according to the embodiment of the application includes the steps of: s110, acquiring vehicle distance values of a plurality of preset time points in a preset time period acquired by a laser range finder arranged on a vehicle; s120, acquiring vehicle speed values of the plurality of preset time points acquired by speed sensors deployed on the vehicle; s130, performing joint analysis on the vehicle distance values of the plurality of preset time points and the vehicle speed values of the plurality of preset time points to obtain a vehicle distance-vehicle speed time sequence interaction feature vector; and S140, determining whether collision early warning prompt is generated or not based on the vehicle distance-vehicle speed time sequence interaction feature vector.
Specifically, in the technical scheme of the application, firstly, vehicle distance values of a plurality of preset time points in a preset time period acquired by a laser range finder deployed on a vehicle are acquired; meanwhile, vehicle speed values at the plurality of predetermined time points acquired by speed sensors disposed at the vehicle are acquired. Wherein the distance value is a distance value between the current vehicle and the preceding vehicle.
It should be appreciated that a smaller vehicle distance generally means a closer distance between vehicles and a higher risk of collision. By analyzing the time sequence change of the vehicle distance, the vehicle distance can be evaluated to judge whether potential collision risks exist. The vehicle speed refers to the distance traveled by the vehicle per unit time. Higher vehicle speeds can increase the severity and risk of collisions. Thus, combining the vehicle distance and vehicle speed information can help to perform collision risk assessment.
And then, passing the distance values of the plurality of preset time points through a distance time sequence feature extractor comprising an upsampling layer and a one-dimensional convolution layer to obtain distance time sequence feature vectors. Namely, the distance time sequence characteristics of the distance values of the plurality of preset time points are extracted to capture the dynamic change rule and fluctuation mode of the distance values in the time dimension.
And simultaneously, passing the vehicle speed values at a plurality of preset time points through a vehicle speed time sequence feature extractor comprising an up-sampling layer and a one-dimensional convolution layer to obtain a vehicle speed time sequence feature vector. That is, the vehicle speed time sequence features of the vehicle speed values at the plurality of predetermined time points are extracted to capture the dynamic change rule and the fluctuation mode of the vehicle speed values in the time dimension.
Then, the inter-feature attention layer is used for carrying out feature interaction based on an attention mechanism on the distance time sequence feature vector and the speed time sequence feature vector so as to obtain a distance-speed time sequence interaction feature vector. It is worth mentioning that the goal of the traditional attention mechanism is to learn an attention weight matrix, applied to the individual neural nodes of the current layer, giving them greater weight for those important nodes and less weight for those secondary nodes. Because each neural node contains certain characteristic information, the neural network can select information which is more critical to the current task target from a plurality of characteristic information through the operation. The attention layers among the features are different, and the dependency relationship among the feature information is focused more.
Accordingly, as shown in fig. 3, performing joint analysis on the vehicle distance values at the plurality of predetermined time points and the vehicle speed values at the plurality of predetermined time points to obtain a vehicle distance-vehicle speed time sequence interaction feature vector, including: s131, extracting the distance time sequence characteristics of the distance values of the plurality of preset time points to obtain distance time sequence characteristic vectors; s132, extracting vehicle speed time sequence features of the vehicle speed values at a plurality of preset time points to obtain vehicle speed time sequence feature vectors; and S133, fusing the vehicle distance time sequence feature vector and the vehicle speed time sequence feature vector to obtain the vehicle distance-vehicle speed time sequence interaction feature vector. It should be appreciated that in step S130, there are three steps, each of which has its particular use. In step S131, the distance values at a plurality of predetermined time points are extracted and combined into a distance timing feature vector, which represents the distance change between vehicles at different time points. This feature vector can be used to analyze the relative position and distance trend between vehicles. In step S132, vehicle speed values at a plurality of predetermined time points are extracted and combined into a vehicle speed time series feature vector, which represents the speed change condition of the vehicle at different time points, and which can be used for analyzing the speed change trend and the running state of the vehicle. In step S133, the distance-speed time sequence feature vector and the speed time sequence feature vector are fused to obtain a distance-speed time sequence interaction feature vector, where the feature vector includes interaction information of distance and speed between vehicles, and can be used to analyze the interaction between vehicles and the feature of cooperative driving. By jointly analyzing the time sequence characteristics of the distance and the speed of the vehicle, more comprehensive information about the running state and the interactive behavior of the vehicle can be obtained, and the characteristic vectors can be used for a plurality of applications, such as traffic flow prediction, vehicle behavior recognition, intelligent driving systems and the like.
More specifically, in step S131, extracting the distance timing feature of the distance values at the plurality of predetermined time points to obtain a distance timing feature vector includes: and passing the distance values of the plurality of preset time points through a distance time sequence feature extractor comprising an upsampling layer and a one-dimensional convolution layer to obtain the distance time sequence feature vector. It is worth mentioning that the Upsampling Layer (Upsampling Layer) is an operation for increasing the data dimension, which is generally used for restoring low resolution data to high resolution data. In the range timing feature extractor, the upsampling layer may be used to increase the resolution of the time dimension to better capture details and variations in range. The one-dimensional convolution layer (1D Convolutional Layer) is a convolution operation for processing sequence data. Unlike conventional two-dimensional convolution layers, one-dimensional convolution layers perform sliding window convolution operations in one dimension, and local features in sequence data can be effectively extracted. In the distance sequential feature extractor, a one-dimensional convolution layer can be used for extracting features of the distance sequence and capturing sequential patterns and associated information of the distances. In the car distance time sequence feature extractor, the combination of the up-sampling layer and the one-dimensional convolution layer can realize the following functions: 1. the upsampling layer may increase the resolution of the time dimension, making details in the train distance sequence more apparent, helping to more accurately capture the distance variations between the vehicles. 2. The one-dimensional convolution layer can conduct convolution operation of a sliding window in the time dimension, and local features in the vehicle distance sequence, such as information of intervals between vehicles, relative speeds and the like, are extracted. 3. The up-sampling layer and the one-dimensional convolution layer are combined, so that important features in the vehicle distance sequence can be extracted while details are maintained, and a more representative vehicle distance time sequence feature vector is obtained. Through the use of the up-sampling layer and the one-dimensional convolution layer, the vehicle distance time sequence feature extractor can effectively extract key features in the vehicle distance sequence, and provides more useful information for subsequent analysis and processing.
More specifically, in step S132, extracting vehicle speed timing characteristics of the vehicle speed values at the plurality of predetermined time points to obtain a vehicle speed timing characteristic vector includes: and passing the vehicle speed values at the plurality of preset time points through a vehicle speed time sequence feature extractor comprising an upsampling layer and a one-dimensional convolution layer to obtain the vehicle speed time sequence feature vector. It will be appreciated that the vehicle speed timing feature extractor is operative to extract vehicle speed timing features and to convert vehicle speed values at a plurality of predetermined points in time into vehicle speed timing feature vectors, which function to capture timing patterns and associated information of vehicle speed, providing useful features for subsequent analysis and processing. The vehicle speed timing feature extractor typically includes an upsampling layer and a one-dimensional convolution layer, similar to the vehicle distance timing feature extractor. The up-sampling layer is used for increasing the resolution of the time dimension, so that details in the vehicle speed sequence are more obvious, and the change and trend of the vehicle speed can be better captured through up-sampling, so that more accurate time sequence features are extracted. The one-dimensional convolution layer carries out convolution operation of a sliding window in a time dimension, local features in a vehicle speed sequence can be extracted, the local features can comprise a change trend, periodicity or other important modes of the speed, and the one-dimensional convolution layer can capture time sequence information in the vehicle speed sequence and help to extract useful features. Through the combination of the upsampling layer and the one-dimensional convolution layer, the vehicle speed timing feature extractor may convert the vehicle speed sequence into a vehicle speed timing feature vector having higher dimensionality and richer information. Such feature vectors may better describe time-series variations in vehicle speed and provide more useful input for subsequent analysis tasks such as distance-to-speed interaction feature extraction.
More specifically, in step S133, fusing the distance-to-vehicle speed timing feature vector and the vehicle speed timing feature vector to obtain the distance-to-vehicle speed timing interaction feature vector includes: and performing feature interaction based on an attention mechanism on the distance time sequence feature vector and the speed time sequence feature vector by using an inter-feature attention layer to obtain the distance-speed time sequence interaction feature vector. It is worth mentioning that the attention mechanism (Attention Mechanism) is a method for weighting different parts of the attention input. In deep learning, an attention mechanism may be used to learn the correlation between different elements in the input and weight focus the different elements according to the weight of the correlation. In the distance-speed time sequence interaction feature extraction, an inter-feature attention layer is used for carrying out feature interaction based on an attention mechanism on a distance time sequence feature vector and a speed time sequence feature vector. In particular, the attention mechanism may assign different weights to the features at each point in time according to the correlation of the time series features of the vehicle distance and the vehicle speed. In this way important features will get higher weights, while unimportant features will get lower weights. The main roles of the attention mechanisms are as follows: 1. feature interaction: through the attention mechanism, the vehicle distance time sequence feature vector and the vehicle speed time sequence feature vector can pay attention to and interact with each other, so that the correlation and interaction between the vehicle distance time sequence feature vector and the vehicle speed time sequence feature vector are captured. This helps extract richer and more accurate inter-vehicle distance-speed interaction features, revealing patterns of cooperative travel and interaction between vehicles. 2. Self-adaptive weight: the attention mechanism may adaptively learn weights based on the importance and relevance of the different parts of the input. This means that the model can dynamically adjust the importance of the features according to the specific task and the characteristics of the data, and the expressive power and performance of the model are improved. 3. Feature selection: through the attentiveness mechanism, the model may selectively focus on the most relevant features in the input, ignoring the features that are not relevant or noisy. This helps to improve the robustness and generalization ability of the model, reducing interference of irrelevant information. In other words, the attention mechanism can provide a flexible way for the model to adaptively focus on different features according to the importance and relevance of the input, thereby improving the performance and expressive power of the model. In the distance-speed time sequence interaction feature extraction, an attention mechanism can capture the important relation between the distance and the speed of the vehicle, and more useful feature representation is provided for subsequent tasks and analysis.
And then, the vehicle distance-vehicle speed time sequence interaction feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether collision early warning prompt is generated or not. Correspondingly, based on the vehicle distance-vehicle speed time sequence interaction feature vector, determining whether to generate collision early warning prompt comprises the following steps: and the vehicle distance-vehicle speed time sequence interaction feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether collision early warning prompt is generated or not.
More specifically, the vehicle distance-vehicle speed time sequence interaction feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether collision early warning prompt is generated or not, and the method comprises the following steps: performing full-connection coding on the vehicle distance-vehicle speed time sequence interaction feature vector by using a full-connection layer of the classifier to obtain a coding classification feature vector; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
That is, in the technical solution of the present disclosure, the label of the classifier includes a collision pre-warning prompt (first label) generated after the weaving is completed, and no collision pre-warning prompt (second label) is generated, where the classifier determines, through a soft maximum function, to which classification label the inter-vehicle distance-vehicle speed time sequence interaction feature vector belongs. It should be noted that the first tag p1 and the second tag p2 do not include a manually set concept, and in fact, during the training process, the computer model does not have a concept of "whether collision warning is generated", which is only two kinds of classification tags, and the probability that the output feature is under the two classification tags, that is, the sum of p1 and p2 is one. Therefore, the classification result of whether the collision early warning prompt is generated is actually converted into the classified probability distribution conforming to the natural rule through classifying the tags, and the physical meaning of the natural probability distribution of the tags is essentially used instead of the language text meaning of whether the collision early warning prompt is generated.
It should be appreciated that the role of the classifier is to learn the classification rules and classifier using a given class, known training data, and then classify (or predict) the unknown data. Logistic regression (logistics), SVM, etc. are commonly used to solve the classification problem, and for multi-classification problems (multi-class classification), logistic regression or SVM can be used as well, but multiple bi-classifications are required to compose multiple classifications, but this is error-prone and inefficient, and the commonly used multi-classification method is the Softmax classification function.
Further, in the technical solution of the present application, the automatic driving safety control method based on the deep learning algorithm is characterized by further comprising a training step: and training the vehicle distance time sequence feature extractor comprising the upsampling layer and the one-dimensional convolution layer, the vehicle speed time sequence feature extractor comprising the upsampling layer and the one-dimensional convolution layer, the inter-feature attention layer and the classifier. It should be understood that in the automatic driving safety control method based on the deep learning algorithm, the training step is a very important step, and its function is as follows: 1. parameter learning: the training step is used for learning parameters of each component in the model, including a vehicle distance time sequence feature extractor, a vehicle speed time sequence feature extractor, an inter-feature attention layer, a classifier and the like. Through training, the model can adaptively adjust parameters according to given training data, so that the model can better fit input data, and the performance and accuracy of the model are improved. 2. Feature extractor optimization: the training step can optimize the vehicle distance time sequence feature extractor and the vehicle speed time sequence feature extractor so that the time sequence features of the vehicle distance and the vehicle speed can be better extracted. Through a back propagation algorithm and an optimizer, the training step can adjust parameters of the feature extractor so that the parameters can capture the association information and important modes between vehicles, and therefore the expression capacity and the discrimination of the features are improved. 3. Attention weight learning: the training step is also used to learn weights in the inter-feature attention layer, which are used to adaptively assign attention according to the importance and relevance of the features. Through training, the attention mechanism can learn which features are more important for a specific task, so that the attention and utilization degree of the model on key features is improved. 4. Training a classifier: the training step further comprises training of a classifier for mapping the extracted features to specific safety control decisions. Through training, the classifier can learn the mapping relation between the input characteristics and the safety control decision, so that the safety control of the automatic driving system is realized. In other words, the training step plays a key role in the automatic driving safety control method based on the deep learning algorithm, and the performance and accuracy of the model are improved through the processes of parameter learning, feature extractor optimization, attention weight learning, classifier training and the like, so that more reliable and safe automatic driving control is realized.
Wherein, more specifically, the training step comprises: acquiring training data, wherein the training data comprises training vehicle distance values of a plurality of preset time points in a preset time period, training vehicle speed values of the preset time points and a true value of whether collision early warning prompt is generated or not; passing the training distance values of the plurality of preset time points through the distance time sequence feature extractor comprising the up-sampling layer and the one-dimensional convolution layer to obtain training distance time sequence feature vectors; passing the training vehicle speed values at the plurality of predetermined time points through the vehicle speed time sequence feature extractor comprising an up-sampling layer and a one-dimensional convolution layer to obtain training vehicle speed time sequence feature vectors; performing feature interaction based on an attention mechanism on the training distance time sequence feature vector and the training vehicle speed time sequence feature vector by using the inter-feature attention layer to obtain a training distance-vehicle speed time sequence interaction feature vector; the training vehicle distance-vehicle speed time sequence interaction feature vector passes through a classifier to obtain a classification loss function value; and training the distance time sequence feature extractor comprising the up-sampling layer and the one-dimensional convolution layer, the vehicle speed time sequence feature extractor comprising the up-sampling layer and the one-dimensional convolution layer, the attention layer between features and the classifier by using the classification loss function value, wherein in each round of iteration of training, weight space exploration constraint iteration based on class matrix regularization is carried out on the training distance-vehicle speed time sequence interaction feature vector.
In the technical scheme of the application, the training distance time sequence feature vector and the training vehicle speed time sequence feature vector respectively express time sequence multiscale neighborhood associated features of a training distance value and a training vehicle speed value, so that after feature interaction based on an attention mechanism is carried out on the training distance time sequence feature vector and the training vehicle speed time sequence feature vector by using an attention layer, the training distance-vehicle speed time sequence interaction feature vector comprises the time sequence multiscale neighborhood associated features of the training distance value and the training vehicle speed value, and also comprises the dependency relationship features of the training distance value and the training vehicle speed value in a time sequence distribution direction, namely, the training distance-vehicle speed time sequence interaction feature vector simultaneously comprises feature representations of diversified time sequence associated dimensions corresponding to time sequence association and sample time sequence association in a sample, and therefore, when the training distance-vehicle speed time sequence interaction feature vector carries out time sequence associated feature representation under a cross dimension, the training distance-vehicle speed time sequence interaction feature vector carries out classification and the cross dimension, the probability distribution feature vector in the cross dimension of the cross dimension is enabled to have a classification effect on the classification matrix classification and the classification effect of the cross dimension.
Based on the above, when classifying the training distance-speed time sequence interaction feature vector by the classifier, the applicant of the application performs weight space exploration constraint based on regularization of the class matrix on the training distance-speed time sequence interaction feature vector during each iteration of the weight matrix.
Accordingly, in one specific example, in each iteration of the training, performing a weight space exploration constraint iteration based on class matrix regularization on the training distance-speed time sequence interaction feature vector, including: performing weight space exploration constraint iteration based on class matrix regularization on the training vehicle distance-vehicle speed time sequence interaction feature vector by using the following optimization formula to obtain an optimized training vehicle distance-vehicle speed time sequence interaction feature vector; wherein, the optimization formula is:wherein (1)>Is the time sequence interaction characteristic vector of the training vehicle distance and the vehicle speed, andfor column vector, +.>Representing matrix multiplication +.>Weight matrix representing last iteration, +.>Representing the optimized training distance-speed time sequence interaction characteristic vector +.>Is a row vector, +.>For a learnable domain transfer matrix, for example, the weight matrix can be initially set to the last iteration +.>Diagonal matrix of diagonal elements, +.>Representing the weight matrix after iteration.
Here, consider the weight space domain of the weight matrix and the training distance-speed time sequence interaction feature vectorDomain differences (domain gap) between probability distribution domains of classification results of (2) by weight matrix +.>Time sequence interaction characteristic vector relative to the training vehicle distance-vehicle speed>The regularized representation of the class matrix of (2) is used as an inter-domain migration agent (inter-domain transferring agent) to transfer the probability distribution of valuable label constraint into a weight space, so that excessive exploration (over-explloit) of the weight distribution in the weight space by a rich labeled probability distribution domain in the classification process based on the weight space is avoided, the convergence effect of the weight matrix is improved, and the training effect of the training distance-speed time sequence interaction feature vector in classification regression through a classifier is improved.
In summary, the automatic driving safety control method based on the deep learning algorithm according to the embodiments of the present application is illustrated, which can automatically determine whether to generate a collision early warning prompt in combination with the deep learning algorithm, so as to avoid dangerous situations.
Fig. 4 is a block diagram of an autopilot safety control system 100 based on a deep learning algorithm in accordance with an embodiment of the present application. As shown in fig. 4, the automatic driving safety control system 100 based on the deep learning algorithm according to the embodiment of the present application includes: a distance value acquisition module 110 for acquiring distance values at a plurality of predetermined time points within a predetermined time period acquired by a laser range finder disposed in a vehicle; a speed value acquisition module 120 for acquiring vehicle speed values at the plurality of predetermined time points acquired by a speed sensor disposed at the vehicle; the joint analysis module 130 is configured to perform joint analysis on the distance values at the plurality of predetermined time points and the vehicle speed values at the plurality of predetermined time points to obtain a distance-speed time sequence interaction feature vector; and a prompt confirmation module 140, configured to determine whether to generate a collision early warning prompt based on the distance-speed time sequence interaction feature vector.
In one example, in the deep learning algorithm-based autopilot safety control system 100 described above, the joint analysis module 130 includes: the vehicle distance time sequence feature extraction unit is used for extracting vehicle distance time sequence features of vehicle distance values of the plurality of preset time points to obtain vehicle distance time sequence feature vectors; a vehicle speed time sequence feature extraction unit for extracting vehicle speed time sequence features of the vehicle speed values at a plurality of preset time points to obtain vehicle speed time sequence feature vectors; and a fusion unit for fusing the distance time sequence feature vector and the speed time sequence feature vector to obtain the distance-speed time sequence interaction feature vector.
Here, it will be appreciated by those skilled in the art that the specific functions and operations of the respective modules in the above-described deep learning algorithm-based automatic driving safety control system 100 have been described in detail in the above description of the deep learning algorithm-based automatic driving safety control method with reference to fig. 1 to 3, and thus, repetitive descriptions thereof will be omitted.
As described above, the deep learning algorithm-based automatic driving safety control system 100 according to the embodiment of the present application may be implemented in various wireless terminals, for example, a server or the like having the deep learning algorithm-based automatic driving safety control algorithm. In one example, the deep learning algorithm-based autopilot safety control system 100 according to embodiments of the present application may be integrated into a wireless terminal as one software module and/or hardware module. For example, the deep learning algorithm-based autopilot safety control system 100 may be a software module in the operating system of the wireless terminal, or may be an application developed for the wireless terminal; of course, the deep learning algorithm-based autopilot safety control system 100 may also be one of a number of hardware modules of the wireless terminal.
Alternatively, in another example, the deep learning algorithm-based autopilot safety control system 100 and the wireless terminal may be separate devices, and the deep learning algorithm-based autopilot safety control system 100 may be connected to the wireless terminal through a wired and/or wireless network and transmit interactive information in a agreed data format.
Fig. 5 is an application scenario diagram of an automatic driving safety control method based on a deep learning algorithm according to an embodiment of the present application. As shown in fig. 5, in this application scenario, first, vehicle distance values (e.g., D1 illustrated in fig. 5) at a plurality of predetermined time points within a predetermined time period acquired by a laser range finder (e.g., C1 illustrated in fig. 5) disposed in a vehicle are acquired; and a vehicle speed value (e.g., D2 illustrated in fig. 5) at the plurality of predetermined time points acquired by a speed sensor (e.g., C2 illustrated in fig. 5) disposed in the vehicle, and then inputting the vehicle distance values at the plurality of predetermined time points and the vehicle speed value at the plurality of predetermined time points into a server (e.g., S illustrated in fig. 5) disposed with an automatic driving safety control algorithm based on a deep learning algorithm, wherein the server is capable of processing the vehicle distance values at the plurality of predetermined time points and the vehicle speed value at the plurality of predetermined time points using the automatic driving safety control algorithm based on the deep learning algorithm to obtain a classification result for indicating whether or not to generate a collision warning prompt.
Furthermore, those skilled in the art will appreciate that the various aspects of the invention are illustrated and described in the context of a number of patentable categories or circumstances, including any novel and useful procedures, machines, products, or materials, or any novel and useful modifications thereof. Accordingly, aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, micro-code, etc.) or by a combination of hardware and software. The above hardware or software may be referred to as a "data block," module, "" engine, "" unit, "" component, "or" system. Furthermore, aspects of the present application may take the form of a computer product, comprising computer-readable program code, embodied in one or more computer-readable media.
Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The foregoing is illustrative of the present invention and is not to be construed as limiting thereof. Although a few exemplary embodiments of this invention have been described, those skilled in the art will readily appreciate that many modifications are possible in the exemplary embodiments without materially departing from the novel teachings and advantages of this invention. Accordingly, all such modifications are intended to be included within the scope of this invention as defined in the following claims. It is to be understood that the foregoing is illustrative of the present invention and is not to be construed as limited to the specific embodiments disclosed, and that modifications to the disclosed embodiments, as well as other embodiments, are intended to be included within the scope of the appended claims. The invention is defined by the claims and their equivalents.
Claims (8)
1. An automatic driving safety control method based on a deep learning algorithm is characterized by comprising the following steps:
acquiring vehicle distance values of a plurality of preset time points in a preset time period acquired by a laser range finder arranged on a vehicle;
acquiring vehicle speed values at the plurality of predetermined time points acquired by speed sensors disposed at the vehicle;
performing joint analysis on the vehicle distance values of the plurality of preset time points and the vehicle speed values of the plurality of preset time points to obtain a vehicle distance-vehicle speed time sequence interaction feature vector; and
determining whether collision early warning prompt is generated or not based on the vehicle distance-vehicle speed time sequence interaction feature vector;
the training method further comprises the following training steps: training a vehicle distance time sequence feature extractor comprising an up-sampling layer and a one-dimensional convolution layer, a vehicle speed time sequence feature extractor comprising an up-sampling layer and a one-dimensional convolution layer, an inter-feature attention layer and a classifier;
wherein the training step comprises:
acquiring training data, wherein the training data comprises training vehicle distance values of a plurality of preset time points in a preset time period, training vehicle speed values of the preset time points and a true value of whether collision early warning prompt is generated or not;
passing the training distance values of the plurality of preset time points through the distance time sequence feature extractor comprising the up-sampling layer and the one-dimensional convolution layer to obtain training distance time sequence feature vectors;
passing the training vehicle speed values at the plurality of predetermined time points through the vehicle speed time sequence feature extractor comprising an up-sampling layer and a one-dimensional convolution layer to obtain training vehicle speed time sequence feature vectors;
performing feature interaction based on an attention mechanism on the training distance time sequence feature vector and the training vehicle speed time sequence feature vector by using the inter-feature attention layer to obtain a training distance-vehicle speed time sequence interaction feature vector;
the training vehicle distance-vehicle speed time sequence interaction feature vector passes through a classifier to obtain a classification loss function value; and
training the distance time sequence feature extractor comprising an up-sampling layer and a one-dimensional convolution layer, the vehicle speed time sequence feature extractor comprising an up-sampling layer and a one-dimensional convolution layer, the attention layer between features and the classifier by using the classification loss function value, wherein in each round of iteration of the training, a weight space exploration constraint iteration based on regularization of a class matrix is carried out on the training distance-vehicle speed time sequence interaction feature vector;
in each iteration of the training, performing a weight space exploration constraint iteration based on class matrix regularization on the training vehicle distance-vehicle speed time sequence interaction feature vector, wherein the method comprises the following steps of:
performing weight space exploration constraint iteration based on class matrix regularization on the training vehicle distance-vehicle speed time sequence interaction feature vector by using the following optimization formula to obtain an optimized training vehicle distance-vehicle speed time sequence interaction feature vector;
wherein, the optimization formula is:
wherein (1)>Is the training distance-speed time sequence interaction characteristic vector and +.>For column vector, +.>Representing matrix multiplication +.>Weight matrix representing last iteration, +.>Representing the optimized training distance-speed time sequence interaction characteristic vector +.>Is a row vector, +.>Is a domain transfer matrix which can be learned, +.>Representing the weight matrix after iteration.
2. The deep learning algorithm-based automatic driving safety control method according to claim 1, wherein performing joint analysis on the vehicle distance values at the plurality of predetermined time points and the vehicle speed values at the plurality of predetermined time points to obtain a vehicle distance-vehicle speed time sequence interaction feature vector comprises:
extracting the distance time sequence characteristics of the distance values of the plurality of preset time points to obtain distance time sequence characteristic vectors;
extracting vehicle speed time sequence characteristics of the vehicle speed values at a plurality of preset time points to obtain a vehicle speed time sequence characteristic vector; and
and fusing the vehicle distance time sequence feature vector and the vehicle speed time sequence feature vector to obtain the vehicle distance-vehicle speed time sequence interaction feature vector.
3. The automatic driving safety control method based on the deep learning algorithm according to claim 2, wherein extracting the distance time sequence features of the distance values at the plurality of predetermined time points to obtain the distance time sequence feature vector comprises:
and passing the distance values of the plurality of preset time points through a distance time sequence feature extractor comprising an upsampling layer and a one-dimensional convolution layer to obtain the distance time sequence feature vector.
4. The automatic driving safety control method based on the deep learning algorithm according to claim 3, wherein extracting vehicle speed time series characteristics of the vehicle speed values at the plurality of predetermined time points to obtain a vehicle speed time series characteristic vector comprises:
and passing the vehicle speed values at the plurality of preset time points through a vehicle speed time sequence feature extractor comprising an upsampling layer and a one-dimensional convolution layer to obtain the vehicle speed time sequence feature vector.
5. The deep learning algorithm-based automatic driving safety control method according to claim 4, wherein fusing the distance-from-vehicle timing feature vector and the vehicle speed timing feature vector to obtain the distance-from-vehicle speed timing interaction feature vector, comprises:
and performing feature interaction based on an attention mechanism on the distance time sequence feature vector and the speed time sequence feature vector by using an inter-feature attention layer to obtain the distance-speed time sequence interaction feature vector.
6. The deep learning algorithm-based automatic driving safety control method according to claim 5, wherein determining whether to generate a collision warning prompt based on the inter-vehicle distance-vehicle speed timing interaction feature vector comprises:
and the vehicle distance-vehicle speed time sequence interaction feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether collision early warning prompt is generated or not.
7. An automatic driving safety control system based on a deep learning algorithm, using the automatic driving safety control method based on a deep learning algorithm as claimed in claim 1, comprising:
the vehicle distance value acquisition module is used for acquiring vehicle distance values of a plurality of preset time points in a preset time period acquired by a laser range finder arranged on a vehicle;
a speed value acquisition module for acquiring vehicle speed values at the plurality of predetermined time points acquired by a speed sensor disposed at the vehicle;
the joint analysis module is used for performing joint analysis on the vehicle distance values of the plurality of preset time points and the vehicle speed values of the plurality of preset time points to obtain a vehicle distance-vehicle speed time sequence interaction feature vector; and
and the prompt confirmation module is used for determining whether collision early warning prompts are generated or not based on the vehicle distance-vehicle speed time sequence interaction feature vector.
8. The deep learning algorithm-based autopilot safety control system of claim 7 wherein the joint analysis module comprises:
the vehicle distance time sequence feature extraction unit is used for extracting vehicle distance time sequence features of vehicle distance values of the plurality of preset time points to obtain vehicle distance time sequence feature vectors;
a vehicle speed time sequence feature extraction unit for extracting vehicle speed time sequence features of the vehicle speed values at a plurality of preset time points to obtain vehicle speed time sequence feature vectors; and
and the fusion unit is used for fusing the vehicle distance time sequence feature vector and the vehicle speed time sequence feature vector to obtain the vehicle distance-vehicle speed time sequence interaction feature vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311248489.XA CN116985793B (en) | 2023-09-26 | 2023-09-26 | Automatic driving safety control system and method based on deep learning algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311248489.XA CN116985793B (en) | 2023-09-26 | 2023-09-26 | Automatic driving safety control system and method based on deep learning algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116985793A CN116985793A (en) | 2023-11-03 |
| CN116985793B true CN116985793B (en) | 2023-12-22 |
Family
ID=88525167
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311248489.XA Active CN116985793B (en) | 2023-09-26 | 2023-09-26 | Automatic driving safety control system and method based on deep learning algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116985793B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117429419B (en) * | 2023-09-13 | 2024-05-10 | 江苏大块头智驾科技有限公司 | Automatic driving method applied to port and driving vehicle |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102287065B1 (en) * | 2020-02-28 | 2021-08-05 | 아주대학교산학협력단 | Crash probability prediction method and apparatus for vehicle using tda technique |
| CN116135640A (en) * | 2021-11-18 | 2023-05-19 | 广州汽车集团股份有限公司 | Anti-collision early warning method and system for vehicle and vehicle |
| CN116279504A (en) * | 2023-03-29 | 2023-06-23 | 重庆赛力斯新能源汽车设计院有限公司 | AR-based vehicle speed assist system and method thereof |
| CN116383685A (en) * | 2023-03-09 | 2023-07-04 | 浙江工业大学 | Vehicle lane change detection method based on space-time interaction diagram attention network |
-
2023
- 2023-09-26 CN CN202311248489.XA patent/CN116985793B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102287065B1 (en) * | 2020-02-28 | 2021-08-05 | 아주대학교산학협력단 | Crash probability prediction method and apparatus for vehicle using tda technique |
| CN116135640A (en) * | 2021-11-18 | 2023-05-19 | 广州汽车集团股份有限公司 | Anti-collision early warning method and system for vehicle and vehicle |
| CN116383685A (en) * | 2023-03-09 | 2023-07-04 | 浙江工业大学 | Vehicle lane change detection method based on space-time interaction diagram attention network |
| CN116279504A (en) * | 2023-03-29 | 2023-06-23 | 重庆赛力斯新能源汽车设计院有限公司 | AR-based vehicle speed assist system and method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116985793A (en) | 2023-11-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gu et al. | A novel lane-changing decision model for autonomous vehicles based on deep autoencoder network and XGBoost | |
| WO2018121690A1 (en) | Object attribute detection method and device, neural network training method and device, and regional detection method and device | |
| CN116985793B (en) | Automatic driving safety control system and method based on deep learning algorithm | |
| Liu et al. | DSDCLA: Driving style detection via hybrid CNN-LSTM with multi-level attention fusion | |
| CN112036435A (en) | Brushless direct current motor sensor fault detection method based on convolutional neural network | |
| CN116767204B (en) | Vehicle speed control method and system based on vehicle load | |
| CN117104377B (en) | Intelligent management system and method for electric bicycle | |
| CN116930042B (en) | Building waterproof material performance detection equipment and method | |
| CN116934304A (en) | Intelligent power distribution room equipment operation maintenance management system and method thereof | |
| CN116872961B (en) | Control system for intelligent driving vehicle | |
| Aksoy et al. | See, attend and brake: An attention-based saliency map prediction model for end-to-end driving | |
| CN117676099B (en) | Security early warning method and system based on Internet of things | |
| CN117784710B (en) | Remote state monitoring system and method for numerical control machine tool | |
| Westny et al. | Vehicle behavior prediction and generalization using imbalanced learning techniques | |
| CN116777183B (en) | Unmanned ship cluster intelligent scheduling method and system | |
| Haque et al. | Driving maneuver classification from time series data: a rule based machine learning approach | |
| Albahri et al. | Fuzzy decision-making framework for explainable golden multi-machine learning models for real-time adversarial attack detection in Vehicular Ad-hoc Networks | |
| Fayyad | Out-of-distribution detection using inter-level features of deep neural networks | |
| Priya et al. | Vehicle Detection in Autonomous Vehicles Using Computer Vision Check for updates | |
| De Arriba López et al. | Development of a deep learning model for recognising traffic sings focused on difficult cases | |
| Kaur et al. | Pedestrian and Vehicle detection in automotive embedded systems using deep neural networks | |
| Sun et al. | Intent-Aware Conditional Generative Adversarial Network for Pedestrian Path Prediction | |
| CN117429419B (en) | Automatic driving method applied to port and driving vehicle | |
| Prasad | Pattern recognition: possible research areas and issues | |
| US20240212361A1 (en) | Method and apparatus for determining driving risks by using deep learning algorithms |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |