CN116386387A - Method and device for predicting following behavior of driving vehicle of hybrid queue person - Google Patents
Method and device for predicting following behavior of driving vehicle of hybrid queue person Download PDFInfo
- Publication number
- CN116386387A CN116386387A CN202310422466.XA CN202310422466A CN116386387A CN 116386387 A CN116386387 A CN 116386387A CN 202310422466 A CN202310422466 A CN 202310422466A CN 116386387 A CN116386387 A CN 116386387A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- driving
- queue
- following behavior
- sub
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 230000001133 acceleration Effects 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 27
- 238000004590 computer program Methods 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 6
- 230000006399 behavior Effects 0.000 description 98
- 230000006870 function Effects 0.000 description 13
- 238000004891 communication Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 12
- 230000008569 process Effects 0.000 description 10
- 238000013528 artificial neural network Methods 0.000 description 8
- 210000002569 neuron Anatomy 0.000 description 7
- 230000015654 memory Effects 0.000 description 5
- 230000008447 perception Effects 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000006855 networking Effects 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 230000004913 activation Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 208000037262 Hepatitis delta Diseases 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 125000004122 cyclic group Chemical group 0.000 description 2
- 208000029570 hepatitis D virus infection Diseases 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 238000013076 uncertainty analysis Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 241000724709 Hepatitis delta virus Species 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000035939 shock Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000011144 upstream manufacturing Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/16—Anti-collision systems
- G08G1/166—Anti-collision systems for active traffic, e.g. moving vehicles, pedestrians, bikes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Traffic Control Systems (AREA)
- Hybrid Electric Vehicles (AREA)
Abstract
The invention relates to the technical field of vehicle control, in particular to a method and a device for predicting the following behavior of a driving vehicle of a hybrid queue person. Firstly, acquiring driving state information of a first person driving vehicle and a last person driving vehicle in a to-be-predicted person driving vehicle sub-queue in a mixed queue in a current time period; and then, the driving state information of the first and last driving vehicles in the current time period is input into a trained vehicle following behavior prediction model to obtain the predicted driving state information of the last driving vehicle in the future time period of the sub-queue of the driving vehicles to be predicted. According to the method and the device, the driving state information of the human driving vehicle in the future time period can be accurately and rapidly predicted by adopting the vehicle following behavior prediction model, reliable data basis is provided for traffic system control, and the safety and the high efficiency of road traffic control are ensured.
Description
Technical Field
The invention relates to the technical field of vehicle control, in particular to a method and a device for predicting the following behavior of a driving vehicle of a hybrid queue person.
Background
With the rapid development of world economy, the demands for transportation industry, in particular road transportation, are continuously increasing. The huge traffic demand makes the highway traffic of key areas in a state of supply and demand for a long time, and the road traffic runs close to the maximum traffic capacity for a long time. In this state, any small disturbance is very likely to cause traffic shock and propagation, resulting in traffic congestion. Along with the continuous development of the internet of vehicles and intelligent network car-connection technologies, the real-time sensing capability, data sharing and accurate control capability of road sides and vehicle-mounted equipment are continuously improved, and the queue control technology of the intelligent network car-connection is used for carrying out cooperative control based on the real-time shared sensing data, so that the efficient and stable operation of queues and traffic flows is ensured, and the overall efficiency, safety and comfort of a road traffic system are improved.
However, due to the huge conservation amount of the traditional automobile, the coexistence period of the intelligent network car and the traditional driving car is inevitably long in the popularization process of the car networking and intelligent network car technologies. In a mixed traffic flow in a car networking environment, an intelligent network vehicle (CAV) and a conventional Human-driven vehicle (HDV) are dispersed in a queue, and a long queue is divided into a plurality of intelligent network vehicle sub-queues and Human-driven vehicle sub-queues. Under the condition that the road side equipment cannot realize holographic perception, the perception network composed of intelligent network coupling and part of road side equipment can only carry out limited perception on the driving vehicles of people. At the same time, when the front/rear vehicles are intelligent network linked vehicles, the following behavior of the driver and the uncertainty of the driver can be influenced. In this case, uncertainty of human behavior may cause a great deal of uncertainty and random disturbance in the control process, so that the control efficiency of the conventional means is reduced, and it is difficult to ensure the stability of traffic flow. Therefore, the hybrid queue for the driving vehicles and the intelligent network vehicle connection is a necessary requirement for improving the efficiency of the traffic system, and the short-term following behavior of the driving vehicles and the uncertainty in the queue are predicted. Therefore, how to accurately predict the following behavior of a human driving vehicle in a hybrid queue is a technical problem to be solved.
Disclosure of Invention
The invention mainly solves the technical problem of accurately predicting the following behavior of the driving vehicles in the mixed queue.
In one aspect, the invention provides a method for predicting the following behavior of a driving vehicle of a hybrid queue person, which comprises the following steps:
acquiring driving state information of a first person driving vehicle and a last person driving vehicle in a to-be-predicted person driving vehicle sub-queue in a mixed queue in a current time period;
and inputting the driving state information of the first and last driving vehicles in the current time period into a trained vehicle following behavior prediction model to obtain the driving state information of the last driving vehicle in the future time period of the sub-queue of the driving vehicles to be predicted.
In one embodiment, the vehicle following behavior prediction model is obtained through training by the following method:
acquiring a plurality of sub-queues of the driving vehicles;
for each of the human-driven vehicle sub-queues, performing the following processing:
acquiring running state information of a first person driving vehicle and a last person driving vehicle in the person driving vehicle sub-queue in a last time period as input sample data;
acquiring driving state information of the last vehicle in the sub-queue of the driving vehicle of the person in the next time period as output sample data;
and training an initial vehicle following behavior prediction model by adopting the input sample data and the output sample data to obtain the trained vehicle following behavior prediction model.
In one embodiment, after obtaining the plurality of personal drive vehicle sub-queues, further comprising:
and processing the plurality of man-driving vehicle subqueues, and removing the man-driving vehicle subqueues with the sample size smaller than a preset value and the man-driving vehicle subqueues which do not accord with the following condition.
In one embodiment, the obtaining a plurality of personal drive vehicle sub-queues includes:
and when the sub-queues of the driving vehicles are obtained, removing the sub-queues of the driving vehicles with the conditions of lane changing, lane entrance and exit or too short lanes.
In one embodiment, the vehicle following behavior prediction model includes an encoder and a decoder;
the step of inputting the driving state information of the first and last vehicles in the current time period into a trained vehicle following behavior prediction model to obtain the driving state information of the last vehicle in the sub-queue of the to-be-predicted vehicles in the future time period, including:
the encoder converts the input driving state information of the first driving vehicle and the last driving vehicle in the current time period into a state vector with a fixed length;
the decoder decodes the fixed-length state vector into the predicted running state information.
In one embodiment, training an initial vehicle following behavior prediction model using the input sample data and the output sample data to obtain the trained vehicle following behavior prediction model includes:
the loss function is set as follows:
where loss represents a training function, t 2 Representing the predicted time step, t representing the current time,
indicating predicted acceleration +.>
Indicating the actual acceleration of the vehicle, I omega I 2 Representing a regularization term;
and determining whether the value of the loss function is smaller than a preset value, if so, determining that training is completed, and obtaining the trained vehicle following behavior prediction model.
In one embodiment, the vehicle-following behavior prediction model is a multilayer LSTM model;
the two LSTM layers comprise a dropout layer, and the setting value of the LSTM layer is 0.5.
On the other hand, the invention also provides a device for predicting the following behavior of the driving vehicle of the hybrid queue person, which is characterized by comprising the following components:
the system comprises an acquisition unit, a prediction unit and a control unit, wherein the acquisition unit is used for acquiring driving state information of a first person driving vehicle and a last person driving vehicle in a to-be-predicted person driving vehicle sub-queue in a mixed queue in a current time period;
the prediction unit is used for inputting the driving state information of the first and last driving vehicles in the current time period into a trained vehicle following behavior prediction model to obtain the predicted driving state information of the last driving vehicle in the future time period of the sub-queue of the driving vehicles to be predicted.
In another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method for predicting the heel-and-toe behavior of a hybrid passenger according to any one of the above.
In another aspect, the present invention also provides a computer program product comprising a computer program which when executed by a processor implements a method of predicting the following behavior of a hybrid passenger vehicle according to any one of the preceding claims.
According to the method for predicting the following behavior of the hybrid queue human-driven vehicle, firstly, the driving state information of a first human-driven vehicle and a last human-driven vehicle in a current time period of a sub-queue of the human-driven vehicles to be predicted in the hybrid queue is obtained; and then, the driving state information of the first and last driving vehicles in the current time period is input into a trained vehicle following behavior prediction model to obtain the predicted driving state information of the last driving vehicle in the future time period of the sub-queue of the driving vehicles to be predicted. According to the method and the device, the driving state information of the human driving vehicle in the future time period can be accurately and rapidly predicted by adopting the vehicle following behavior prediction model, reliable data basis is provided for traffic system control, and the safety and the high efficiency of road traffic control are ensured.
Drawings
FIG. 1 is a flowchart of a method for predicting the following behavior of a hybrid-queue human-driven vehicle according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a hybrid queue model according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a process for establishing a sub-queue of a person driving a vehicle under a mixed traffic flow according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a prediction of a driving vehicle of a person according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a vehicle following behavior prediction model according to an embodiment of the present invention;
FIG. 6 is a comparison of predicted and actual values provided by an embodiment of the present invention;
FIG. 7 is a schematic structural diagram of a device for predicting the following behavior of a hybrid-queue human-driven vehicle according to an embodiment of the present invention;
fig. 8 is a schematic entity structure diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The invention will be described in further detail below with reference to the drawings by means of specific embodiments. Wherein like elements in different embodiments are numbered alike in association. In the following embodiments, numerous specific details are set forth in order to provide a better understanding of the present application. However, one skilled in the art will readily recognize that some of the features may be omitted, or replaced by other elements, materials, or methods in different situations. In some instances, some operations associated with the present application have not been shown or described in the specification to avoid obscuring the core portions of the present application, and may not be necessary for a person skilled in the art to describe in detail the relevant operations based on the description herein and the general knowledge of one skilled in the art.
Furthermore, the described features, operations, or characteristics of the description may be combined in any suitable manner in various embodiments. Also, various steps or acts in the method descriptions may be interchanged or modified in a manner apparent to those of ordinary skill in the art. Thus, the various orders in the description and drawings are for clarity of description of only certain embodiments, and are not meant to be required orders unless otherwise indicated.
In the mixed traffic flow in the car networking environment, the intelligent network train and the man-driven vehicles divide the long queue into a plurality of intelligent network train sub-queues and man-driven vehicle sub-queues. The intelligent network vehicles can mutually sense through the vehicle-mounted sensor, and the following behavior of the vehicle in the sensing range is known. However, the driving vehicles do not have sensors, the following behavior of the adjacent driving vehicles and the number of vehicles in the sub-queues of the driving vehicles can not be known, and only the running state of the HDV adjacent to the CAV can be obtained.
With the development of artificial intelligence technology in recent years, machine learning and deep learning algorithms are continuously enriched, and data-driven relaxation-following models are rapidly developed. The method utilizes deep learning network models such as a cyclic neural network (RNN, recurrent Neural Network), a gate-controlled cyclic unit (GRU, gate Recurrent Unit), a Long Short-Term Memory neural network (LSTM), and the like to construct a model with a complex hidden layer, and focuses on mining hidden internal rules in real vehicle driving data, so that accuracy of following behavior prediction is improved.
Specifically, based on a deep learning algorithm, the application provides a method for predicting the following behavior of a driving vehicle of a hybrid queue, which comprises the steps of firstly obtaining driving state information of a first driving vehicle and a last driving vehicle of a sub-queue of the driving vehicles of the to-be-predicted persons in the hybrid queue in a current time period; and then, the driving state information of the first and last vehicles in the current time period is input into a trained vehicle following behavior prediction model to obtain the driving state information of the last vehicle in the future time period of the sub-queue of the to-be-predicted vehicles.
Furthermore, in order to enable the prediction result to have better robustness so as to be more accurate for predicting the driving vehicles in the sub-queues of the driving vehicles, the uncertainty of the vehicle following behavior prediction model is considered, namely a Dropout layer is added in the vehicle following behavior prediction model, so that the vehicle following behavior prediction model is not sensitive to certain specific characteristics any more, the model robustness is improved, and meanwhile uncertainty analysis on the prediction result can be realized.
The following description will be made in detail on the following method for predicting the heel-and-toe behavior of a hybrid passenger car according to several specific embodiments of the present invention. It is to be understood that the following embodiments may be combined with each other and that some embodiments may not be repeated for the same or similar concepts or processes.
Fig. 1 is a flowchart of a method for predicting the following behavior of a hybrid-queue human-driven vehicle according to an embodiment of the present invention, where the method for predicting the following behavior of a human-driven vehicle may be executed by a software and/or hardware device. Referring to fig. 1, the method for predicting the following behavior of a driving vehicle of a hybrid train comprises the following steps:
s101, acquiring driving state information of a first person driving vehicle and a last person driving vehicle of a to-be-predicted person driving vehicle sub-queue in the mixed queue in a current time period.
In one embodiment, the traffic detection system may obtain the driving state information of the first and last vehicles in the sub-queues of the to-be-predicted vehicles in the mixed queue in the current time period.
In an embodiment, fig. 2 is a schematic diagram of a hybrid queue model provided in an embodiment of the present invention, please refer to fig. 2, in which the position, speed and acceleration information of a first driving vehicle (HDV 1) and a last driving vehicle (HDVn) in a sub-queue of driving vehicles are used as input data of a vehicle following behavior prediction model, and the acceleration information of the last driving vehicle is used as output data of the vehicle following behavior prediction model, so as to analyze the following behavior of a driver. And then, according to the existing vehicle state and analysis model, predicting the acceleration of the last person driving vehicle in the queue in a short period, and converting the predicted acceleration into the position and the speed by utilizing a kinematic formula to obtain the following state of the last person driving vehicle in the short period.
In one embodiment, the travel state information may generally include current position, speed, acceleration, and the like.
S102, the driving state information of the first and last vehicles in the current time period is input into a trained vehicle following behavior prediction model, and the driving state information of the last vehicle in the future time period of the sub-queue of the to-be-predicted vehicles is obtained.
In one embodiment, fig. 4 is a schematic diagram of prediction of a person driving vehicle provided in the embodiment of the present invention, and please refer to fig. 4, in which positions, speeds and accelerations of a first person driving vehicle and a last person driving vehicle in a sub-queue of the person driving vehicles to be predicted in a current time period are input into a trained vehicle following behavior prediction model, so as to obtain positions, speeds and accelerations of the last person driving vehicle in the sub-queue of the person driving vehicles to be predicted in a future time period, and provide data support for queue cooperative control in a subsequent mixed flow environment.
In one embodiment, the vehicle following behavior prediction model is obtained by training in the following way:
acquiring a plurality of sub-queues of the driving vehicles;
for each individual driving vehicle sub-queue, the following processing is performed:
acquiring running state information of a first person driving vehicle and a last person driving vehicle in a person driving vehicle sub-queue in a last time period as input sample data;
acquiring running state information of the last vehicle in the sub-queue of the human driving vehicle in the next time period as output sample data;
and training an initial vehicle following behavior prediction model by adopting the input sample data and the output sample data to obtain a trained vehicle following behavior prediction model.
In an embodiment, when data of a plurality of sub-queues of driving vehicles are acquired, fig. 3 is a schematic diagram of a process for establishing a sub-queue of driving vehicles under a mixed traffic flow provided in an embodiment of the present invention, and please refer to fig. 3, which is shown in the following description.
Referring to fig. 3, in the hybrid queue shown in fig. 3, since the conventional driving vehicle HDV does not have an on-vehicle sensor and an on-vehicle communication device, the conventional driving vehicle HDV cannot acquire the running state of the peripheral HDV, and therefore the conventional driving vehicle HDV cannot acquire how many HDVs are in the current sub-queue.
However, in the hybrid train, the control of the intelligent network train (such as the CAV sub-train in fig. 2) is affected by the running state of the upstream driving vehicle (i.e., the HDVn in fig. 3), so that in this scenario, the running state of the last driving vehicle in the driving vehicle sub-train needs to be predicted.
Meanwhile, in the actual prediction process, the motion states of the first and last person driving vehicles in the HDV subqueue cannot be obtained, so that the motion state information of only the two vehicles is input in the embodiment. When the prediction method of the embodiment is adopted, the specific number of vehicles in the HDV sub-queue is not required to be considered, and only the track and the motion state of the last person driving vehicle are predicted through the perceived motion relationship between the first person driving vehicle and the last person driving vehicle.
In one embodiment, when data of a plurality of sub-queues of the driving vehicles are acquired, an I-80 dataset in a NGSIM (Next Generation Simulation) dataset is used as vehicle following behavior prediction model construction and verification data, then the dataset is processed, sub-queue data which has smaller sample size and does not accord with the following condition is removed, and initial data for training the vehicle following behavior prediction model is formed. The NGSIM data sets describe interactions between intermodal passengers, vehicles, and highway systems, as well as the interaction devices, profiles, congestion, and other environmental features provided to them by traffic control. The I-80 dataset is a relatively high quality and detailed dataset collected under NGSIM effort that can support algorithms that develop driver behavior on a microscopic level. The invention analyzes the relatively common following behavior of a person driving a vehicle, and avoids the fluctuation of the data set caused by the conditions of lane changing, lane entrance and exit, too short lane and the like as far as possible in the data set selection. Therefore, the initial data set (i.e., training data set) of the present embodiment is processed with the I-80 data set trajectors-0400-0415 time period, vehicle_id (ID of current Vehicle) in lane 2, precedent (ID of current Vehicle), frame_id (sampling time), local_y (Vehicle position), v_vel (Vehicle speed), v_acc (Vehicle acceleration) data.
In the mixed flow queue, the human-driven vehicles do not have sensors, so that related data cannot be perceived, and the human-driven vehicles HDV real-time perception data (position, speed and acceleration) close to the CAV can be obtained only through the CAV vehicle-mounted sensors. However, the short-range communication equipment configured by the intelligent network vehicle has a certain communication range, and the vehicle distance for vehicle-to-vehicle communication should be kept within the communication range, and then the position interval of the HDV sub-queues in the middle of the two CAV queues should be smaller than the communication range. In this embodiment, the position difference between the current vehicle and the preceding vehicle is calculated, the current vehicle and the preceding vehicle, the position difference of which is smaller than the sensing range of the sensor, form the last person driving vehicle and the first person driving vehicle of the HDV sub-queue, and the vehicle position difference is calculated iteratively to establish all possible HDV sub-queues.
In one embodiment, after the plurality of sub-queues of the driving vehicles are obtained, the plurality of sub-queues of the driving vehicles are further processed, and the sub-queues of the driving vehicles with the sample size smaller than a preset value and the sub-queues of the driving vehicles not conforming to the following condition are removed. For example, counting the number of data pairs of each 'HDVn-HDV 1', removing the data pairs with the number smaller than 150 and larger variation, for example, removing the data pairs with the headway longer than 4 seconds, wherein the headway refers to the time required for driving the distance between the heads of two vehicles according to the set speed, and avoiding the prediction error caused by the fact that the data quantity is too small or the vehicles are separated from the following behavior.
The pre-processed human drive vehicle sub-queue data is a simple time series data that needs to be converted into supervised learning data with input and output components. In this embodiment, the position, speed and acceleration of the first HDV driving vehicle and the last HDV driving vehicle at 10s (i.e., the first 100 sampling times and the sampling interval of 0.1 second) before the current time are taken as input sample data, and the acceleration of the last HDV driving vehicle at 3s (i.e., the last 30 sampling times) is taken as predicted output sample data. And training an initial vehicle following behavior prediction model by adopting the input sample data and the output sample data to obtain a trained vehicle following behavior prediction model.
When the vehicle following behavior prediction model is trained, since the following behavior prediction is a regression problem, a mean square error is adopted as a loss function of the model in the process of fitting the vehicle following behavior prediction model. In order to prevent overfitting, an L2 regularization term is added into the loss function, so that weight attenuation in model training is effectively realized. The loss function after adding the L2 regularization term is as follows:
where loss represents a training function, t 2 Representing the predicted time step, t representing the current time,
indicating predicted acceleration +.>
Indicating the actual acceleration of the vehicle, I omega I 2 Representing regularization terms.
In one embodiment, 70% of the sample data is used as a training set and 30% of the sample data is used as a validation set to fit and train the vehicle following behavior prediction model. And determining that the training is completed until the loss values of the training set and the verification set are small and within an acceptable range, and obtaining a trained vehicle following behavior prediction model.
In one embodiment, fig. 5 is a schematic structural diagram of a vehicle following behavior prediction model provided by the embodiment of the present invention, where as shown in fig. 5, the vehicle following behavior prediction model is a Decoder-Encoder LSTM model, and the model uses a tensorflow and a Keras framework as a basis, and uses a Sequential model in Keras to stack some network layers through an add () interface, so as to construct a complete vehicle following behavior prediction model according to a stacking sequence.
In one embodiment, a vehicle following behavior prediction model includes an encoder and a decoder; position speed and acceleration data of first and last person driving vehicles in HDV queue
As the running state information in the current time period to be input in a predicting way, the encoder converts the running state information of the first and last person driving vehicles in the current time period into a fixed-length state vector; (i.e., the encoder processes the input sequence into a state vector and a hidden state, providing a fixed-length output vector), the decoder decodes the fixed-length state vector into the last human-driven vehicle in the futureThe driving state information in the time period, specifically, the driving state information of the last person driving the vehicle in the future time period may be acceleration +.>
In one embodiment, the vehicle following behavior prediction model may be a multi-layer LSTM model, and a dropout layer is included in two LSTM layers, where the LSTM layer has a setting value of 0.5.
Specifically, the vehicle following behavior prediction model of the present embodiment is designed as follows:
(1) Encoder LSTM. LSTM needs to explicitly set the input_shape attribute, set the input time step to sample time 100, and the feature dimension to 6. 128 neurons were set in the first hidden layer, tanh was used as the activation function, and the weight decay rate was set to 0.0002. Each time step of the LSTM encoder outputs a hidden vector, but only the last hidden state is output to the next layer for decoding by the decoder.
(2) Repeat vector layer (repeat input n times). The parameter value for this layer is set to output the time step number 30 and the hidden vector received from the LSTM is repeated 30 times, with each time step having the same vector.
(3) Dropout layer. The dropout layer parameter value was set to 0.5. In the training model stage, each neuron has 50% probability of terminating the work, namely, the probability of setting the activation function value to be 0 is 50%, in this case, any two neurons can not necessarily appear in the same dropout network each time, so that the updating of the weight value is not dependent on the coaction of hidden nodes with fixed relations, the neural network is not sensitive to certain specific characteristics, and the model robustness is improved.
(4) Decoder LSTM. The LSTM layer construction continues in the decoder, but differs from the encoding layer in that the return_sequences attribute value is set to True so that each time step has an output vector and is passed to the next layer for prediction in the next fully-connected layer.
(5) And (5) a full connection layer. The full-connection layer sets the neuron number as the characteristic dimension 1 of the output vector, namely the acceleration of the last human-driven vehicle of the HDV. Each time step in the LSTM decoder outputs a vector, and the layer uses a fully-concatenated layer at each time step, so that only one acceleration amount is predicted per time step. And calling a TimeDistributed layer developed by Keras, and applying the same full connection layer to each output time step to obtain the output of the multi-time-step prediction result.
The parameters of each layer of the vehicle following behavior prediction model of the present embodiment are shown in the following table 1:
TABLE 1 vehicle following behavior prediction model parameters for each layer
The method for predicting the following behavior of the hybrid-queue human-driven vehicle provided by the embodiment considers short-term uncertainty prediction analysis. The basic LSTM model does not actually take into account the specific value of the prediction uncertainty, which in this embodiment compensates for this technical deficiency. For the uncertainty of the deep learning model, parameters such as initial value, weight and the like of the model are fixed, and the uncertainty mainly comes from the uncertainty of a prediction result, namely the error existing between the prediction value and the true value. This embodiment uses the theoretical framework proposed by GAL et al, and this framework has been validated in the RNN model.
The existing theoretical framework provides a theoretical proof of uncertainty in modeling of dropout as a tool for extracting information from the existing neural network model, and the form of the neural network model is not changed, but the average value of the prediction result and uncertainty (namely variance) of the prediction result are estimated and calculated by collecting the results of multiple random forward traversal of the neural network, and the formula is as follows:
wherein, in the above formula, x * Representing the actual sample value, y * The predicted value is represented by a value of the prediction,
and (5) representing the weight of the neural network of the i layer, and T representing the prediction duration.
In this embodiment, by using the existing dropout theory basis, the dropout layer is innovatively introduced into the constructed multilayer LSTM model, so as to obtain uncertainty of LSTM prediction, so that the model prediction is closer to the real scene, and the predicted result is more accurate. In the middle of two LSTM layers, dropout is introduced, and parameter values are set to be constant values between 0 and 1 generally, so that neurons in the model stop working with a certain probability, and the generalization of the model is stronger. For example, setting the parameter value to 0.5 indicates that the activation value of the neuron ceases to function with a 50% probability, such that the prediction is ambiguous. In the model prediction process, setting a training attribute value as True, namely setting a dropout to be opened in the prediction process, removing different hidden neurons is equivalent to using different neural network structures, and carrying out multiple-time prediction on the average value and variance of the prediction results of a plurality of trained neural networks to obtain uncertainty of short-term behavior prediction.
Further, the trained vehicle following behavior prediction model is used for predicting the acceleration of the last driving vehicle of the HDV queue, and then the motion equation v=v is used for 0 Sum of +at
The predicted acceleration is converted into velocity and position information.
Fig. 6 is a schematic diagram showing comparison of predicted values and actual values of a set of random samples, and fig. 6 shows comparison of predicted values and actual values of a set of random samples of future 3 seconds, and shows acceleration, speed and position of the last human-driven vehicle of the HDV from top to bottom. From extensive predictive analysis, acceleration is knownThe error is generally 0.6m/s 2 Within 0.15m/s, and within 0.3 m. Therefore, the predicted value is almost fitted with the true value, so that the accuracy of the vehicle following behavior prediction model provided by the embodiment is higher, and the feasibility of the model is verified.
It can be seen that by the hybrid queue human-driven vehicle following behavior prediction method provided by the embodiment, the running state information of the human-driven vehicle in the future time period can be accurately and rapidly predicted, a reliable data basis is provided for traffic system control, and the safety and the high efficiency of road traffic control are ensured.
In addition, compared with the prior art, the method for predicting the following behavior of the hybrid queue human driving vehicle has the following beneficial effects:
1. according to the invention, under the environment of limited perception and information sharing, the following behavior of the human driving vehicle in the new traffic environment is explored, a short-term vehicle following behavior prediction model is constructed, and a basis is provided for the application development of the human driving vehicle in the mixed flow in the future vehicle networking environment.
2. The invention introduces uncertainty analysis in the vehicle following behavior prediction model, analyzes the uncertainty of the driving behavior of the driving vehicle and the propagation state of the driving behavior of the driving vehicle in the sub-queue of the driving vehicle; and estimating the following behavior habit of the driver and the best distribution fitting result of uncertainty of the following behavior habit of the driver, and providing data support for queue cooperative control in a subsequent mixed flow environment.
The following description is made of the hybrid train-driving vehicle following behavior prediction device provided by the invention, and the hybrid train-driving vehicle following behavior prediction device described below and the hybrid train-driving vehicle following behavior prediction method described above can be correspondingly referred to each other.
Fig. 7 is a schematic structural diagram of a hybrid-queue vehicle following behavior prediction apparatus according to an embodiment of the present invention, referring to fig. 7, the vehicle following behavior prediction apparatus 70 may include:
an obtaining unit 701, configured to obtain driving state information of a first person driving vehicle and a last person driving vehicle in a to-be-predicted person driving vehicle sub-queue in the hybrid queue in a current time period;
the prediction unit 702 is configured to input driving state information of the first and last vehicles in the current time period into a trained vehicle following behavior prediction model, and obtain predicted driving state information of the last vehicle in the future time period of the sub-queue of the to-be-predicted vehicles.
In one embodiment, the human-driven vehicle following behavior prediction apparatus 70 may include a training unit, specifically configured to:
acquiring a plurality of sub-queues of the driving vehicles;
for each individual driving vehicle sub-queue, the following processing is performed:
acquiring running state information of a first person driving vehicle and a last person driving vehicle in a person driving vehicle sub-queue in a last time period as input sample data;
acquiring running state information of the last vehicle in the sub-queue of the human driving vehicle in the next time period as output sample data;
and training an initial vehicle following behavior prediction model by adopting the input sample data and the output sample data to obtain a trained vehicle following behavior prediction model.
In one embodiment, the training unit is specifically configured to:
the loss function is set as follows:
where loss represents a training function, t 2 Representing the predicted time step, t representing the current time,
indicating predicted acceleration +.>
Indicating the actual acceleration of the vehicle, I omega I 2 Representing positiveThen the item is converted;
and determining whether the value of the loss function is smaller than a preset value, if so, determining that training is completed, and obtaining a trained vehicle following behavior prediction model.
In an embodiment, the obtaining unit 70 is further configured to:
after the plurality of man-driven vehicle subqueues are obtained, the plurality of man-driven vehicle subqueues are processed, and man-driven vehicle subqueues with the sample size smaller than a preset value and man-driven vehicle subqueues which do not accord with the following condition are removed.
In an embodiment, the obtaining unit 70 is further configured to:
and when the sub-queues of the driving vehicles are obtained, removing the sub-queues of the driving vehicles with the conditions of lane changing, lane entrance and exit or too short lanes.
In one embodiment, the vehicle following behavior prediction model includes an encoder and a decoder.
The prediction unit 702 is specifically configured to:
the method comprises the steps that input driving state information of a first person driving vehicle and a last person driving vehicle in a current time period is transcoded into a fixed-length state vector through an encoder; the fixed-length state vector is decoded by a decoder into predicted running state information.
Fig. 8 is a schematic physical structure of an electronic device according to an embodiment of the present invention, as shown in fig. 8, where the electronic device may include: processor 810, communication interface (Communications Interface) 820, memory 830, and communication bus 840, wherein processor 810, communication interface 820, memory 830 accomplish communication with each other through communication bus 840. Processor 810 may invoke logic instructions in memory 830 to perform a hybrid-queue human drive-train behavior prediction method comprising: acquiring driving state information of a first person driving vehicle and a last person driving vehicle in a to-be-predicted person driving vehicle sub-queue in a mixed queue in a current time period; and inputting the driving state information of the first and last vehicles in the current time period into a trained vehicle following behavior prediction model to obtain the driving state information of the last vehicle in the future time period of the sub-queue of the to-be-predicted vehicles.
Further, the logic instructions in the memory 830 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, where the computer program product includes a computer program, where the computer program can be stored on a non-transitory computer readable storage medium, and when the computer program is executed by a processor, the computer can execute the method for predicting the following behavior of a hybrid-queue human driving vehicle provided by the above methods, and the method includes: acquiring driving state information of a first person driving vehicle and a last person driving vehicle in a to-be-predicted person driving vehicle sub-queue in a mixed queue in a current time period; and inputting the driving state information of the first and last vehicles in the current time period into a trained vehicle following behavior prediction model to obtain the driving state information of the last vehicle in the future time period of the sub-queue of the to-be-predicted vehicles.
In yet another aspect, the present invention further provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the hybrid-queue human-drive vehicle following behavior prediction method provided by the above methods, the method comprising: acquiring driving state information of a first person driving vehicle and a last person driving vehicle in a to-be-predicted person driving vehicle sub-queue in a mixed queue in a current time period; and inputting the driving state information of the first and last vehicles in the current time period into a trained vehicle following behavior prediction model to obtain the driving state information of the last vehicle in the future time period of the sub-queue of the to-be-predicted vehicles.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for predicting the following behavior of a hybrid passenger vehicle, comprising:
acquiring driving state information of a first person driving vehicle and a last person driving vehicle in a to-be-predicted person driving vehicle sub-queue in a mixed queue in a current time period;
and inputting the driving state information of the first and last driving vehicles in the current time period into a trained vehicle following behavior prediction model to obtain the driving state information of the last driving vehicle in the future time period of the sub-queue of the driving vehicles to be predicted.
2. The hybrid-queue human-driven vehicle following behavior prediction method according to claim 1, wherein the vehicle following behavior prediction model is obtained by training in the following manner:
acquiring a plurality of sub-queues of the driving vehicles;
for each of the human-driven vehicle sub-queues, performing the following processing:
acquiring running state information of a first person driving vehicle and a last person driving vehicle in the person driving vehicle sub-queue in a last time period as input sample data;
acquiring driving state information of the last vehicle in the sub-queue of the driving vehicle of the person in the next time period as output sample data;
and training an initial vehicle following behavior prediction model by adopting the input sample data and the output sample data to obtain the trained vehicle following behavior prediction model.
3. The hybrid-queue human-driven vehicle following behavior prediction method of claim 2, further comprising, after obtaining the plurality of human-driven vehicle subqueues:
and processing the plurality of man-driving vehicle subqueues, and removing the man-driving vehicle subqueues with the sample size smaller than a preset value and the man-driving vehicle subqueues which do not accord with the following condition.
4. A method of predicting the following behavior of a hybrid-tandem human drive vehicle as set forth in any one of claims 1-3, wherein said obtaining a plurality of sub-queues of human drive vehicles comprises:
and when the sub-queues of the driving vehicles are obtained, removing the sub-queues of the driving vehicles with the conditions of lane changing, lane entrance and exit or too short lanes.
5. The hybrid-queue human-driven vehicle following behavior prediction method of any one of claims 1-4, wherein the vehicle following behavior prediction model comprises an encoder and a decoder;
the step of inputting the driving state information of the first and last vehicles in the current time period into a trained vehicle following behavior prediction model to obtain the driving state information of the last vehicle in the sub-queue of the to-be-predicted vehicles in the future time period, including:
the encoder converts the input driving state information of the first driving vehicle and the last driving vehicle in the current time period into a state vector with a fixed length;
the decoder decodes the fixed-length state vector into the predicted running state information.
6. The hybrid-queue human-driven vehicle following behavior prediction method of claim 2, wherein training an initial vehicle following behavior prediction model using the input sample data and output sample data to obtain the trained vehicle following behavior prediction model comprises:
the loss function is set as follows:
where loss represents a training function, t 2 Representing the predicted time step, t representing the current time,
representing the predicted acceleration rate of the vehicle,
indicating the actual acceleration of the vehicle, I omega I 2 Representing a regularization term;
and determining whether the value of the loss function is smaller than a preset value, if so, determining that training is completed, and obtaining the trained vehicle following behavior prediction model.
7. The hybrid-queue human-driven vehicle following behavior prediction method of claim 2, wherein the vehicle following behavior prediction model is a multilayer LSTM model;
the two LSTM layers comprise a dropout layer, and the setting value of the LSTM layer is a constant value between 0 and 1.
8. A hybrid passenger vehicle following behavior prediction apparatus, comprising:
the system comprises an acquisition unit, a prediction unit and a control unit, wherein the acquisition unit is used for acquiring driving state information of a first person driving vehicle and a last person driving vehicle in a to-be-predicted person driving vehicle sub-queue in a mixed queue in a current time period;
the prediction unit is used for inputting the driving state information of the first and last driving vehicles in the current time period into a trained vehicle following behavior prediction model to obtain the predicted driving state information of the last driving vehicle in the future time period of the sub-queue of the driving vehicles to be predicted.
9. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the hybrid queued human driving vehicle following behavior prediction method of any of claims 1 to 7.
10. A computer program product comprising a computer program which when executed by a processor implements a method of hybrid queued human driving vehicle following behaviour prediction as claimed in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310422466.XA CN116386387B (en) | 2023-04-19 | 2023-04-19 | Method and device for predicting following behavior of driving vehicle of hybrid queue person |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310422466.XA CN116386387B (en) | 2023-04-19 | 2023-04-19 | Method and device for predicting following behavior of driving vehicle of hybrid queue person |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116386387A true CN116386387A (en) | 2023-07-04 |
| CN116386387B CN116386387B (en) | 2024-03-08 |
Family
ID=86980556
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310422466.XA Active CN116386387B (en) | 2023-04-19 | 2023-04-19 | Method and device for predicting following behavior of driving vehicle of hybrid queue person |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116386387B (en) |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9494940B1 (en) * | 2015-11-04 | 2016-11-15 | Zoox, Inc. | Quadrant configuration of robotic vehicles |
| WO2019246246A1 (en) * | 2018-06-20 | 2019-12-26 | Cavh Llc | Connected automated vehicle highway systems and methods related to heavy vehicles |
| US20200021961A1 (en) * | 2018-07-10 | 2020-01-16 | Cavh Llc | Vehicle on-board unit for connected and automated vehicle systems |
| CN111582586A (en) * | 2020-05-11 | 2020-08-25 | 长沙理工大学 | Multi-fleet driving risk prediction system and method for reducing jitter |
| CN112445229A (en) * | 2020-11-04 | 2021-03-05 | 清华大学 | Single-lane multi-queue hierarchical control method for piloting motorcade cooperation |
| CN112466119A (en) * | 2020-11-26 | 2021-03-09 | 清华大学 | Method and system for predicting vehicle following speed of vehicle by using vehicle-road cooperative data |
| CN113219962A (en) * | 2021-02-26 | 2021-08-06 | 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) | Control method and system for hybrid queue following safety and storage medium |
| CN113808438A (en) * | 2021-09-18 | 2021-12-17 | 长安大学 | Vehicle queue control mode switching system and method based on neural network |
| CN114387787A (en) * | 2022-03-24 | 2022-04-22 | 华砺智行(武汉)科技有限公司 | Vehicle track control method and device, electronic equipment and storage medium |
| CN114973733A (en) * | 2022-04-29 | 2022-08-30 | 北京交通大学 | Method for optimizing and controlling track of networked automatic vehicle under mixed flow at signal intersection |
| CN115032931A (en) * | 2022-06-15 | 2022-09-09 | 东南大学 | Intelligent network united queue converging control method based on clustering theory |
| CN115083156A (en) * | 2022-06-14 | 2022-09-20 | 北京航空航天大学 | Vehicle track prediction method based on Newell following model |
| US11521396B1 (en) * | 2019-02-02 | 2022-12-06 | Uatc, Llc | Probabilistic prediction of dynamic object behavior for autonomous vehicles |
| CN115497281A (en) * | 2022-08-16 | 2022-12-20 | 清华大学 | Robust model prediction control method for multi-queue pilot vehicle in mixed traffic scene |
-
2023
- 2023-04-19 CN CN202310422466.XA patent/CN116386387B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9494940B1 (en) * | 2015-11-04 | 2016-11-15 | Zoox, Inc. | Quadrant configuration of robotic vehicles |
| WO2019246246A1 (en) * | 2018-06-20 | 2019-12-26 | Cavh Llc | Connected automated vehicle highway systems and methods related to heavy vehicles |
| US20200021961A1 (en) * | 2018-07-10 | 2020-01-16 | Cavh Llc | Vehicle on-board unit for connected and automated vehicle systems |
| US11521396B1 (en) * | 2019-02-02 | 2022-12-06 | Uatc, Llc | Probabilistic prediction of dynamic object behavior for autonomous vehicles |
| CN111582586A (en) * | 2020-05-11 | 2020-08-25 | 长沙理工大学 | Multi-fleet driving risk prediction system and method for reducing jitter |
| CN112445229A (en) * | 2020-11-04 | 2021-03-05 | 清华大学 | Single-lane multi-queue hierarchical control method for piloting motorcade cooperation |
| CN112466119A (en) * | 2020-11-26 | 2021-03-09 | 清华大学 | Method and system for predicting vehicle following speed of vehicle by using vehicle-road cooperative data |
| CN113219962A (en) * | 2021-02-26 | 2021-08-06 | 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) | Control method and system for hybrid queue following safety and storage medium |
| CN113808438A (en) * | 2021-09-18 | 2021-12-17 | 长安大学 | Vehicle queue control mode switching system and method based on neural network |
| CN114387787A (en) * | 2022-03-24 | 2022-04-22 | 华砺智行(武汉)科技有限公司 | Vehicle track control method and device, electronic equipment and storage medium |
| CN114973733A (en) * | 2022-04-29 | 2022-08-30 | 北京交通大学 | Method for optimizing and controlling track of networked automatic vehicle under mixed flow at signal intersection |
| CN115083156A (en) * | 2022-06-14 | 2022-09-20 | 北京航空航天大学 | Vehicle track prediction method based on Newell following model |
| CN115032931A (en) * | 2022-06-15 | 2022-09-09 | 东南大学 | Intelligent network united queue converging control method based on clustering theory |
| CN115497281A (en) * | 2022-08-16 | 2022-12-20 | 清华大学 | Robust model prediction control method for multi-queue pilot vehicle in mixed traffic scene |
Non-Patent Citations (1)
| Title |
|---|
| 安实;姚焓东;姜慧夫;崔建勋;: "信号交叉口绿色驾驶车速控制方法", 交通运输系统工程与信息, no. 05, 15 October 2015 (2015-10-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116386387B (en) | 2024-03-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhou et al. | Development of an efficient driving strategy for connected and automated vehicles at signalized intersections: A reinforcement learning approach | |
| Bai et al. | Hybrid reinforcement learning-based eco-driving strategy for connected and automated vehicles at signalized intersections | |
| Xiong et al. | A new framework of vehicle collision prediction by combining SVM and HMM | |
| Lee et al. | Integrated deep learning and stochastic car-following model for traffic dynamics on multi-lane freeways | |
| CN109910909B (en) | Automobile track internet interactive prediction method for multi-automobile motion state | |
| CN112215337A (en) | Vehicle trajectory prediction method based on environment attention neural network model | |
| US11465650B2 (en) | Model-free reinforcement learning | |
| CN109711557A (en) | A kind of wheelpath prediction technique, system, computer equipment and storage medium | |
| el abidine Kherroubi et al. | Novel decision-making strategy for connected and autonomous vehicles in highway on-ramp merging | |
| CN112734808B (en) | Trajectory prediction method for vulnerable road users in vehicle driving environment | |
| JP2020059497A (en) | Method and device of controlling automatic driving introducing "detection acquisition" model, and program constituted thereby | |
| KR101742043B1 (en) | Apparatus and method for travel mode choice prediction | |
| CN112487954B (en) | Pedestrian crossing behavior prediction method for plane intersection | |
| CN110196587A (en) | Vehicular automatic driving control strategy model generating method, device, equipment and medium | |
| Katariya et al. | Deeptrack: Lightweight deep learning for vehicle trajectory prediction in highways | |
| CN113901718A (en) | Deep reinforcement learning-based driving collision avoidance optimization method in following state | |
| Zhang et al. | Predictive trajectory planning for autonomous vehicles at intersections using reinforcement learning | |
| Chen et al. | Follownet: A comprehensive benchmark for car-following behavior modeling | |
| Wu et al. | Lane-GNN: Integrating GNN for Predicting Drivers' Lane Change Intention | |
| Lima et al. | Driver behavior analysis based on numerical data using deep neural networks | |
| CN116386387B (en) | Method and device for predicting following behavior of driving vehicle of hybrid queue person | |
| Kuutti et al. | Adversarial mixture density networks: Learning to drive safely from collision data | |
| CN115981302A (en) | Vehicle following lane change behavior decision-making method and device and electronic equipment | |
| Chen et al. | Platoon separation strategy optimization method based on deep cognition of a driver’s behavior at signalized intersections | |
| Lv et al. | A lane-changing decision-making model of bus entering considering bus priority based on GRU neural network |
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 |