CN114852099A - Method for predicting lane changing behavior of motor vehicle - Google Patents
Method for predicting lane changing behavior of motor vehicle Download PDFInfo
- Publication number
- CN114852099A CN114852099A CN202110151796.0A CN202110151796A CN114852099A CN 114852099 A CN114852099 A CN 114852099A CN 202110151796 A CN202110151796 A CN 202110151796A CN 114852099 A CN114852099 A CN 114852099A
- Authority
- CN
- China
- Prior art keywords
- lane
- vehicle
- target vehicle
- probability
- expected utility
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 64
- 230000008859 change Effects 0.000 claims abstract description 93
- 241001127637 Plantago Species 0.000 claims abstract description 6
- 230000001133 acceleration Effects 0.000 claims description 68
- 238000004364 calculation method Methods 0.000 claims description 14
- 239000000126 substance Substances 0.000 claims description 9
- 230000006399 behavior Effects 0.000 abstract description 83
- 230000009471 action Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 16
- 230000008569 process Effects 0.000 description 16
- 238000011160 research Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 238000007476 Maximum Likelihood Methods 0.000 description 3
- 230000003631 expected effect Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000004936 stimulating effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000002939 conjugate gradient method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
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
- B60W60/00—Drive control systems specially adapted for autonomous road vehicles
- B60W60/001—Planning or execution of driving tasks
- B60W60/0027—Planning or execution of driving tasks using trajectory prediction for other traffic participants
Abstract
The invention relates to a method for predicting lane change behaviors of a motor vehicle, and belongs to the technical field of automatic driving. The method comprises the following steps: according to a revenue function
Calculating an expected utility value of a left lane changing, an expected utility value of a lane keeping and an expected utility value of a right lane changing of a target vehicle; wherein u is the expected utility value;
is the plantago exercisable space index;
is a collision risk index; c is a comfort index; omega 1 ，ω 2 ，ω 3 Is the corresponding coefficient; comparing the expected utility value of the left lane changing, the expected utility value of the lane keeping and the expected utility value of the right lane changing, wherein the corresponding lane changing behavior with the maximum expected utility value is the predicted lane changing behaviorA lane change action. According to the method and the device, when the target vehicle lane changing behavior is predicted, the relationship between the target vehicle and surrounding vehicles is considered, the lane changing behavior prediction of the target vehicle under a complex traffic scene is realized, and the accuracy of the lane changing behavior prediction of the target vehicle is improved.
Description
Technical Field
The invention relates to a method for predicting lane change behaviors of a motor vehicle, and belongs to the technical field of automatic driving.
Background
With the development of sensor technology, computer and automotive electronics, automotive applications are becoming more and more widespread. In the automatic driving process, the automatic driving vehicle detects the driving condition around the driving vehicle through various sensors, radars, cameras and other equipment arranged on the vehicle so as to ensure the safe driving of the vehicle.
In order to further ensure the driving safety, the automatic driving vehicle needs to predict the lane changing behavior of other vehicles, and further determine the motion state of the vehicle. At present, the Baidu Apollo predicts the behavior of a vehicle by using a recurrent neural network (deep learning), which trains a prediction model through a large amount of historical data to predict a target vehicle, but the method is only suitable for a trained scene, and fails in a new traffic scene, and the method cannot well explain the behavior of the target vehicle and cannot perform reverse analysis according to a prediction result.
For this reason, studies on the single-vehicle state of the target vehicle are proposed to predict the lane change behavior, however, most studies do not consider the interaction influence between vehicles in a complex traffic scene or do not study the vehicle interaction deeply enough, and are only applicable to a simple traffic scene, so that the predicted lane change behavior has a large deviation from the actual lane change behavior.
Disclosure of Invention
The application aims to provide a method for predicting lane changing behaviors of a motor vehicle, which is used for solving the problem of low accuracy of the conventional lane changing behavior prediction.
In order to achieve the above object, the present application proposes a technical solution of a first method for predicting a lane change behavior of a motor vehicle, the method comprising the steps of:
1) according to a revenue function
Calculating the expected effect value of the left lane change of the target vehicle and the expected effect of lane keepingChanging the expected utility value of the channel by the value and the right;
wherein u is the expected utility value;
is the plantago exercisable space index;
is a collision risk index; c is a comfort index; omega 1 ，ω 2 ，ω 3 Is the corresponding coefficient;
2) comparing the expected utility value of the left lane changing, the expected utility value of the lane keeping and the expected utility value of the right lane changing, wherein the corresponding lane changing behavior with the maximum expected utility value is the predicted lane changing behavior;
the space index of the front feasible vehicle is obtained according to the distance between the target vehicle and the front vehicle; the collision risk index is obtained according to the distance between the target vehicle and other vehicles; the comfort index is obtained from the longitudinal acceleration and the lateral acceleration of the target vehicle.
The technical scheme of the first method for predicting the lane change behavior of the motor vehicle has the beneficial effects that: the method and the device realize the prediction of the lane change behavior by calculating the expected utility value of the target vehicle under different lane change behaviors, wherein the expected utility value consists of a vehicle-ahead feasible space index, a collision danger index and a comfort index, and the vehicle-ahead feasible space index and the collision danger index both represent the distance relationship between the target vehicle and the surrounding vehicles.
Further, the calculation process of the space index of the vehicle front driving is as follows:
wherein, the first and the second end of the pipe are connected with each other,
is the index of the space available for driving before the vehicle; d r The distance between the target vehicle and the front vehicle; d v Is the visible distance.
Further, the calculation process of the collision risk index is as follows:
wherein the content of the first and second substances,
is a collision risk index;
representing the minimum value of the distance between the target vehicle and the other vehicle.
Further, the comfort index is calculated by the following steps:
wherein c is a comfort index; a is x Is the longitudinal acceleration of the target vehicle; a is y Is the lateral acceleration of the target vehicle; t is the prediction time period.
Further, the expected utility value of the left lane change is obtained by predicting the left lane change of the target vehicle and the track of the lane keeping of the surrounding vehicles; obtaining an expected utility value of a lane keeping through predicting the tracks of the lane keeping of a target vehicle and the lane keeping of surrounding vehicles; and obtaining an expected utility value of right lane changing by predicting the tracks of the right lane changing of the target vehicle and the lane keeping of the surrounding vehicles.
In addition, the application also provides a technical scheme of a second method for predicting the lane change behavior of the motor vehicle, and the method comprises the following steps:
acquiring a motion track of a target vehicle; obtaining the lateral offset and the lateral offset speed of the target vehicle relative to the central line of the lane according to the motion track of the target vehicle; inputting the lateral offset and the lateral offset speed into a pre-established behavior recognition probability model to obtain a first left lane changing probability, a first lane keeping probability and a first right lane changing probability of the target vehicle; the behavior recognition probability model is a continuous hidden Markov model;
according to a revenue function
Calculating an expected utility value of a left lane changing, an expected utility value of a lane keeping and an expected utility value of a right lane changing of a target vehicle; normalizing each expected utility value to obtain corresponding left lane change intention probability, lane keeping intention probability and right lane change intention probability;
wherein u is the expected utility value;
is the plantago exercisable space index;
is a collision risk index; c is a comfort index; omega 1 ，ω 2 ，ω 3 Is the corresponding coefficient;
weighting and superposing the first left lane changing probability and the left lane changing intention probability to obtain a second left lane changing probability; weighting and superposing the first lane keeping probability and the lane keeping intention probability to obtain a second lane keeping probability; weighting and superposing the first right lane changing probability and the right lane changing intention probability to obtain a second right lane changing probability;
and comparing the second left lane changing probability, the second lane keeping probability and the second right lane changing probability, wherein the corresponding lane changing behavior with the maximum probability is the predicted lane changing behavior.
The second method for predicting the lane change behavior of the motor vehicle has the following beneficial effects: the method comprises the steps of calculating expected utility values of a target vehicle under different lane changing behaviors, and further realizing prediction of the lane changing behaviors, wherein the expected utility values consist of a vehicle-front feasible space index, a collision danger index and a comfort index, and the vehicle-front feasible space index and the collision danger index both represent the distance relationship between the target vehicle and surrounding vehicles; according to the method and the device, when the target vehicle lane changing behavior is predicted, the relationship between the target vehicle and surrounding vehicles is considered, the lane changing behavior prediction of the target vehicle under a complex traffic scene is realized, and the accuracy of the lane changing behavior prediction of the target vehicle is improved. Meanwhile, under a complex traffic scene, the lane change probability of the target vehicle is predicted based on the behavior recognition probability model of the continuous hidden Markov model, and the lane change probability and the behavior prediction under the complex traffic scene are weighted and superposed to obtain the final lane change probability. The method comprehensively considers the influence of complex traffic and historical data, and further improves the accuracy of the lane change behavior prediction.
Further, the observation probability in the continuous hidden markov model is a probability density function of gaussian distribution.
Further, the calculation process of the space index of the vehicle front driving is as follows:
wherein the content of the first and second substances,
is the index of the space available for driving before the vehicle; d r The distance between the target vehicle and the front vehicle; d v Is the visible distance.
Further, the calculation process of the collision risk index is as follows:
wherein the content of the first and second substances,
is a collision risk index;
representing the minimum value of the distance between the target vehicle and the other vehicle.
Further, the comfort index is calculated by the following steps:
wherein c is a comfort index; a is x Is the longitudinal acceleration of the target vehicle; a is y Is the lateral acceleration of the target vehicle; t is the prediction time period.
Drawings
FIG. 1 is a flow chart of embodiment 1 of the method for predicting lane change behavior of a motor vehicle according to the present invention;
FIG. 2 is a flow chart of embodiment 2 of the method for predicting lane change behavior of a motor vehicle according to the present invention;
FIG. 3 is a schematic illustration of an I80 road segment research area of the present invention;
fig. 4 is a graph of the weight coefficient of the intention probability of the present invention and the intention probability of keeping a lane.
Detailed Description
Method for predicting lane change behavior of motor vehicle example 1:
the method mainly includes the steps that based on the situation of complex actual traffic, the distance between a target vehicle and surrounding vehicles and the longitudinal acceleration and the lateral acceleration of the target vehicle are determined by predicting the left lane changing, lane keeping, right lane changing and the running tracks of the surrounding vehicles of the target vehicle, then a first expected utility value after the left lane changing, a second expected utility value after the lane keeping and a third expected utility value after the right lane changing are obtained by combining an expected utility theory respectively, and the predicted lane changing behavior is determined by comparing the expected utility values under different conditions.
The invention uses the expected utility theory to calculate the driving intention of the target vehicle, and the expected utility is the sum of products between the income brought by all possible scenes and the occurrence probability. According to the theory of expected utility, the vehicle selects the action that produces the highest expected utility. The expected utility generated by a certain behavior of the target vehicle is related to the behavior probability distribution of the vehicles around the target vehicle and the revenue function of each scene, which accords with a rational thinking process when a driver makes a behavior decision: the method comprises the steps of judging the ongoing behaviors of other surrounding vehicles, evaluating the income brought by a scene which possibly occurs, and finally selecting the behavior with the highest expected effectiveness to make a final decision.
For a predicted target vehicle, the expected utility value produced by each of its behaviors represents, in essence, the magnitude of the probability that the driver of the vehicle will select such behavior at the intended level for a future period of time. Therefore, the invention can obtain the intention probability of the driver generating the behavior in the future period by normalizing the expected utility value under the condition that all drivers are rational drivers. Therefore, the profit of the target vehicle under all possible scenes needs to be calculated, and the revenue function is defined to comprise three parts: the vehicle front driving space index, the collision risk index and the comfort index belong to positive earnings, and the collision risk index and the comfort index belong to negative earnings.
Accordingly, a revenue function is designed to model the driver's intent, utilizing the expected effect to express the probability that the transportation vehicle will produce each behavioral intent. Based on one such assumption: the behavior decision of normal and rational drivers during driving can be abstracted into a process of continuously seeking maximum benefit. The designed revenue function is as follows:
wherein u is the expected utility value;
is the plantago exercisable space index;
is a collision risk index; c is a comfort index; omega 1 ，ω 2 ，ω 3 Is the corresponding probability coefficient.
Specifically, the method for predicting the lane change behavior of the motor vehicle is shown in fig. 1, and comprises the following steps:
1) the method comprises the steps that after a lane is changed on the left side of a target vehicle and tracks of surrounding vehicles for keeping lanes are predicted, the distance between the target vehicle and the surrounding vehicles and the longitudinal acceleration and the lateral acceleration of the target vehicle are obtained; predicting the tracks of a target vehicle keeping lane and surrounding vehicles keeping lane to obtain the distance between the target vehicle and the surrounding vehicles, the longitudinal acceleration and the lateral acceleration of the target vehicle after lane keeping is achieved; and predicting the tracks of the right lane change of the target vehicle and the lane keeping of the surrounding vehicles to obtain the distance between the target vehicle and the surrounding vehicles after the right lane change, and the longitudinal acceleration and the lateral acceleration of the target vehicle.
The lane keeping trajectory prediction method, whether it is the target vehicle or the surrounding vehicle, is as follows:
predicting the future track of the vehicle through a constant angular velocity acceleration model, wherein the constant angular velocity acceleration model is as follows:
wherein X is the size of the vehicle in the X direction in the global coordinate system; y is the size of the vehicle in the Y direction in the global coordinate system;
is the heading angle of the vehicle; v. of x The longitudinal speed of the vehicle under a vehicle coordinate system; v. of y The lateral speed of the vehicle under a vehicle coordinate system; omega is the yaw angular velocity of the vehicle; a is x Is the longitudinal acceleration of the vehicle; a is y Is the lateral acceleration of the vehicle; w is a ω A disturbance value that is a derivative of the yaw angular acceleration of the vehicle; w is a ax A disturbance value that is a derivative of the longitudinal acceleration of the vehicle; w is a ay A disturbance value that is a derivative of the lateral acceleration of the vehicle. After detecting the vehicle motion state by the vehicle's own on-board sensors,the vehicle network technology is used for transmitting the vehicle to the vehicle, and the vehicle can predict the motion trail of the lane kept by other vehicles by using the above formula. And when the obstacle is shielded, the position information and the like of other obstacles can be obtained according to the drive test equipment.
In the lane keeping trajectory prediction, the motion parameters of the vehicle are kept unchanged in order to simplify the calculation process.
The lane change trajectory prediction method of the target vehicle is as follows:
the target vehicle performing the lane-change maneuver uses a fifth-order polynomial to fit the lane-change trajectory of the vehicle. Establishing a coordinate system (x, y) by using the direction and the position of the lane change starting point of the target vehicle, wherein the equation of the central line is y and ax 2 + bx + c, where a, b, c are coefficients, assuming (x) D ，y D ) For a target point for a target vehicle lane change, the lane change trajectory of the target vehicle may be expressed as:
y＝a 5 x 5 +a 4 x 4 +a 3 x 3 +a 2 x 2 +a 1 x+a 0 ；
suppose that the vehicle speed at the present time is v x The coefficient a of the fifth-order polynomial is ω 5 、a 4 、a 3 、a 2 、a 1 、a 0 Comprises the following steps:
wherein:
to simplify the calculation process, assume the vehicle speed v of the vehicle x And the yaw rate omega is not changed, and the vehicle speed v of the vehicle is x When the yaw angular velocity omega is constant, the lane-changing track can be changed from the lane-changing target point (x) D ，y D ) Abscissa x of D And (4) uniquely determining.
In the track prediction, firstly, a set of discrete x is used D And generating a group of tracks, and evaluating the group of predicted tracks through a defined evaluation function to select an optimal predicted track.
The evaluation function J of the predicted trajectory of the present invention is:
wherein, a ymax Maximum lateral acceleration generated when the vehicle runs along each predicted track; t is the time spent;
maximum lateral acceleration in all predicted trajectories; t is a unit of max The maximum elapsed time in all predicted trajectories;
and
for the weight coefficient, the invention selects
The evaluation function reflects the cost of the track in terms of comfort and driving efficiency, so that the x with the minimum evaluation function value at each vehicle speed is used D And the optimal predicted driving track can be selected by calculating the coefficient of the fifth-order polynomial of the track according to the longitudinal lane changing distance under the vehicle speed.
In order to simplify the calculation model, the longitudinal acceleration and the lateral acceleration of the target vehicle, whether the lane change is left, the lane keeping is right, or the lane change is right, are the longitudinal acceleration and the lateral acceleration acquired at the present time, and are calculated on the basis of assuming that the longitudinal acceleration and the lateral acceleration are unchanged.
The lane-keeping trajectory prediction and the lane-change trajectory prediction are both calculated in the host vehicle, and the host vehicle predicts the lane-change behavior of the target vehicle.
2) Obtaining a first vehicle-front travelable space index according to the distance between a target vehicle and a front vehicle (the front of the target vehicle is not the right front in an absolute sense and is a vehicle in a certain range in front of the target vehicle) after the target vehicle changes lanes left, obtaining a first collision risk index according to the distance between the target vehicle and other vehicles (the other vehicles are other surrounding vehicles except the front vehicle), and obtaining a first comfort index according to the longitudinal acceleration and the lateral acceleration of the target vehicle; obtaining a second front-of-vehicle travelable space index according to the distance between the target vehicle and the front vehicle after the target vehicle keeps the lane, obtaining a second collision risk index according to the distance between the target vehicle and other vehicles, and obtaining a second comfort index according to the longitudinal acceleration and the lateral acceleration of the target vehicle; and obtaining a third front travelable space index according to the distance between the target vehicle and the front vehicle, obtaining a third collision risk index according to the distance between the target vehicle and other vehicles, and obtaining a third comfort index according to the longitudinal acceleration and the lateral acceleration of the target vehicle after the target vehicle changes lanes to the right.
The calculation process of the front driving space index is as follows:
wherein the content of the first and second substances,
is the index of the space available for driving before the vehicle; d r The distance between the target vehicle and the front vehicle; d v Is a visible distance; the visible distance is a quantity related to the visible distance, the visible range of young drivers is about 70m-200m, and D is set v ＝150m。
The collision risk index is calculated by the following process:
wherein the content of the first and second substances,
is a collision risk index;
representing the minimum value of the distance between the target vehicle and the other vehicle.
The satisfaction of the safety condition means that the lateral distance or the longitudinal distance of the target vehicle from the other vehicle within the predicted time period T is within a safety distance range. Suppose a target vehicle v o Respectively has a length and a width of L o 、D o Other surrounding vehicles v t Respectively has a length and a width of L t 、D t The coordinates and course angles of two vehicles at a certain time t are
And
difference between course angles of two vehicles
The condition that no collision occurs between the two vehicles can be obtained through the distance between the central points of the two vehicles in the x direction and the y direction, namely that the condition that the collision does not occur between the two vehicles at any time t in the prediction time period is met:
wherein Δ S is a longitudinal safety distance; Δ W is a lateral safety distance, Δ S ═ 2 Δ v, Δ W ═ 0.3 m;
The comfort index is calculated by the following steps:
wherein c is a comfort index; a is x Is the longitudinal acceleration of the target vehicle; a is y Is the lateral acceleration of the target vehicle; t is the prediction time period. The comfort index may be understood as the negative of the integral of the square of the longitudinal acceleration and the lateral acceleration of the target vehicle over the prediction time period T.
The positions of the target vehicle and the surrounding vehicles in the prediction time period can be obtained according to the track prediction, and meanwhile, a first vehicle-front travelable space index, a first collision risk index and a first comfort index are respectively calculated according to the formula; a second front-of-vehicle travelable space index, a second collision risk index, a second comfort index; a third vehicle-front travelable space index, a third collision risk index, and a third comfort index.
The collision risk index and the comfort index are process quantities in a prediction time period, and the index of the space available for driving in front is a state quantity which is directly a state that the target vehicle and other surrounding vehicles reach the position of the predicted track end point.
3) Obtaining a first expected utility value of the target vehicle after the left lane change according to the first vehicle front travelable space index, the first collision risk index and the first comfort index and by combining an expected utility theory; obtaining a second expected utility value of the target vehicle after keeping the lane according to the second front-vehicle travelable space index, the second collision risk index and the second comfort index by combining an expected utility theory; and obtaining a third expected utility value of the target vehicle after the left lane change according to the third vehicle front travelable space index, the third collision risk index and the third comfort index and by combining an expected utility theory.
Substituting the first in-front travelable space index, the first collision risk index and the first comfort index into
Obtaining a first expected utility value; substituting the second front-of-vehicle travelable space index, the second collision risk index and the second comfort index into
Obtaining a second expected utility value; substituting the third front-vehicle travelable space index, the third collision risk index and the third comfort index into
A third expected utility value is obtained.
The parameter to be calibrated in the gain function is recorded as theta ═ omega 1 ，ω 2 ，ω 3 The invention calibrates the parameters through maximum likelihood estimation, and the calibration process is as follows:
for the target vehicle v o Its certain behavior m o，i The desired effect produced is
The state information of the vehicle (including the target vehicle and the surrounding vehicles) is required for simultaneous calculation of the revenue function and the intention probability of the vehicle
And a parameter theta, so to make the parameter calibration method easier to understand, the vehicle v is used o Behavior m of o，i Expected utility of
Given a
And theta, the intention probability p Intention to Can be written as
Wherein m is o，i Is a target vehicle v o 1, 2, and 3, which respectively represent the behaviors of changing lanes left, keeping lanes, and changing lanes right;
the state information of the target vehicle and the surrounding vehicles from the time t-h to the time t,
(x, y) is the global coordinates of the vehicle, v x Is the longitudinal speed, v, of the vehicle y As the lateral speed of the vehicle, a x Is the longitudinal acceleration of the vehicle, a y Is the lateral acceleration of the vehicle, delta is the heading angle of the vehicle; θ ═ ω 1 ，ω 2 ，ω 3 The parameter to be calibrated in the gain function is obtained;
is a collection of vehicles, in common
A trolley.
The above formula is satisfied:
the invention calibrates the parameters by maximum likelihood estimation, and the likelihood function is as follows:
l (theta) expresses the parameter theta toThe likelihood of the set is assumed to be estimated as θ
Then
Can be obtained by the maximum likelihood method:
the likelihood function L (θ) is a function relating to the parameter θ ═ ω 1 ，ω 2 ，ω 3 The multivariate nonlinear equation of the method can be estimated by a mature conjugate gradient method
The value of (c).
4) And normalizing the first expected utility value, the second expected utility value and the third expected utility value to obtain corresponding left lane-changing intention probability, lane-keeping intention probability and right lane-changing intention probability, and comparing the left lane-changing intention probability, the lane-keeping intention probability and the right lane-changing intention probability, wherein the corresponding lane-changing behavior with the highest intention probability is the predicted lane-changing behavior.
In the above embodiment, in order to obtain the predicted lane change behavior more intuitively, the expected utility values are normalized and then compared, as another implementation, the expected utility values may also be directly compared, and the lane change behavior corresponding to the maximum expected utility value is the predicted lane change behavior.
In order to simplify the calculation model, the trajectories of all the vehicles around the target vehicle are predicted to keep the lane, and as another embodiment, after the information of the vehicles around is acquired through the vehicle network technology, how to calculate that the vehicles around have obvious lane changing behaviors, the prediction of the lane changing trajectories of the vehicles around can be carried out.
According to the method, the track of the target vehicle and surrounding vehicles is predicted, and the left lane change intention probability, the lane keeping intention probability and the right lane change intention probability of the target vehicle are obtained by combining an expected utility theory, so that the lane change behavior of the target vehicle is predicted. According to the invention, the error between the predicted lane change behavior and the actual lane change behavior is reduced by predicting the complex traffic scene, and the prediction accuracy is improved.
the method for predicting the lane change behavior of the motor vehicle in the embodiment is different from the method for predicting the lane change behavior of the motor vehicle in the embodiment 1 in that on the basis of the embodiment 1, the lane change probability of the target vehicle is identified through a behavior identification probability model, and the lane change probability and the intention probability are weighted to obtain the final lane change probability, namely, the influence of a complex traffic scene and a historical track is comprehensively considered, and the lane change prediction of the target vehicle is accurately performed.
Specifically, the method for predicting the lane change behavior of the motor vehicle is shown in fig. 2, and comprises the following steps:
1) acquiring the distance between a target vehicle and surrounding vehicles and the longitudinal acceleration and the lateral acceleration of the target vehicle after the left lane change; after keeping the lane, the distance between the target vehicle and the surrounding vehicles, and the longitudinal acceleration and the lateral acceleration of the target vehicle; after right lane changing, the distance between the target vehicle and the surrounding vehicles, and the longitudinal acceleration and the lateral acceleration of the target vehicle; and simultaneously acquiring the motion trail of the target vehicle.
Regarding the distance between the target vehicle and the surrounding vehicle after the left lane change, and the longitudinal acceleration and the lateral acceleration of the target vehicle; after keeping the lane, the distance between the target vehicle and the surrounding vehicles, and the longitudinal acceleration and the lateral acceleration of the target vehicle; after the lane change, the distance between the target vehicle and the surrounding vehicles, and the acquisition manner of the longitudinal acceleration and the lateral acceleration of the target vehicle are the same as those in embodiment 1, and details are not described here.
The motion trail of the target vehicle can be sensed and acquired by a sensor of the host vehicle, such as: a millimeter wave radar of the host vehicle, and the like. The data that the host vehicle can acquire the target vehicle and the surrounding vehicles includes: position, heading angle, velocity, and acceleration. And superposing the obtained positions to obtain the motion trail of the target vehicle.
2) Obtaining a first vehicle front travelable space index according to the distance between the target vehicle and the front vehicle after the lane is changed on the left side, obtaining a first collision risk index according to the distance between the target vehicle and other vehicles, and obtaining a first comfort index according to the longitudinal acceleration and the lateral acceleration of the target vehicle; obtaining a second front-of-vehicle travelable space index according to the distance between the target vehicle and the front vehicle after the lane is kept, obtaining a second collision risk index according to the distance between the target vehicle and other vehicles, and obtaining a second comfort index according to the longitudinal acceleration and the lateral acceleration of the target vehicle; obtaining a third front travelable space index according to the distance between the target vehicle and the front vehicle after the lane change on the right, obtaining a third collision risk index according to the distance between the target vehicle and other vehicles, and obtaining a third comfort index according to the longitudinal acceleration and the lateral acceleration of the target vehicle; and obtaining the lateral offset and the lateral offset speed of the target vehicle relative to the central line of the lane according to the motion trail of the target vehicle.
The calculation processes of the vehicle front driving space index, the collision risk index, and the comfort index have been described in embodiment 1, and are not described herein again.
3) Obtaining a first expected utility value of the target vehicle after the left lane change according to the first vehicle front travelable space index, the first collision risk index and the first comfort index and by combining an expected utility theory; obtaining a second expected utility value of the target vehicle after keeping the lane according to the second front-vehicle travelable space index, the second collision risk index and the second comfort index by combining an expected utility theory; obtaining a third expected utility value of the target vehicle after the left lane change according to the third vehicle-front travelable space index, the third collision risk index and the third comfort index and by combining an expected utility theory; normalizing the first expected utility value, the second expected utility value and the third expected utility value to obtain corresponding left lane-changing intention probability, lane-keeping intention probability and right lane-changing intention probability; and simultaneously inputting the lateral offset and the lateral offset speed into a pre-established behavior recognition probability model to obtain a first left lane changing probability, a first lane keeping probability and a first right lane changing probability of the target vehicle.
The behavior recognition probability model reflects the judgment of the lane changing behavior of the target vehicle through the historical track, and is used for judging the possible behavior of the target vehicle through the motion state of a single vehicle of the target vehicle. The behavior recognition probability model is used for modeling the behavior of a target vehicle based on a hidden Markov model, and because the motion state of the vehicle is a continuous variable, a continuous hidden Markov model is adopted, and the parameters of the continuous hidden Markov model comprise 2 state sets and 3 probability matrixes, and are expressed as follows:
λ＝{S，O，A，B，π}；
S＝{s 1 ，s 2 ，s 3 }；
A＝[a ij ]，a ij ＝p(q t+1 ＝s j |q t ＝s i )；
B＝[b i (O t )]，b i (O t )＝p(O t |q t ＝s i )；
π′＝{π 1 ，π 2 ，π 3 }；
wherein S is an implicit state, namely a channel switching state; s 1 Changing the lane for the left; s 2 To keep the lane; s 3 Changing the lane for the right; o is an observable state; d is a lateral offset;
is the lateral offset velocity; a is the state transition probability; a is ij Is in a state s of i Transition to state s j The probability of (d); q. q.s t+1 The state at the time t + 1; q. q.s t Is the state at time t; b is output observation probability; b is a mixture of i (O t ) To be in an implicit state of s i Time-out observation state O t The probability of (d); pi' is the initial state probability; pi i Is at s at the initial time i Probability of state.
The output observation probability B is expressed by using a probability density function of a gaussian distribution:
wherein, mu i Is the mean vector of the Gaussian distribution; sigma i A covariance matrix which is a Gaussian distribution; t is the length of the observed time series.
In the continuous hidden Markov model, S and O are defined according to the actual traffic scene, so that the complete continuous hidden Markov model can be obtained only by solving { A, B, omega }, namely solving a ij 、μ i 、∑ i 、ω i 。
Training a continuous hidden Markov model by an NGSIM (Next Generation subscriber identity Module) traffic data set by adopting a Baum-Welch algorithm, and iteratively solving a ij 、μ i 、∑ i 、ω i The specific training process is as follows:
a. and after filtering the data in the NGSIM traffic data set, extracting lane-changing and straight-going data to obtain a sample data set.
An I80 road section in the NGSIM data set is selected as a research object, the road section is located in Moliville city, san Francisco, Calif., as shown in FIG. 3, a research area is a road section from south to north, a main line is 503 meters long and comprises 6 lanes, a junction of an entrance ramp and the main line is located at 128 meters of the main line, an acceleration lane is about 94 meters long, the total length of ramps is 200 meters, the distance between the entrance ramp and the exit ramp is greater than 375 meters, an exit ramp is located outside the research area, lane numbers are sequentially increased from left to right and from 1 to 7, wherein 6 lanes are auxiliary lanes, and 7 lanes are entrance ramps.
The track information of the vehicle in the NGSIM data is obtained through an image recognition technology, and certain errors and noises exist in the data, so that abnormal data such as vehicle coordinates, acceleration, vehicle speed and the like are generated. In order to reduce the measurement noise of the data, a logarithmic exponential moving average filtering algorithm is adopted to preprocess the data. And then, extracting sample track data under three behaviors of changing lanes on the left, keeping lanes and changing lanes on the right as a sample data set. Since the data of the NGSIM do not all satisfy the requirements, before extracting the data, the rules of the extraction are explicit: considering only vehicles with initial lanes of 2, 3, 4 and 5; the NGSIM data comprises data of three vehicles, namely a car, a motorcycle and a truck, wherein the data of the car accounts for about 97 percent, so that only the data of the car is extracted as a research object; and thirdly, because 50 frames of lane change data need to be extracted when lane change data are extracted, the amount of front and rear data of a lane change point of a vehicle is more than 25 frames.
b. Setting a ij 、μ i 、∑ i 、ω i Initial value, importing a sample data set, and obtaining a by iteration of Baum-Welch algorithm ij 、μ i 、∑ i 、ω i And obtaining the behavior recognition probability model according to the optimal value of the data.
Inputting the calculated lateral offset and lateral offset speed into a behavior recognition probability model i (O t ) Obtaining the recognition probability p of the target vehicle at the time t Identification Including a first left lane change probability p 1 First probability of keeping lane p 2 And a first right lane change probability p 3 I.e. the probability that the target vehicle is in a left lane change, the probability of being in a lane keeping, and the probability of being in a right lane change at time t.
The calculation process of the expected utility value is already described in embodiment 1, and is not described here. The probability of the left lane change intention, the probability of the lane keeping intention and the probability of the right lane change intention are written into a vector in the form of p Intention to 。
4) P is to be Identification And p Intention to Weighted superposition is carried out to obtain the final prediction probability p Prediction 。
The weighted overlap-add is of the form:
p prediction ＝τ 1 p Identification +τ 2 p Intention to ；
Wherein, tau 1 A weight coefficient for identifying the probability; tau is 2 Is a weight coefficient of the intention probability.
τ 1 And τ 2 Satisfy the relationship ofτ 1 +τ 2 ＝1，τ 1 And τ 2 The different values of (a) reflect whether to believe more about the intended result or believe more about the result of identifying the historical track when predicting the lane change behavior of the target vehicle. For most vehicles keeping lanes, if no factors stimulating the driver to generate lane changing behaviors or the condition of lane changing is not met because of safety problems, the intention presumption result is more believable, so that the behavior recognition result error caused by the fluctuation of the vehicle track can be effectively avoided; and when the factors stimulating the vehicle to change the lane exist and the lane change safety condition is met, the behavior recognition result is more credible. Thus parameter τ 2 The value of (c) can be determined by the probability of intention to keep the lane, which is determined by a cubic spline as shown in fig. 4.
In particular, p Identification Including a first left lane change probability p 1 First probability of keeping lane p 2 And a first right lane change probability p 3 ；p Intention to The lane change intention probability comprises a left lane change intention probability, a lane keeping intention probability and a right lane change intention probability; p is a radical of Prediction The lane change probability comprises a second left lane change probability, a second lane keeping probability and a second right lane change probability. Weighting and superposing the first left lane changing probability and the left lane changing intention probability of the target vehicle to obtain a second left lane changing probability of the target vehicle; weighting and superposing the first lane keeping probability and the lane keeping intention probability of the target vehicle to obtain a second lane keeping probability of the target vehicle; and performing weighted superposition on the first right lane changing probability and the right lane changing intention probability of the target vehicle to obtain a second right lane changing probability of the target vehicle.
5) And comparing the second left lane changing probability, the second lane keeping probability and the second right lane changing probability, wherein the corresponding lane changing behavior with the maximum probability is the predicted lane changing behavior.
According to the method for predicting the lane change behavior of the target vehicle, on the basis of researching the single-vehicle state of the target vehicle, complex traffic scenes such as interaction and game among vehicles in the traffic scene are considered, and the intention of a driver of the target vehicle is modeled, so that the method has good performance in the aspects of predicting the advance degree, reflecting interaction and robustness, and the prediction success probability is higher.
Claims (10)
1. A method for predicting lane change behavior of a motor vehicle, comprising the steps of:
1) according to a revenue function
Calculating an expected utility value of a left lane changing, an expected utility value of a lane keeping and an expected utility value of a right lane changing of a target vehicle;
wherein u is the expected utility value;
is the plantago exercisable space index;
is a collision risk index; c is a comfort index; omega 1 ，ω 2 ，ω 3 Is the corresponding coefficient;
2) comparing the expected utility value of the left lane changing, the expected utility value of the lane keeping and the expected utility value of the right lane changing, wherein the corresponding lane changing behavior with the maximum expected utility value is the predicted lane changing behavior;
the space index of the front feasible vehicle is obtained according to the distance between the target vehicle and the front vehicle; the collision risk index is obtained according to the distance between the target vehicle and other vehicles; the comfort index is obtained from the longitudinal acceleration and the lateral acceleration of the target vehicle.
2. The method of predicting a lane change behavior of a motor vehicle of claim 1, wherein the calculation of the index of space available for vehicle travel comprises:
wherein the content of the first and second substances,
is the index of the space available for driving before the vehicle; d r The distance between the target vehicle and the front vehicle; d v Is the visible distance.
3. The method of predicting a lane change behavior of a motor vehicle of claim 1, wherein the collision risk index is calculated by:
wherein the content of the first and second substances,
is a collision risk index;
representing the minimum value of the distance between the target vehicle and the other vehicle.
4. The method of predicting a lane change behavior of a motor vehicle of claim 1, wherein the comfort index is calculated by:
wherein c is a comfort index; a is x Is the longitudinal acceleration of the target vehicle; a is y Is the lateral acceleration of the target vehicle; t is the prediction time period.
5. The method for predicting a lane change behavior of a motor vehicle according to claim 1, wherein an expected utility value of a left lane change is obtained by predicting a left lane change of a target vehicle and a track of a lane kept by surrounding vehicles; obtaining an expected utility value of a lane keeping through predicting the tracks of the lane keeping of a target vehicle and the lane keeping of surrounding vehicles; and obtaining an expected utility value of right lane changing by predicting the tracks of the right lane changing of the target vehicle and the lane keeping of the surrounding vehicles.
6. A method for predicting lane change behavior of a motor vehicle, comprising the steps of:
acquiring a motion track of a target vehicle; obtaining the lateral offset and the lateral offset speed of the target vehicle relative to the central line of the lane according to the motion track of the target vehicle; inputting the lateral offset and the lateral offset speed into a pre-established behavior recognition probability model to obtain a first left lane changing probability, a first lane keeping probability and a first right lane changing probability of the target vehicle; the behavior recognition probability model is a continuous hidden Markov model;
according to a revenue function
Calculating an expected utility value of a left lane change, an expected utility value of a lane keeping and an expected utility value of a right lane change of a target vehicle; normalizing each expected utility value to obtain corresponding left lane change intention probability, lane keeping intention probability and right lane change intention probability;
wherein u is the expected utility value;
is the plantago exercisable space index;
is a collision risk index; c is a comfort index; omega 1 ，ω 2 ，ω 3 Is the corresponding coefficient;
weighting and superposing the first left lane changing probability and the left lane changing intention probability to obtain a second left lane changing probability; weighting and superposing the first lane keeping probability and the lane keeping intention probability to obtain a second lane keeping probability; weighting and superposing the first right lane changing probability and the right lane changing intention probability to obtain a second right lane changing probability;
and comparing the second left lane changing probability, the second lane keeping probability and the second right lane changing probability, wherein the corresponding lane changing behavior with the maximum probability is the predicted lane changing behavior.
7. The method of predicting a motor vehicle lane-change behavior of claim 6, wherein the probability observed in the continuous hidden Markov model is a probability density function of a Gaussian distribution.
8. The method of predicting a lane change behavior of a motor vehicle of claim 6, wherein the calculation of the index of space available for vehicle travel comprises:
wherein the content of the first and second substances,
is the index of the space available for driving before the vehicle; d r The distance between the target vehicle and the front vehicle; d v Is the visible distance.
9. The method of predicting a lane change behavior of a motor vehicle of claim 6, wherein the collision risk index is calculated by:
wherein the content of the first and second substances,
is a collision risk index;
representing the minimum value of the distance between the target vehicle and the other vehicle.
10. The method of predicting a lane change behavior of a motor vehicle of claim 6, wherein the comfort index is calculated by:
wherein c is a comfort index; a is x Is the longitudinal acceleration of the target vehicle; a is y Is the lateral acceleration of the target vehicle; t is the prediction time period.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110151796.0A CN114852099A (en) | 2021-02-03 | 2021-02-03 | Method for predicting lane changing behavior of motor vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110151796.0A CN114852099A (en) | 2021-02-03 | 2021-02-03 | Method for predicting lane changing behavior of motor vehicle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114852099A true CN114852099A (en) | 2022-08-05 |
Family
ID=82623703
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110151796.0A Pending CN114852099A (en) | 2021-02-03 | 2021-02-03 | Method for predicting lane changing behavior of motor vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114852099A (en) |
Citations (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5521579A (en) * | 1993-04-26 | 1996-05-28 | Mercedes-Benz Ag | Method for providing guiding assistance for a vehicle in changing lane |
| US20130090825A1 (en) * | 2011-10-10 | 2013-04-11 | Mando Corporation | Lane-change assistance system of vehicle and lane-change assistance method thereof |
| CN103754224A (en) * | 2014-01-24 | 2014-04-30 | 清华大学 | Vehicle multi-target coordinating lane changing assisting adaptive cruise control method |
| KR101399026B1 (en) * | 2012-11-23 | 2014-05-27 | 현대엠엔소프트 주식회사 | Method for notification warning when vhicles change lanes system |
| US20140207325A1 (en) * | 2013-01-21 | 2014-07-24 | GM Global Technology Operations LLC | Efficient data flow algorithms for autonomous lane changing, passing and overtaking behaviors |
| US20140309889A1 (en) * | 2013-04-11 | 2014-10-16 | Hyundai Motor Company | Method and system for controlling change of lane |
| CN104442812A (en) * | 2013-10-16 | 2015-03-25 | 陈武强 | Automobile lane change assist detection device and method |
| CN105480229A (en) * | 2015-11-24 | 2016-04-13 | 大连楼兰科技股份有限公司 | Intelligent lane change assisting system and method based on information fusion |
| CN107672592A (en) * | 2017-08-22 | 2018-02-09 | 吉利汽车研究院（宁波）有限公司 | It is a kind of that accessory system is kept based on the track of vehicle active lane change or steering |
| CN107702716A (en) * | 2017-08-31 | 2018-02-16 | 广州小鹏汽车科技有限公司 | A kind of unmanned paths planning method, system and device |
| CN108313054A (en) * | 2018-01-05 | 2018-07-24 | 北京智行者科技有限公司 | The autonomous lane-change decision-making technique of automatic Pilot and device and automatic driving vehicle |
| JP2018177179A (en) * | 2017-04-12 | 2018-11-15 | トヨタ自動車株式会社 | Lane change support device |
| CN109398355A (en) * | 2017-08-18 | 2019-03-01 | 郑州宇通客车股份有限公司 | It is a kind of for lane keeping method and vehicle based on distributed motor-driven vehicle |
| CN109460023A (en) * | 2018-11-09 | 2019-03-12 | 上海理工大学 | Driver's lane-changing intention recognition methods based on Hidden Markov Model |
| US20190100211A1 (en) * | 2017-09-29 | 2019-04-04 | Neusoft Corporation | Vehicle lane-changing control method, vehicle lanechanging control device and related equipment |
| CN109727490A (en) * | 2019-01-25 | 2019-05-07 | 江苏大学 | A kind of nearby vehicle behavior adaptive corrective prediction technique based on driving prediction field |
| CN109835339A (en) * | 2019-03-21 | 2019-06-04 | 北京经纬恒润科技有限公司 | A kind of lane-change decision-making technique and device |
| CN109949611A (en) * | 2019-03-28 | 2019-06-28 | 百度在线网络技术（北京）有限公司 | The lane change method, apparatus and storage medium of unmanned vehicle |
| CN110525436A (en) * | 2019-08-27 | 2019-12-03 | 中国第一汽车股份有限公司 | Vehicle lane-changing control method, device, vehicle and storage medium |
| CN110949391A (en) * | 2019-11-04 | 2020-04-03 | 广州小鹏汽车科技有限公司 | Vehicle lane change control method and system, vehicle and storage medium |
| CN111209838A (en) * | 2019-12-31 | 2020-05-29 | 清华大学 | Driving intention-based dynamic identification method for lane change behavior of surrounding vehicles |
| CN111238520A (en) * | 2020-02-06 | 2020-06-05 | 北京百度网讯科技有限公司 | Lane change path planning method and device, electronic equipment and computer readable medium |
| CN111267846A (en) * | 2020-02-11 | 2020-06-12 | 南京航空航天大学 | Game theory-based peripheral vehicle interaction behavior prediction method |
| CN111361564A (en) * | 2020-04-29 | 2020-07-03 | 吉林大学 | Lane change system considering benefit maximization and comprehensive decision method |
| US20200262448A1 (en) * | 2019-02-19 | 2020-08-20 | Baidu Online Network Technology (Beijing) Co., Ltd. | Decision method, device, equipment in a lane changing process and storage medium |
| CN111559388A (en) * | 2020-03-26 | 2020-08-21 | 吉利汽车研究院（宁波）有限公司 | Target vehicle screening method, device, equipment and storage medium |
| CN111845787A (en) * | 2020-08-03 | 2020-10-30 | 北京理工大学 | Lane change intention prediction method based on LSTM |
| CN111994089A (en) * | 2020-09-02 | 2020-11-27 | 中国科学技术大学 | Driver lane change intention identification method and system based on hybrid strategy game |
| CN111994088A (en) * | 2020-09-02 | 2020-11-27 | 中国科学技术大学 | Driver lane change intention identification method and system based on hybrid strategy game |
| CN112053589A (en) * | 2020-08-18 | 2020-12-08 | 北京航空航天大学 | Target vehicle lane changing behavior adaptive identification model construction method |
| CN112124313A (en) * | 2020-09-29 | 2020-12-25 | 重庆长安汽车股份有限公司 | Target vehicle selection method based on automatic driving lane changing, vehicle and storage medium |
| CN112203916A (en) * | 2018-11-08 | 2021-01-08 | 宝马股份公司 | Method and device for determining lane change related information of target vehicle, method and device for determining vehicle comfort measure for predicting driving maneuver of target vehicle, and computer program |
| CN112258893A (en) * | 2020-09-01 | 2021-01-22 | 华南理工大学 | Intelligent vehicle lane change collision probability assessment method based on track prediction |
-
2021
- 2021-02-03 CN CN202110151796.0A patent/CN114852099A/en active Pending
Patent Citations (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5521579A (en) * | 1993-04-26 | 1996-05-28 | Mercedes-Benz Ag | Method for providing guiding assistance for a vehicle in changing lane |
| US20130090825A1 (en) * | 2011-10-10 | 2013-04-11 | Mando Corporation | Lane-change assistance system of vehicle and lane-change assistance method thereof |
| KR101399026B1 (en) * | 2012-11-23 | 2014-05-27 | 현대엠엔소프트 주식회사 | Method for notification warning when vhicles change lanes system |
| US20140207325A1 (en) * | 2013-01-21 | 2014-07-24 | GM Global Technology Operations LLC | Efficient data flow algorithms for autonomous lane changing, passing and overtaking behaviors |
| US20140309889A1 (en) * | 2013-04-11 | 2014-10-16 | Hyundai Motor Company | Method and system for controlling change of lane |
| CN104442812A (en) * | 2013-10-16 | 2015-03-25 | 陈武强 | Automobile lane change assist detection device and method |
| CN103754224A (en) * | 2014-01-24 | 2014-04-30 | 清华大学 | Vehicle multi-target coordinating lane changing assisting adaptive cruise control method |
| CN105480229A (en) * | 2015-11-24 | 2016-04-13 | 大连楼兰科技股份有限公司 | Intelligent lane change assisting system and method based on information fusion |
| JP2018177179A (en) * | 2017-04-12 | 2018-11-15 | トヨタ自動車株式会社 | Lane change support device |
| CN109398355A (en) * | 2017-08-18 | 2019-03-01 | 郑州宇通客车股份有限公司 | It is a kind of for lane keeping method and vehicle based on distributed motor-driven vehicle |
| CN107672592A (en) * | 2017-08-22 | 2018-02-09 | 吉利汽车研究院（宁波）有限公司 | It is a kind of that accessory system is kept based on the track of vehicle active lane change or steering |
| CN107702716A (en) * | 2017-08-31 | 2018-02-16 | 广州小鹏汽车科技有限公司 | A kind of unmanned paths planning method, system and device |
| US20190100211A1 (en) * | 2017-09-29 | 2019-04-04 | Neusoft Corporation | Vehicle lane-changing control method, vehicle lanechanging control device and related equipment |
| CN108313054A (en) * | 2018-01-05 | 2018-07-24 | 北京智行者科技有限公司 | The autonomous lane-change decision-making technique of automatic Pilot and device and automatic driving vehicle |
| CN112203916A (en) * | 2018-11-08 | 2021-01-08 | 宝马股份公司 | Method and device for determining lane change related information of target vehicle, method and device for determining vehicle comfort measure for predicting driving maneuver of target vehicle, and computer program |
| CN109460023A (en) * | 2018-11-09 | 2019-03-12 | 上海理工大学 | Driver's lane-changing intention recognition methods based on Hidden Markov Model |
| CN109727490A (en) * | 2019-01-25 | 2019-05-07 | 江苏大学 | A kind of nearby vehicle behavior adaptive corrective prediction technique based on driving prediction field |
| US20200262448A1 (en) * | 2019-02-19 | 2020-08-20 | Baidu Online Network Technology (Beijing) Co., Ltd. | Decision method, device, equipment in a lane changing process and storage medium |
| CN109835339A (en) * | 2019-03-21 | 2019-06-04 | 北京经纬恒润科技有限公司 | A kind of lane-change decision-making technique and device |
| CN109949611A (en) * | 2019-03-28 | 2019-06-28 | 百度在线网络技术（北京）有限公司 | The lane change method, apparatus and storage medium of unmanned vehicle |
| CN110525436A (en) * | 2019-08-27 | 2019-12-03 | 中国第一汽车股份有限公司 | Vehicle lane-changing control method, device, vehicle and storage medium |
| CN110949391A (en) * | 2019-11-04 | 2020-04-03 | 广州小鹏汽车科技有限公司 | Vehicle lane change control method and system, vehicle and storage medium |
| CN111209838A (en) * | 2019-12-31 | 2020-05-29 | 清华大学 | Driving intention-based dynamic identification method for lane change behavior of surrounding vehicles |
| CN111238520A (en) * | 2020-02-06 | 2020-06-05 | 北京百度网讯科技有限公司 | Lane change path planning method and device, electronic equipment and computer readable medium |
| CN111267846A (en) * | 2020-02-11 | 2020-06-12 | 南京航空航天大学 | Game theory-based peripheral vehicle interaction behavior prediction method |
| CN111559388A (en) * | 2020-03-26 | 2020-08-21 | 吉利汽车研究院（宁波）有限公司 | Target vehicle screening method, device, equipment and storage medium |
| CN111361564A (en) * | 2020-04-29 | 2020-07-03 | 吉林大学 | Lane change system considering benefit maximization and comprehensive decision method |
| CN111845787A (en) * | 2020-08-03 | 2020-10-30 | 北京理工大学 | Lane change intention prediction method based on LSTM |
| CN112053589A (en) * | 2020-08-18 | 2020-12-08 | 北京航空航天大学 | Target vehicle lane changing behavior adaptive identification model construction method |
| CN112258893A (en) * | 2020-09-01 | 2021-01-22 | 华南理工大学 | Intelligent vehicle lane change collision probability assessment method based on track prediction |
| CN111994089A (en) * | 2020-09-02 | 2020-11-27 | 中国科学技术大学 | Driver lane change intention identification method and system based on hybrid strategy game |
| CN111994088A (en) * | 2020-09-02 | 2020-11-27 | 中国科学技术大学 | Driver lane change intention identification method and system based on hybrid strategy game |
| CN112124313A (en) * | 2020-09-29 | 2020-12-25 | 重庆长安汽车股份有限公司 | Target vehicle selection method based on automatic driving lane changing, vehicle and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112347567B (en) | Vehicle intention and track prediction method | |
| CN111144432B (en) | Method for eliminating fuzzy detection in sensor fusion system | |
| US11794757B2 (en) | Systems and methods for prediction windows for optimal powertrain control | |
| EP3745158B1 (en) | Methods and systems for computer-based determining of presence of dynamic objects | |
| Althoff et al. | Comparison of Markov chain abstraction and Monte Carlo simulation for the safety assessment of autonomous cars | |
| CN109572694A (en) | It is a kind of to consider probabilistic automatic Pilot methods of risk assessment | |
| CN111079590A (en) | Peripheral vehicle behavior pre-judging method of unmanned vehicle | |
| CN111746559B (en) | Method and system for predicting lane changing intention of front vehicle | |
| CN111667512B (en) | Multi-target vehicle track prediction method based on improved Kalman filtering | |
| Peng et al. | Intelligent method for identifying driving risk based on V2V multisource big data | |
| Chae et al. | Design and vehicle implementation of autonomous lane change algorithm based on probabilistic prediction | |
| CN114462667A (en) | SFM-LSTM neural network model-based street pedestrian track prediction method | |
| CN108733042B (en) | Target tracking method and device for automatic driving vehicle | |
| CN115056776A (en) | Adaptive driver following method comprehensively considering perception process and driver behavior | |
| Jeong | Predictive lane change decision making using bidirectional long shot-term memory for autonomous driving on highways | |
| Guo et al. | A fusion estimation of the peak tire–road friction coefficient based on road images and dynamic information | |
| CN110843767A (en) | Spline theory-based automatic parking method and system | |
| EP3845981B1 (en) | Methods and processors for controlling steering of self-driving car | |
| US20220126826A1 (en) | Enhanced vehicle operation | |
| CN117376920A (en) | Intelligent network connection automobile network attack detection, safety state estimation and control method | |
| Choi et al. | Radar-based lane estimation with deep neural network for lane-keeping system of autonomous highway driving | |
| CN114620059A (en) | Automatic driving method and system thereof, and computer readable storage medium | |
| CN114852099A (en) | Method for predicting lane changing behavior of motor vehicle | |
| Guo et al. | Forwarding collision assessment with the localization information using the machine learning method | |
| CN113761715B (en) | Method for establishing personalized vehicle following model based on Gaussian mixture and hidden Markov |
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 |