CN114519931B - Method and device for predicting behavior of target vehicle in intersection environment - Google Patents
Method and device for predicting behavior of target vehicle in intersection environment Download PDFInfo
- Publication number
- CN114519931B CN114519931B CN202011288883.2A CN202011288883A CN114519931B CN 114519931 B CN114519931 B CN 114519931B CN 202011288883 A CN202011288883 A CN 202011288883A CN 114519931 B CN114519931 B CN 114519931B
- Authority
- CN
- China
- Prior art keywords
- target
- target vehicle
- lane
- vehicle
- probability
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
-
- 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
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- 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
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
- G08G1/0962—Arrangements for giving variable traffic instructions having an indicator mounted inside the vehicle, e.g. giving voice messages
- G08G1/0967—Systems involving transmission of highway information, e.g. weather, speed limits
- G08G1/096708—Systems involving transmission of highway information, e.g. weather, speed limits where the received information might be used to generate an automatic action on the vehicle control
- G08G1/096725—Systems involving transmission of highway information, e.g. weather, speed limits where the received information might be used to generate an automatic action on the vehicle control where the received information generates an automatic action on the vehicle control
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
- G08G1/0962—Arrangements for giving variable traffic instructions having an indicator mounted inside the vehicle, e.g. giving voice messages
- G08G1/0968—Systems involving transmission of navigation instructions to the vehicle
- G08G1/096833—Systems involving transmission of navigation instructions to the vehicle where different aspects are considered when computing the route
- G08G1/096844—Systems involving transmission of navigation instructions to the vehicle where different aspects are considered when computing the route where the complete route is dynamically recomputed based on new data
-
- 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
-
- 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
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Theoretical Computer Science (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Game Theory and Decision Science (AREA)
- Tourism & Hospitality (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Mathematical Physics (AREA)
- Atmospheric Sciences (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention belongs to the technical field of automatic driving target vehicle prediction, and particularly relates to a method and a device for predicting target vehicle behaviors in an intersection environment. Firstly, extracting characteristics of a target vehicle capable of driving towards each target lane according to state information of the target vehicle, further calculating the probability corresponding to each characteristic, and performing weighted summation on each characteristic and the corresponding weight to obtain the characteristic probability; then determining the road relation probability according to the states of traffic lights or the attribute relation between lanes; multiplying the road relation probability and the characteristic probability to obtain the comprehensive probability of each target lane; and finally, selecting the target lane with the highest comprehensive probability as an intention lane of the target vehicle so as to predict the running track of the target vehicle. When the target vehicle running track is predicted, the self motion condition of the target vehicle and the conditions of surrounding lanes and traffic lights are considered, so that the predicted track of the predicted target vehicle is closer to the actual track.
Description
Technical Field
The invention belongs to the technical field of automatic driving target vehicle prediction, and particularly relates to a method and a device for predicting target vehicle behaviors in an intersection environment.
Background
With the development of various technologies such as computer technology, 5G communication technology, and sensor technology, the autonomous vehicle is gradually becoming a development trend of future vehicles. An autonomous vehicle is a vehicle that can travel safely and automatically by techniques such as artificial intelligence techniques without any human active operation. The automatic driving brings convenience to life of people, but frequent traffic accidents also make people realize that the automatic driving has some safety risks, and the safety and the trafficability of the automatic driving vehicle are always important in the field of automatic driving vehicles.
Particularly, in some intersection regions, as shown in fig. 1, road conditions are complex, and the intersection regions are areas with multiple traffic accidents, although traffic signals, stop signs and the like can effectively solve traffic flows, the performances of road safety and traffic efficiency are still unsatisfactory, how to enable the automatic driving vehicle to obtain more accurate driving decision at the intersection becomes a key problem concerned at present, and the method has important significance for improving the automatic driving user experience and the initiative safety.
The Chinese patent with application publication number CN110275531A discloses a method for predicting the track of an obstacle, which comprises the steps of firstly determining the positioning information of unmanned equipment and each obstacle at the last moment of a target moment, and determining a global interaction vector (which is used for representing the common influence factors of the unmanned equipment and a plurality of obstacles on the movement) at the last moment based on the positioning information so as to predict the track point of each obstacle at the target moment. According to the method, only the common influence factors of the unmanned equipment and the obstacles on the movement are considered, but the influence of the movement track of the obstacles in the actual environment is not considered, so that the movement track of the obstacles is not accurately predicted, and the situation that the self-vehicle can avoid the obstacles according to the planned track of the movement track of the obstacles cannot be ensured.
Disclosure of Invention
The invention provides a method and a device for predicting target vehicle behaviors in an intersection environment, which are used for solving the problem that the track of a target vehicle cannot be predicted inaccurately in the prior art.
In order to solve the technical problem, the technical scheme of the invention comprises the following steps:
the invention provides a target vehicle behavior prediction method under a crossing environment, which comprises the following steps:
1) When a vehicle is about to arrive at an intersection, acquiring state information of a target vehicle within a set range of the intersection; the target vehicle is a vehicle about to enter the intersection and drive away from the intersection;
2) Extracting at least two characteristics that the target vehicle can travel towards each target lane according to the state information of the target vehicle; the target lane is at least two lanes connected at the intersection;
3) For one target lane, calculating probabilities corresponding to the characteristics that the target vehicle can drive towards the target lane according to the characteristics that the target vehicle can drive towards the target lane and the characteristics that the target vehicle can drive towards other target lanes; weighting and summing each feature and the corresponding weight to obtain the feature probability that the target vehicle can drive towards the target lane;
judging whether the state of the traffic signal lamp viewed by the vehicle can be acquired:
if the traffic signal light state is obtained, the traffic signal light state of the target vehicle entering the intersection is deduced according to the traffic signal light state seen by the vehicle, whether the target vehicle can drive from the lane to which the target vehicle is located to the target lane is determined according to the deduced traffic signal light state of the target vehicle entering the intersection, the road relation probability of the target vehicle which can drive to the target lane under the traffic signal light state is obtained through calculation, and the road relation probability is multiplied by the corresponding characteristic probability, so that the comprehensive probability of the target vehicle which can drive to the target lane is obtained;
if the target vehicle can not be obtained, calculating the road relation probability that the target vehicle can drive towards the target lane under the state of the unavailable traffic light according to the attribute relation between the lane where the target vehicle is located and the target lane, and multiplying the road relation probability by the corresponding characteristic probability to obtain the comprehensive probability that the target vehicle can drive towards the target lane; the attribute relation is the relation of whether the vehicle can enter another lane from one lane according to the road traffic rule;
4) Calculating the comprehensive probability that the target vehicle can drive towards other target lanes according to the method in the step 3); selecting a target lane with higher comprehensive probability as an intention lane of a target vehicle; and predicting the running track of the target vehicle according to the lane where the target vehicle is located and the intended lane, so as to plan the running track of the self vehicle according to the running track of the target vehicle.
The beneficial effects of the above technical scheme are: the method comprises the steps of discretizing possible action behaviors of a target vehicle at an intersection, firstly determining each characteristic that the target vehicle can drive towards a target lane according to state information of the target vehicle, and calculating the probability corresponding to each characteristic according to the characteristics; besides the characteristics, determining the road relation probability according to the state of a traffic signal lamp or the attribute relation between lanes; combining the characteristic probability and the road relation probability to obtain the comprehensive probability of the target vehicle towards each target lane; and then, the target lane which the target vehicle is most likely to travel towards can be determined, and after the travel track of each target vehicle is predicted, the route of the vehicle can be planned, so that the travel route avoiding the target vehicles is planned. When the target vehicle driving track is predicted, various factors are considered, the predicted track of the target vehicle obtained through prediction is closer to the actual track according to the characteristics obtained according to the self state information of the target vehicle and the surrounding traffic signal lamp conditions or road attribute relations, so that the path planned by the vehicle is more reasonable and safer, the vehicle can be ensured to avoid the target vehicles at intersections with complex road conditions, and the method can be applied to various scenes.
Further, in order to accurately predict the travel track of the target vehicle, the state information of the target vehicle includes a history track of the target vehicle; the feature that the target vehicle can travel toward the respective target lanes includes a first feature that: calculating included angles between the historical track of the target vehicle and the center lines of all the target lanes, wherein each included angle is a first characteristic corresponding to each target lane; the probability that the target vehicle may travel toward the target lane corresponding to the first feature is:
wherein, P c1_m A probability corresponding to the first feature that the target vehicle can travel toward the target lane m; theta m As historical track and purpose of target vehicleMarking the included angle of the center line of the lane m; n represents the total number of target lanes.
Further, in order to accurately predict the running track of the target vehicle, the state information of the target vehicle comprises historical motion state information of the target vehicle; the feature that the target vehicle can travel toward the respective target lanes includes a second feature that: calculating a characteristic angle corresponding to each historical moment of the target vehicle, wherein the characteristic angle is an included angle difference between a vector formed by the historical track point and the key point of the target lane and a historical speed vector; calculating the characteristic angle variation corresponding to the adjacent historical moments, wherein the obtained characteristic angle variation is the second characteristic corresponding to each target lane; the probability that the target vehicle may travel toward the target lane corresponding to the second feature is:
wherein, P agl_m A probability corresponding to the second feature that the target vehicle can travel toward the target lane m;the characteristic angle variation corresponding to the jth historical moment under the target lane m; k is the total number of the target vehicles at the historical time; k-1 is the total number of the characteristic angle variation; n represents the total number of target lanes; the function f (x) is:
further, in order to accurately predict the running track of the target vehicle, the state information of the target vehicle comprises historical motion state information of the target vehicle; the feature that the target vehicle can travel toward the respective target lanes includes a third feature that: calculating the distance between each historical moment of the target vehicle and each key point of the target lane, and calculating the distance difference corresponding to the adjacent historical moments, wherein the obtained distance difference is a third characteristic corresponding to each target lane; the probability that the target vehicle may travel toward the target lane corresponding to the third feature is:
wherein, P dis_m A probability corresponding to the third feature that the target vehicle can travel toward the target lane m;obtaining the distance difference between the target vehicle and the m key points of the target lane at the jth adjacent historical moment; k is the total number of the target vehicles at the historical moment; k-1 is the total number of distance differences; n represents the total number of the target lanes; the function f (x) is:
further, in order to accurately predict the running track of the target vehicle, the state information of the target vehicle includes current motion state information of the target vehicle; the feature that the target vehicle can travel toward the respective target lanes includes a fourth feature that: calculating included angles between the current position of the target vehicle and the center lines of all the target lanes and distances between the current position of the target vehicle and key points of all the target lanes, wherein all the included angles and the distances are fourth features corresponding to all the target lanes; the probability that the target vehicle may travel toward the target lane corresponding to the fourth feature is:
wherein, P x_m A probability corresponding to the fourth feature that the target vehicle can travel toward the target lane m; dis' m The distance between the current position of the target vehicle and the key point m of the target lane is taken as the distance; theta' m The included angle between the current position of the target vehicle and the center line m of the target lane is set; n represents the total number of the target lanes; the softmax function is:
further, the probability of the relationship between the road where the target vehicle can travel toward the target lane in the unavailable traffic light state is as follows:
wherein, P road_m The probability of the road relation that the target lane m can drive towards the target lane in the traffic signal lamp state cannot be obtained; l is the number of target lanes having an exit attribute corresponding to the lane in which the target vehicle is located; p is a radical of thre Is a designed probability threshold.
Further, the probability of the relationship between the road, which can be obtained when the target vehicle can drive towards the target lane under the traffic light state, is as follows:
wherein, P light_m The road relation probability that the target lane m can drive towards the target lane in the traffic light state can be obtained; r is the number of target lanes having an exit attribute corresponding to the lane in which the target vehicle is located; p is a radical of formula thre Is a designed probability threshold.
Furthermore, the key point of the target lane is the intersection point of the stop line of the target lane at the intersection and the central line of the target lane.
The invention also provides a device for predicting the target vehicle behavior in the intersection environment, which comprises a memory and a processor, wherein the processor is used for executing the instructions stored in the memory to realize the introduced method for predicting the target vehicle behavior in the intersection environment.
Drawings
FIG. 1 is a schematic illustration of an intersection traffic environment;
FIG. 2 is a schematic view of the road relationship of the present invention;
FIG. 3 is a flow chart of a method of predicting the behavior of a target vehicle in an intersection environment of the present invention;
FIG. 4 is a schematic illustration of a cluster of predicted trajectories of a pre-generated target vehicle of the present invention;
fig. 5 is a structural diagram of a target vehicle behavior prediction apparatus in an intersection environment of the present invention;
FIG. 6 is a schematic illustration of the characteristic angle definition of the present invention.
Detailed Description
The basic idea of the invention is as follows: the present invention is directed to predicting the behavior of a target vehicle arriving at an intersection of an automatically driven vehicle (hereinafter, referred to as a self vehicle) in the vicinity of the intersection, in view of the phenomenon that the behavior of the vehicle at the intersection is complicated. The method specifically comprises the following steps: acquiring state information of a target vehicle, and extracting characteristics S related to crossing driving behaviors and used for judging that the vehicle can drive towards each target lane according to the state information of the target vehicle i (ii) a And then calculating to obtain the feature probability P corresponding to each feature i Combining the weights omega corresponding to the feature probabilities i Obtaining the comprehensive probability P of the target vehicle driving from the lane to each target lane when reaching the intersection all_m (m =1,2, \8230;, N), wherein N is the total number of the target lanes, and the target lane with the highest probability is selected as the intended lane of the target vehicle; and predicting the running track of the target vehicle according to the lane where the target vehicle is located and the intended lane of the target vehicle. After the driving tracks of the target vehicles are predicted, the driving tracks of the vehicles can be planned to avoid the target vehicles.
The following describes a method for predicting a target vehicle behavior in an intersection environment and a device for predicting a target vehicle behavior in an intersection environment in detail with reference to the drawings and embodiments.
The method comprises the following steps:
in the embodiment of the method for predicting the target vehicle behavior in the intersection environment, provided that the vehicle is about to arrive at the intersection shown in fig. 2, 16 lanes, namely lane 1, lane 2, \8230 \ lane 15 and lane 16, are connected at the intersection, the whole method flow is shown in fig. 3, and the process is as follows:
step one, determining target vehicles near an intersection to which the own vehicle is about to arrive, wherein the target vehicles comprise MB1 located in a lane 13, MB2 located in a lane 9 and MB3 located in a lane 8. The vehicles have not yet entered the intersection and are ready to exit the intersection. Next, how to predict the behavior of the target vehicle will be described with reference to MB1 as an example, starting from step two.
Step two, acquiring motion state information of the MB1, wherein the motion state information comprises current state information and historical state information of the MB 1; judging whether the vehicle can obtain the state of the traffic light viewed by the road where the vehicle is located: when the traffic light state can be obtained, the traffic light state seen by the MB1 when the MB1 enters the intersection is presumed according to the traffic light state seen by the own vehicle; when the vehicle can not be acquired, determining the attribute relationship between the lane (lane 13) where the MB1 is located and other lanes (lanes except the lane 13) connected with the intersection, wherein the attribute relationship is the relationship whether the vehicle can enter into another lane from one lane or not according to the road traffic rule; specifically, the method comprises the following steps:
1) Historical state information of the MB1 is acquired, and the historical state information includes a historical position and a historical speed of the MB 1.
The historical state information reflects the overall trend of the vehicle motion, so the invention needs to acquire the historical state information of the target vehicle. The state information of MB1 at time t is X t =(x t ,y t ,v t ),x t 、y t Is a geodetic coordinate value v of the target vehicle at time t t The speed of the target vehicle at time t. Assuming that history state information in k consecutive states is used, a history state information series of X = { X = can be obtained 1 ,X 2 ,X 3 ,…,X k }。
2) Current state information of MB1 is obtained, which includes the current position and current velocity of MB 1.
Although the historical state information reflects the overall trend of the target vehicle, the historical state information is only one type of historical motor knowledge of the target vehicle toward the target, and the state of the target vehicle is largely determined by the current state information. The current state information of MB1 is:[v x ,v y ,x,y] T 。
3) And determining the attribute relationship of the lane where the MB1 is located and other lanes connected with the intersection, wherein the attribute relationship comprises two relationships of passable and non-communicable.
Lanes are attributed, with attributes that a lane has including: an entrance attribute and an exit attribute; further excavation is carried out, and each lane also has attributes of left turning, right turning, straight going and the like; for the crossroad, different lanes have the relation of passability or not. For the intersection shown in fig. 1, the road link relationship matrix M obtained according to whether the roads can be driven is:
wherein m is i_j The relationship that whether the lane i to the lane j can pass or not is represented, and the specific definition is as follows:
4) And presuming the traffic light state of the road where the MB1 is located according to the traffic light state of the road where the own vehicle is located.
For the state of fig. 1, the relationship between the traffic light state viewed on the road where the vehicle is located and the traffic light state viewed on the road where MB1 is located in lane 1 is shown in table 1. If the own vehicle is in another lane, the same method is used to estimate that, for example, if the lane where the own vehicle is located is lane 16, the relationship between the traffic light state viewed on the road where the own vehicle is located and the traffic light state viewed on the road where MB1 is located is the same as the content in table 1.
TABLE 1 relationship between traffic light status and vehicle
Step three, determining N lanes (including normal lanes and irregular lanes) which are possible to run by MB1) As shown in fig. 1, the lane 2, the lane 8230the lane 8230, the lane 16, which are called target lanes, are respectively the target lane 1, the target lane 2, the lane 8230the lane 8230and the target lane 16. And for each target lane, extracting features according to various information acquired or calculated in the step two, wherein the extracted features comprise four features which are respectively as follows: historical track dip characteristic S 1 (first feature), historical course angular variation feature S 2 (second feature), historical distance change feature S 3 (third feature) and current state information feature S 4 (fourth feature). The following description specifically describes how to define and calculate these features, wherein the key point of each target lane is defined as the intersection point of the stop line of the target lane at the intersection and the center line of the target lane, and the key points of the target lanes 1,2, 8230are defined, D1, D2, 8230are defined, D16 are defined, and the positions are (x) and (x) are defined 1 ,y 1 )、(x 2 ,y 2 )、……、(x 16 ,y 16 )。
1) Extracting historical track inclination angle characteristic S of MB1 1 Including the historical track inclination angle characteristic S corresponding to the target lane 1 1_1 Historical track inclination angle characteristic S corresponding to target lane 2 1_2 "\8230; \ 8230;, and historical track dip angle feature S corresponding to target lane 16 1_16 。
According to the historical state information of the MB1, fitting the historical track of the MB1 into a linear straight line by using a least square method, namely fitting the historical track, and calculating the road inclination angle theta between the fitted historical track and the central line of each target lane i I =1,2, \ 8230, N, N denotes the total number of target lanes, where N =16. I.e. S 1_1 Is theta 1 ,S 1_2 Is theta 2 And so on.
2) Extracting historical course angle change characteristic S of MB1 2 And historical distance change characteristic S 3 ,S 2 Comprises historical course angle change characteristics S corresponding to the target lane 1 2_1 And historical course angle change characteristics S corresponding to the target lane 2 2_2 "\8230;, and the historical heading angle change characteristic S corresponding to the target lane 16 2_16 ,S 3 Comprises a first part and a second partHistorical distance change characteristic S corresponding to marking lane 1 3_1 Historical distance change characteristic S corresponding to target lane 2 3_2 "\8230; \ 8230;, and the historical distance change characteristic S corresponding to the target lane 16 3_16 。
For a natural language in which a target vehicle travels from a starting point toward a certain target point (assuming that point a is the target point and point B is the initial point), the following are: and when the target vehicle approaches the point A and is far away from the point B, the included angle of the target vehicle towards the point A is gradually reduced or unchanged, and the included angle far away from the point B is increased or unchanged. Therefore, the angle change characteristic and the distance change characteristic at the time points adjacent to the history of MB1 can be obtained from the fitting history trajectory of MB 1.
For the target lane 1, calculating the difference of included angles between vectors formed by all historical moments of the MB1 and the key point D1 and historical speed vectors, wherein the difference of the included angles is a characteristic angle corresponding to all historical moments of the target vehicle, and the difference of the included angles is respectively as follows: agl1_1, agl1_2, \8230, 8230, agl1_ k-1, agl1_ k, and the distances from the key point D1 are respectively as follows: l1_1, L1_2, \8230;, L1_ k-1, L1_ k, respectively, the corresponding adjacent historical course angle variations (characteristic angle variations) are The corresponding historical distance variation amounts are respectively: />I.e. the historical heading angle change sequence ≥ of MB1>Is a historical course angle change characteristic S 2_1 (ii) a Deriving a historical distance change sequence @forMB 1>For historical distance change characteristics S 3_1 . Wherein, the definition of the characteristic angle can be seen in fig. 6, wherein the vector formed by the historical track point and the key point, and the historical speedThe degree vectors are respectively shown as vectors enclosed by a dashed frame in fig. 6, and the characteristic angle is an included angle formed by the two.
For the target lane 2, the included angles between each historical moment of the MB1 and the key point D2 are respectively as follows: agl2_1, agl2_2, \8230;, agl2_ k-1, agl2_ k, the distances from the key point D1 are respectively: l1_1, L1_2, L8230, L1_ k-1, and L1_ k respectively have adjacent historical course angle change quantities The corresponding historical distance variation amounts are respectively: /> I.e. the historical heading angle change sequence ≥ of MB1>Is a historical course angle change characteristic S 2_2 (ii) a Obtaining a historical distance change sequence for MB1For historical distance change characteristics S 3_2 。
For other target lanes, the calculation is performed in the same way as in the previous two paragraphs.
3) Determining the Pre-State information characteristic S of MB1 4 I.e., [ v ] x ,v y ,x,y] T 。
Step four, for a target lane, calculating the probability corresponding to each feature of the target lane according to each feature determined in step three, wherein the probabilities are respectively as follows:
features of inclination angle S with historical track 1 Corresponding historical global change probability P c1 (including P) c1_1 、P c1_2 、……、P c1_N ) And historical headingAngular variation characteristic S 2 Corresponding historical angle change probability P agl (including P) agl_1 、P agl_2 、……、P agl_16 ) And a characteristic S of distance variation from history 3 Corresponding historical distance change probability P dis (including P) dis_1 、P dis_2 、……、P dis_16 ) And with the previous status information characteristic S 4 Corresponding current state intention probability P x (including P) x_1 、P x_2 、……、P x_16 ). The following is specifically described:
1) Determining a characteristic S of inclination angle with the historical track according to the following formula 1_m Corresponding historical global change probability P c1_m :
Wherein, P c1_m Is the historical overall change probability of the target lane m (m =1,2, \8230;, or N); theta m An included angle between the historical track of the target vehicle and the center line m of the target lane is set; n represents the total number of target lanes; the softmax function is:
2) Determining the angular variation characteristic S of the historical course according to the following formula 2_m Corresponding historical angle change probability P agl_m :
Wherein, P agl_m Historical angle change probabilities for target lane m (m =1,2, \8230;, or N);the included angle difference between the target vehicle and the m key points of the target lane at the jth adjacent historical moment is obtained; k is the total number of the target vehicles at the historical time; k-1 is the total number of the included angle differences(ii) a N represents the total number of target lanes; the function f (x) is defined as follows:
3) Determining a distance change characteristic S from history according to the following formula 3_m Corresponding historical distance change probability P dis_m :
Wherein, P dis_m A historical distance change probability for target lane m (m =1,2, \8230;, or N);obtaining the distance difference between the target vehicle and the m key points of the target lane at the jth adjacent historical moment; k is the total number of the target vehicles at the historical moment; k-1 is the total number of distance differences; n represents the total number of target lanes.
4) Determining a pre-state information feature S 4_m Corresponding current state intention probability P x_m 。
Current state information v from MB1 x ,v y ,x,y] T Firstly, a feature angle corresponding to the current time of the MB1 is calculated, wherein the feature angle corresponding to the current time is an included angle difference theta 'between a vector formed by the current position and the key point D1 and the current speed vector' i ;
Then, the distances between MB1 and the key points are calculated as:
then the sequence between the current state information of MB1 and the respective lane key points is:
θ′={θ′ 1 ,θ′ 1 ,……,θ′ N }
dis′={dis′ 1 ,dis′ 2 ,……,dis′ N }
finally, the current state intention probability P of the MB1 is calculated according to the following formula x_m :
Wherein, P x_m A current state intent probability for lane m (m =1,2, \8230;, or N); dis' m The distance between the current position of the target vehicle and the key point m of the target lane is taken as the distance; theta' m The included angle between the current position of the target vehicle and the center line m of the target lane is set; n represents the total number of target lanes.
5) In the case where the traffic light state viewed from the vehicle is not available, the road relationship probability P of the road relationship probability MB1 that the target vehicle m can travel toward the target lane in the traffic light state is determined according to the following formula road_m :
Wherein, P road_m The probability of the road relation that the target lane m can drive towards the target lane in the traffic light state cannot be obtained; l is the number of target lanes with an exit attribute corresponding to the lane in which the target vehicle is located, and l =6 in this embodiment; p is a radical of thre The probability threshold for the design is typically low, but indicates that the target vehicle still has a possibility to travel to other lanes in violation.
6) Under the condition that the state of the traffic light seen by the vehicle can be obtained, the road relation probability P of the road relation probability MB1 that the target vehicle m can drive towards the target lane under the state of the traffic light can be obtained is determined according to the following formula light_m :
Wherein, P light_m The target lane m can be faced to the target lane in the state of acquiring traffic lightsA road relation probability of travel; r is the number of target lanes having exit attributes corresponding to lanes where the target vehicles are located, and traffic light states of roads determine attributes of left-turn, right-turn, straight-going and the like of the vehicles, but the traffic light states when the vehicles enter the intersections cannot be completely determined, so that the probability is determined only by combining table 1 aiming at the condition that the lanes can be determined before the vehicles enter the intersections, and the number of the traveling vehicles in the current state of MB1 is r; p is a radical of thre The probability threshold for the design is typically low, but indicates that the target vehicle is still likely to drive in the other lane violations.
Step five, calculating the comprehensive probability P of the target lane m (m =1,2, \8230;, N) according to the plurality of probabilities calculated in the step four all_m . Obtaining the comprehensive probability P of each target lane matrix ={P all_1 ,P all_2 ,……,P all_N Get the target lane P corresponding to the maximum integrated probability max =maxP matrix 。
When the traffic light state when the MB1 enters the intersection cannot be judged:
P all_m =P road_m *(ω 1 *P c1_m +ω 2 *P agl_m +ω 3 *P dis_m +ω 4 *P x_m )
when the traffic light state when the MB1 enters the intersection can be judged:
P all_m =P light_m *(ω 1 *P c1_m +ω 2 *P agl_m +ω 3 *P dis_m +ω 4 *P x_m )
wherein, P all_m Is the integrated probability of the target lane m; omega 1 、ω 2 、ω 3 、ω 4 The weights are respectively corresponding to the probabilities, are fixed parameters, and are adjusted according to multiple simulation experiments.
And step six, predicting the running track of the target vehicle MB1 according to the lane where the target vehicle MB1 is located and the intended lane.
The target vehicle predicted track is generated by adopting a B-spline, wherein the calculation formula of the B-spline curve is as follows:
in the formula, P i Is the ith control point; p represents the total number of control points; b i,deg (t) is a formula of a basic function table.
The target vehicle generates a plurality of predicted trajectories at each moment, as shown in fig. 4, forming a trajectory cluster representing a path that the target vehicle may travel.
And seventhly, predicting the running track of other target vehicles at the intersection according to the method for other target vehicles including MB2 and MB3. After the driving tracks of the target vehicles are predicted, the driving tracks of the vehicles can be planned to avoid the target vehicles.
Thus, the traveling locus prediction of the target vehicle and the own vehicle can be completed. The method discretizes the possible action behaviors of the target vehicle at the intersection, expresses the possible action behaviors by a plurality of characteristics, considers the conditions of surrounding traffic lights or road attribute relationship before predicting the track, and enables the predicted track of the target vehicle to be closer to the actual track, so that the path planned by the vehicle is more reasonable and safer, the vehicle can be ensured to avoid the target vehicles at the intersection with complex road conditions, and the method can be applied to various scenes.
The embodiment of the device is as follows:
the embodiment provides a target vehicle behavior prediction device in a crossing environment, as shown in fig. 5, which includes a memory, a processor and an internal bus, wherein the processor and the memory are communicated with each other through the internal bus.
The processor can be a microprocessor MCU, a programmable logic device FPGA and other processing devices.
The memory can be various memories for storing information by using an electric energy mode, such as RAM, ROM and the like; various memories for storing information by magnetic energy, such as a hard disk, a floppy disk, a magnetic tape, a core memory, a bubble memory, a usb disk, etc.; various types of memory that store information optically, such as CDs, DVDs, etc., are used. Of course, there are other types of memory, such as quantum memory, graphene memory, and the like.
The processor may invoke logic instructions in the memory to implement a method of predicting behavior of a target vehicle in an intersection environment. The method is described in detail in the method examples.
Claims (9)
1. A method for predicting the behavior of a target vehicle in an intersection environment is characterized by comprising the following steps:
1) When a vehicle is about to arrive at an intersection, acquiring state information of a target vehicle within a set range of the intersection; the target vehicle is a vehicle about to enter the intersection and drive away from the intersection;
2) Extracting at least two features that the target vehicle can travel towards each target lane according to the state information of the target vehicle, wherein the features comprise: historical track inclination angle characteristics, historical course angle change characteristics, historical distance change characteristics and current state information characteristics; the target lane is at least two lanes connected at the intersection;
3) For one target lane, calculating probabilities corresponding to the characteristics that the target vehicle can drive towards the target lane according to the characteristics that the target vehicle can drive towards the target lane and the characteristics that the target vehicle can drive towards other target lanes; weighting and summing each feature and the corresponding weight to obtain the feature probability that the target vehicle can drive towards the target lane;
judging whether the state of the traffic signal lamp viewed by the vehicle can be acquired:
if the traffic signal light state is obtained, the traffic signal light state of the target vehicle entering the intersection is deduced according to the traffic signal light state seen by the vehicle, whether the target vehicle can drive from the lane to which the target vehicle is located to the target lane is determined according to the deduced traffic signal light state of the target vehicle entering the intersection, the road relation probability of the target vehicle which can drive to the target lane under the traffic signal light state is obtained through calculation, and the road relation probability is multiplied by the corresponding characteristic probability, so that the comprehensive probability of the target vehicle which can drive to the target lane is obtained;
if the target vehicle can not be obtained, calculating the road relation probability that the target vehicle can drive towards the target lane under the state of the unavailable traffic light according to the attribute relation between the lane where the target vehicle is located and the target lane, and multiplying the road relation probability by the corresponding characteristic probability to obtain the comprehensive probability that the target vehicle can drive towards the target lane; the attribute relation is the relation of whether the vehicle can enter another lane from one lane according to the road traffic rule;
4) Calculating the comprehensive probability that the target vehicle can drive towards other target lanes according to the method in the step 3); selecting a target lane with higher comprehensive probability as an intention lane of a target vehicle; and predicting the running track of the target vehicle according to the lane where the target vehicle is located and the intended lane, so as to plan the running track of the self vehicle according to the running track of the target vehicle.
2. The crossing environment behavior prediction method of claim 1, wherein the state information of the target vehicle comprises a historical track of the target vehicle; the features that the target vehicle can travel toward the respective target lanes include a first feature that: calculating included angles between the historical track of the target vehicle and the center lines of all the target lanes, wherein each included angle is a first feature corresponding to each target lane; the probability that the target vehicle may travel toward the target lane corresponding to the first feature is:
wherein, P c1_m A probability corresponding to the first feature that the target vehicle can travel toward the target lane m; theta.theta. m An included angle between the historical track of the target vehicle and the center line of the target lane m is set; n represents the total number of target lanes.
3. The method of predicting behavior of a target vehicle in an intersection environment of claim 1, wherein the state information of the target vehicle comprises historical movement state information of the target vehicle; the feature that the target vehicle can travel toward the respective target lanes includes a second feature that: calculating a characteristic angle corresponding to each historical moment of the target vehicle, wherein the characteristic angle is an included angle difference between a vector formed by the historical track point and the key point of the target lane and the historical speed vector; calculating the characteristic angle variation corresponding to the adjacent historical moments, wherein the obtained characteristic angle variation is the second characteristic corresponding to each target lane; the probability that the target vehicle may travel toward the target lane corresponding to the second feature is:
wherein, P agl_m A probability corresponding to the second feature that the target vehicle can travel toward the target lane m;the characteristic angle variation corresponding to the jth historical moment under the target lane m; k is the total number of the target vehicles at the historical moment; k-1 is the total number of the characteristic angle variation; n represents the total number of target lanes; the function f (x) is:
4. the crossing environment behavior prediction method of a target vehicle as claimed in claim 1, wherein the state information of the target vehicle comprises historical movement state information of the target vehicle; the feature that the target vehicle can travel toward the respective target lanes includes a third feature that: calculating the distance between each historical moment of the target vehicle and each key point of the target lane, and calculating the distance difference corresponding to the adjacent historical moments, wherein the obtained distance difference is a third characteristic corresponding to each target lane; the probability that the target vehicle may travel toward the target lane corresponding to the third feature is:
wherein, P dis_m A probability corresponding to the third feature that the target vehicle can travel toward the target lane m;obtaining the distance difference between the target vehicle and the m key points of the target lane at the jth adjacent historical moment; k is the total number of the target vehicles at the historical time; k-1 is the total number of distance differences; n represents the total number of the target lanes; the function f (x) is:
5. the method of predicting behavior of a target vehicle in an intersection environment of claim 1, wherein the state information of the target vehicle comprises current motion state information of the target vehicle; the feature that the target vehicle can travel toward the respective target lanes includes a fourth feature that: calculating included angles between the current position of the target vehicle and the center lines of all the target lanes and distances between the current position of the target vehicle and key points of all the target lanes, wherein the included angles and the distances are fourth features corresponding to all the target lanes; the probability that the target vehicle may travel toward the target lane corresponding to the fourth feature is:
wherein, P x_m A probability corresponding to the fourth feature that the target vehicle can travel toward the target lane m; dis' m The distance between the current position of the target vehicle and the key point m of the target lane is taken as the distance; theta' m An included angle between the current position of the target vehicle and the center line m of the target lane is set; n represents the total number of the target lanes; the softmax function is:
6. the method for predicting the behavior of the target vehicle under the intersection environment according to any one of claims 1 to 5, wherein the probability of the road relationship that the target vehicle can travel towards the target lane under the unavailable traffic light state is:
wherein, P road_m The road relation probability that the target lane m can drive towards the target lane in the traffic light state cannot be obtained; l is the number of target lanes having an exit attribute corresponding to the lane in which the target vehicle is located; p is a radical of formula thre A designed probability threshold; n represents the total number of target lanes.
7. The method for predicting the behavior of the target vehicle under the intersection environment according to any one of claims 1 to 5, wherein the probability of obtaining the relationship between the roads, in which the target vehicle can travel towards the target lane under the traffic light state, is as follows:
wherein, P light_m The road relation probability that the target lane m can drive towards the target lane in the traffic light state can be obtained; r is the number of target lanes having an exit attribute corresponding to the lane in which the target vehicle is located; p is a radical of thre A probability threshold for the design; n represents the total number of target lanes.
8. The method for predicting the behavior of a target vehicle under an intersection environment according to any one of claims 3 to 5, wherein the key point of the target lane is an intersection of a stop line of the target lane at the intersection and a center line of the target lane.
9. An intersection environment target vehicle behavior prediction device, characterized by comprising a memory and a processor, wherein the processor is used for executing instructions stored in the memory to realize the intersection environment target vehicle behavior prediction method according to any one of claims 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011288883.2A CN114519931B (en) | 2020-11-17 | 2020-11-17 | Method and device for predicting behavior of target vehicle in intersection environment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011288883.2A CN114519931B (en) | 2020-11-17 | 2020-11-17 | Method and device for predicting behavior of target vehicle in intersection environment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114519931A CN114519931A (en) | 2022-05-20 |
CN114519931B true CN114519931B (en) | 2023-04-07 |
Family
ID=81594254
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011288883.2A Active CN114519931B (en) | 2020-11-17 | 2020-11-17 | Method and device for predicting behavior of target vehicle in intersection environment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114519931B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115346370B (en) * | 2022-08-10 | 2023-11-03 | 重庆大学 | Intersection anti-collision system and method based on intelligent traffic |
CN115447599B (en) * | 2022-09-22 | 2023-06-27 | 上海保隆汽车科技股份有限公司 | Automatic early warning method, device and equipment for vehicle steering and readable storage medium |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4985232B2 (en) * | 2007-08-27 | 2012-07-25 | トヨタ自動車株式会社 | Collision determination device |
JP5900454B2 (en) * | 2013-10-09 | 2016-04-06 | トヨタ自動車株式会社 | Vehicle lane guidance system and vehicle lane guidance method |
US9409570B2 (en) * | 2014-05-09 | 2016-08-09 | Toyota Motor Engineering & Manufacturing North America, Inc. | Method and apparatus for predicting most probable path of vehicle travel and vehicle control loss preview |
KR20190035159A (en) * | 2017-09-26 | 2019-04-03 | 삼성전자주식회사 | Vehicle motion prediction method and apparatus |
CN107591011B (en) * | 2017-10-31 | 2020-09-22 | 吉林大学 | Intersection traffic signal self-adaptive control method considering supply side constraint |
CN108399740B (en) * | 2018-01-22 | 2020-04-28 | 华南理工大学 | Method for predicting collision probability of motor vehicles at signalized intersection |
CN108022026A (en) * | 2018-01-25 | 2018-05-11 | 戴姆勒股份公司 | A kind of traffic security early warning method of traffic control, device and the system including the device |
DE102018203063A1 (en) * | 2018-03-01 | 2019-09-05 | Honda Motor Co., Ltd. | Collision risk-prediction unit |
CN110969852A (en) * | 2019-12-09 | 2020-04-07 | 上海宝康电子控制工程有限公司 | Method for realizing real-time prediction and investigation and control processing of public security based on control vehicle running path |
-
2020
- 2020-11-17 CN CN202011288883.2A patent/CN114519931B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN114519931A (en) | 2022-05-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11243532B1 (en) | Evaluating varying-sized action spaces using reinforcement learning | |
US11970168B2 (en) | Vehicle trajectory modification for following | |
WO2022052406A1 (en) | Automatic driving training method, apparatus and device, and medium | |
CN110796856B (en) | Vehicle lane change intention prediction method and training method of lane change intention prediction network | |
US20230124864A1 (en) | Graph Representation Querying of Machine Learning Models for Traffic or Safety Rules | |
CN110789528B (en) | Vehicle driving track prediction method, device, equipment and storage medium | |
US20200174490A1 (en) | Neural networks for vehicle trajectory planning | |
CN111583715B (en) | Vehicle track prediction method, vehicle collision early warning method, device and storage medium | |
JP2022544715A (en) | Modeling and predicting yielding behavior | |
Chae et al. | Virtual target-based overtaking decision, motion planning, and control of autonomous vehicles | |
CN115343947A (en) | Motion control decision for autonomous vehicles | |
JP2022546283A (en) | Occupancy prediction neural network | |
CN114519931B (en) | Method and device for predicting behavior of target vehicle in intersection environment | |
CN112947406A (en) | FLOYD and Astar-based hybrid path planning method | |
CN114511999B (en) | Pedestrian behavior prediction method and device | |
CN112327888B (en) | Path planning method and device, electronic equipment and storage medium | |
KR20210118995A (en) | Method and apparatus for generating u-turn path of autonomous vehicle based on deep learning | |
Guo et al. | Toward human-like behavior generation in urban environment based on Markov decision process with hybrid potential maps | |
Althoff et al. | Stochastic reachable sets of interacting traffic participants | |
CN115062202A (en) | Method, device, equipment and storage medium for predicting driving behavior intention and track | |
CN116259185A (en) | Vehicle behavior decision method and device fusing prediction algorithm in parking lot scene | |
AU2021229267A1 (en) | Detecting a construction zone by a lead autonomous vehicle (AV) and updating routing plans for following Autonomous vehicle (AVs) | |
US11436504B1 (en) | Unified scene graphs | |
CN117104272A (en) | Intelligent driving method, system, vehicle and storage medium | |
CN111310919A (en) | Driving control strategy training method based on scene segmentation and local path planning |
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 | ||
CB02 | Change of applicant information | ||
CB02 | Change of applicant information |
Address after: No. 6, Yutong Road, Guancheng Hui District, Zhengzhou, Henan 450061 Applicant after: Yutong Bus Co.,Ltd. Address before: No.1, Shibali Heyu Road, Guancheng Hui District, Zhengzhou City, Henan Province Applicant before: ZHENGZHOU YUTONG BUS Co.,Ltd. |
|
GR01 | Patent grant | ||
GR01 | Patent grant |