CN114863685B - Traffic participant trajectory prediction method and system based on risk acceptance degree - Google Patents

Traffic participant trajectory prediction method and system based on risk acceptance degree Download PDF

Info

Publication number
CN114863685B
CN114863685B CN202210784856.7A CN202210784856A CN114863685B CN 114863685 B CN114863685 B CN 114863685B CN 202210784856 A CN202210784856 A CN 202210784856A CN 114863685 B CN114863685 B CN 114863685B
Authority
CN
China
Prior art keywords
traffic participant
traffic
time
moment
graph model
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
Application number
CN202210784856.7A
Other languages
Chinese (zh)
Other versions
CN114863685A (en
Inventor
龚建伟
文晨旭
李子睿
臧政
吕超
吴绍斌
齐建永
何刚
冯悦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huidong Planet Beijing Technology Co ltd
Beijing Institute of Technology BIT
Original Assignee
Huidong Planet Beijing Technology Co ltd
Beijing Institute of Technology BIT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Huidong Planet Beijing Technology Co ltd, Beijing Institute of Technology BIT filed Critical Huidong Planet Beijing Technology Co ltd
Priority to CN202210784856.7A priority Critical patent/CN114863685B/en
Publication of CN114863685A publication Critical patent/CN114863685A/en
Application granted granted Critical
Publication of CN114863685B publication Critical patent/CN114863685B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • 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
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Human Resources & Organizations (AREA)
  • Evolutionary Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Analytical Chemistry (AREA)
  • Chemical & Material Sciences (AREA)
  • Marketing (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Probability & Statistics with Applications (AREA)
  • Educational Administration (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention relates to a traffic participant trajectory prediction method and system based on risk acceptance degree, comprising the following steps: for collecting different traffic participants in a target traffic scenet‑MIs at the momenttTrack information of the moment is obtained, and the track information is preprocessed; different said traffic participants including pedestrians, bicycles and motor vehicles; for is totClustering the preprocessing track information of the moment, and determining the preprocessing track information according to the clustering resulttA risk acceptance level for each of the transportation participants at a time; by usingt‑MIs at the momenttThe preprocessed track information of a time andttraining a heterogeneous graph model according to the risk acceptance degree of each traffic participant at any moment to obtain a trained heterogeneous graph model; and predicting the track of each traffic participant by using the trained heterogeneous graph model. The method and the system have the advantages that different risk acceptance degrees of different traffic participants are considered, and the future tracks of the traffic participants can be accurately predicted.

Description

Traffic participant trajectory prediction method and system based on risk acceptance degree
Technical Field
The invention relates to the technical field of automobile intelligent interaction, in particular to a traffic participant trajectory prediction method and system based on risk acceptance degree.
Background
At present, with the increasing number of urban road traffic participants, road condition information is becoming more and more complex. The method simulates the behavior of the traffic participants in the complex urban environment, accurately expresses the interactive behavior of the traffic participants, and is important for realizing safe and socially acceptable motion planning of the automatic driving automobile. However, accurate representation of such interactive behavior remains a challenge due to the uncertainty of the movement of the traffic participants and the complexity of the urban environment.
At present, there are many methods for modeling the interaction behavior of traffic participants, but these methods have some respective disadvantages. The uncertainty problem of bicycles and pedestrians with a prediction time of more than 1 second cannot be handled by the physics-based motion model; the "two-phase" prediction framework cannot simultaneously consider interactions between traffic participants and predict future behavior; the "social awareness" approach applies a pool mechanism and concatenation operations to directly fuse the features of the interacting participants, but without interpretability. Meanwhile, the method does not consider different cognitions of different traffic participants on the same interactive scene and different judgments on risks. Therefore, the existing traffic participant interactive behavior prediction methods cannot accurately predict future tracks, and the traffic participant track prediction method and the traffic participant track prediction system based on the risk acceptance degree are provided.
Disclosure of Invention
The invention aims to provide a traffic participant trajectory prediction method and system based on risk acceptance degree, which can accurately predict the trajectory of a traffic participant.
In order to achieve the purpose, the invention provides the following scheme:
a traffic participant trajectory prediction method based on risk acceptance degrees comprises the following steps:
for collecting different traffic participants in a target traffic scenet-MTime of dayTo tTrack information of the moment is obtained, and the track information is preprocessed; different said traffic participants including pedestrians, bicycles and motor vehicles;
to pairtClustering the preprocessing track information of the moment, and determining the preprocessing track information according to the clustering resulttA risk acceptance level for each of the transportation participants at a time;
by usingt-MIs at the momenttThe preprocessed trajectory information of a time andttraining a heterogeneous graph model according to the risk acceptance degree of each traffic participant at any moment to obtain a trained heterogeneous graph model;
and predicting the track of each traffic participant by using the trained heterogeneous graph model.
A system for predicting a trajectory of a transportation participant based on risk acceptance, comprising:
the track information acquisition and processing module is used for acquiring different traffic participants in a target traffic scenet-MIs at the momenttTrack information of the moment is obtained, and the track information is preprocessed; different said transportation participants include pedestrians, bicycles and motor vehicles;
a clustering module for pairingtClustering the preprocessing track information of the moment, and determining the preprocessing track information according to the clustering resulttA risk acceptance level for each of the transportation participants at a time;
a heterogeneous graph model training module for utilizingt-MIs at the momenttThe preprocessed trajectory information of a time andttraining a heterogeneous graph model according to the risk acceptance degree of each traffic participant at any moment to obtain a trained heterogeneous graph model;
and the track prediction module is used for predicting the track of each traffic participant by utilizing the trained heterogeneous graph model.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention provides a traffic participant trajectory prediction method and system based on risk acceptance degree, which comprises the following steps: for collecting different traffic participants in a target traffic scenet-MIs at the momenttTrack information of the moment is obtained, and the track information is preprocessed; different said traffic participants including pedestrians, bicycles and motor vehicles; to pairtClustering the preprocessing track information of the moment, and determining the preprocessing track information according to the clustering resulttA risk acceptance level for each of the transportation participants at a time; by usingt-MIs at the momenttThe preprocessed trajectory information of a time andttraining a heterogeneous graph model according to the risk acceptance degree of each traffic participant at any moment to obtain a trained heterogeneous graph model; and predicting the track of each traffic participant by using the trained heterogeneous graph model. Different risk acceptance degrees of different traffic participants are considered, namely different cognitions of different traffic participants on the same interaction scene and different judgment on dangers are considered, so that the future track of the traffic participants can be accurately predicted.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a flowchart of a traffic participant trajectory prediction method based on risk acceptance degree according to embodiment 1 of the present invention;
FIG. 2 shows the estimated time to collision provided in embodiment 1 of the present inventionTTCSchematic diagram of (1);
fig. 3 is a block diagram of a traffic participant trajectory prediction system based on risk acceptance according to embodiment 2 of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a traffic participant trajectory prediction method and system based on risk acceptance degree, which can accurately predict the trajectory of a traffic participant.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Example 1
As shown in fig. 1, the embodiment provides a traffic participant trajectory prediction method based on risk acceptance degree, including:
s1, collecting different traffic participants in the target traffic scenet-MTime of flightIs carved to tTrack information of the moment is obtained, and the track information is preprocessed; different said traffic participants including pedestrians, bicycles and motor vehicles; the target traffic scenario may be different urban traffic scenarios including highways, town blocks, crossroads, and the like.
The preprocessing of the track information in step S1 specifically includes:
determining a target traffic participant from the different traffic participants in the target traffic scene;
determining a relative position, a relative speed and a collision estimated time of each non-target traffic participant with the target traffic participant; the pre-processing trajectory information includes the relative position, the relative velocity, and the estimated time of collision.
The expression for the relative position is:
Figure 140188DEST_PATH_IMAGE001
(ii) a Wherein the content of the first and second substances,iis shown asiA traffic participant;
Figure 310269DEST_PATH_IMAGE002
and
Figure 737708DEST_PATH_IMAGE003
denotes the firstiThe horizontal and vertical coordinates of each traffic participant relative to the target traffic participant;
the expression for the relative velocity is:
Figure 344139DEST_PATH_IMAGE004
Figure 148015DEST_PATH_IMAGE005
wherein the content of the first and second substances,
Figure 438183DEST_PATH_IMAGE006
the time difference between two adjacent moments;
the expression for the estimated time to collision is:
Figure 302102DEST_PATH_IMAGE007
wherein the content of the first and second substances,
Figure 412141DEST_PATH_IMAGE008
is around the target traffic participantiA non-target traffic participant intThe time of impact at the moment of time,
Figure 19708DEST_PATH_IMAGE009
is the relative velocity
Figure 429961DEST_PATH_IMAGE010
In relative position
Figure 464782DEST_PATH_IMAGE011
Projection of (2). FIG. 2 shows the estimated time of collisionTTCSchematic diagram of (a).
S2, pairtPreprocessing track information of time of day
Figure 327696DEST_PATH_IMAGE012
Clustering is carried out, and the determination is carried out according to the clustering resulttA risk acceptance level for each of the transportation participants at a time; specifically, it can be expressed as:
Figure 4534DEST_PATH_IMAGE013
(ii) a Wherein the content of the first and second substances,
Figure 721823DEST_PATH_IMAGE014
representing the degree of risk acceptance resulting from the clustering,
Figure 412698DEST_PATH_IMAGE015
a representative clustering method is a method of clustering,Sthe feature vector extracted in step S1 ((S))tPre-processing trace information for a time of day),
Figure 277755DEST_PATH_IMAGE016
for approximating the parameters, whether the result is expected or not is judged by comparing the loss in the calculation process with the approximation parameters. The risk acceptance degree of each traffic participant can be understood as the distance between the cluster where the traffic participant is located and the target traffic participant, if the distance between the traffic participant and the host vehicle is short, the risk acceptance degree of the traffic participant is considered to be large, and the situation can also be understood as the cognitive situation and the danger judgment situation of the traffic participant in the interactive scene.
In step S2, clustering the preprocessing track information at time t by using a kernel principal component analysis method in combination with a K-means clustering algorithm;
the Kernel Principal Component Analysis (KPCA) method first maps by non-linearity
Figure 243437DEST_PATH_IMAGE017
Data to be recorded
Figure 877550DEST_PATH_IMAGE018
Conversion to high dimensional space
Figure 739326DEST_PATH_IMAGE019
Using PCA in high dimensional space
Figure 826100DEST_PATH_IMAGE019
Mapping to another low dimensional space
Figure 329894DEST_PATH_IMAGE020
And finally dividing the samples by a linear classifier.tThe preprocessing track information of the time is converted through KPCA and is clustered into K groups by minimizing the square sum in the clusters
Figure 818513DEST_PATH_IMAGE021
。
The target function expression of the K-means clustering algorithm is as follows:
Figure 851191DEST_PATH_IMAGE022
wherein, the first and the second end of the pipe are connected with each other,
Figure 628523DEST_PATH_IMAGE023
representing the clustering result when the latter equation reaches the minimum
Figure 185275DEST_PATH_IMAGE021
；
Figure 482395DEST_PATH_IMAGE024
Is shown askClustering;
Figure 731980DEST_PATH_IMAGE025
representing calculating the square sum in the cluster;
Figure 754466DEST_PATH_IMAGE020
is formed by kernel principalAfter conversion of the analysistPreprocessing track information of the moment;
Figure 600062DEST_PATH_IMAGE026
is the firstkCluster center of individual clusters.
S3, use oft-MIs at the momenttThe preprocessed trajectory information of a time andttraining a heterogeneous graph model according to the risk acceptance degree of each traffic participant at any moment to obtain a trained heterogeneous graph model;
the heterogeneous graph model constructs an instance layer containing each traffic participant, and each traffic participant is an instance node
Figure 266536DEST_PATH_IMAGE027
. At the moment of timetEach traffic participant can be regarded as an example node
Figure 155863DEST_PATH_IMAGE028
Example node characteristicsfThe method comprises the following steps:
Figure 658520DEST_PATH_IMAGE029
wherein the content of the first and second substances,
Figure 557075DEST_PATH_IMAGE030
is the relative position of the instance node with respect to the host vehicle (target traffic participant),ris the degree of risk acceptance resulting from the clustering,
Figure 343634DEST_PATH_IMAGE031
is the firstiThe type of individual traffic participant.
The heterogeneous graph model constructs a class layer containing traffic participants with similar movement modes according to the dynamic characteristics (such as speed, acceleration and the like) of different traffic participants, wherein each movement mode is a class node
Figure 607125DEST_PATH_IMAGE032
Each one of the classesTraffic participants under other nodes have similar dynamic characteristics. The different movement modes can be embodied in the present embodiment as pedestrians, bicycles and motor vehicles.
The relationship between the instance level and the class level is that the model is applied to all instance nodes having the same type
Figure 128236DEST_PATH_IMAGE033
Integration and embedding is performed by slave instance nodes
Figure 564903DEST_PATH_IMAGE034
To category node
Figure 487859DEST_PATH_IMAGE033
The edge guide passes information from the instance layer to the category layer. After computation in the class layer, class nodes
Figure 718989DEST_PATH_IMAGE035
Sending processed valuable information to instance nodes
Figure 727397DEST_PATH_IMAGE036
。
Space edge
Figure 233333DEST_PATH_IMAGE037
And a time edge
Figure 276376DEST_PATH_IMAGE038
The role of (1) is to establish interaction relationships between nodes. At the moment of timetExample node
Figure 960298DEST_PATH_IMAGE039
And
Figure 705269DEST_PATH_IMAGE040
the spatial relationship between the two participants (the spatial relationship between the two participants is derived from the relative position relationship and the relative velocity relationship in step S1) can be described as
Figure 31208DEST_PATH_IMAGE041
Specifically, characteristics of each instance node
Figure 928757DEST_PATH_IMAGE042
The spatial relationship can be expressed as
Figure 501690DEST_PATH_IMAGE043
And
Figure 750268DEST_PATH_IMAGE044
. In the same way, the time edge
Figure 598007DEST_PATH_IMAGE045
May represent the temporal precedence correlation of the same instance node between adjacent frames.
It is further found that step S3 specifically includes:
step S31, taking each traffic participant as an example node, and taking the relative position, the risk acceptance degree and the traffic participant type of the traffic participant as the characteristics of the example node;
step S32, taking the motion mode of each traffic participant as a category node; the motion mode comprises a walking mode, a riding mode and a driving mode;
step S33, taking the space relation between the traffic participants as the space edge, and taking the time correlation of one traffic participant at two adjacent moments as the time edge;
step S34, constructing the heterogeneous graph model according to the instance nodes, the category nodes, the spatial edges and the temporal edges;
Figure 615642DEST_PATH_IMAGE046
。
wherein the content of the first and second substances,
Figure 641367DEST_PATH_IMAGE034
and
Figure 360930DEST_PATH_IMAGE035
as an exampleA node and a category node, wherein,
Figure 763093DEST_PATH_IMAGE037
and
Figure 618922DEST_PATH_IMAGE038
spatial edges and temporal edges.
Step S35, thet-MIs at the momenttThe relative position sum of each of the traffic participants at a timetAnd inputting the risk acceptance degree of each traffic participant into the heterogeneous graph model at any moment to obtain the trained heterogeneous graph model. Specifically, it can be expressed as:
Figure 549969DEST_PATH_IMAGE047
。
wherein the content of the first and second substances,
Figure 756828DEST_PATH_IMAGE048
and
Figure 697102DEST_PATH_IMAGE049
are respectively traffic participants
Figure 158171DEST_PATH_IMAGE050
To
Figure 306124DEST_PATH_IMAGE051
Relative position sequence of time instants and slave
Figure 485433DEST_PATH_IMAGE052
TotThe sequence of relative positions of the time of day,
Figure 229398DEST_PATH_IMAGE053
represents a supervised heterogeneous graph model,
Figure 325399DEST_PATH_IMAGE054
are the weights of the nodes and edges in the graph neural network. The formula means: will be provided with
Figure 863827DEST_PATH_IMAGE052
TotRelative position sequence of time of day
Figure 61590DEST_PATH_IMAGE049
Clustering derived risk acceptance
Figure 327356DEST_PATH_IMAGE055
And weight
Figure 28595DEST_PATH_IMAGE054
Training to obtain a heterogeneous graph model
Figure 987193DEST_PATH_IMAGE053
The model can predict
Figure 672252DEST_PATH_IMAGE050
To
Figure 492441DEST_PATH_IMAGE056
Position of time of day
Figure 563034DEST_PATH_IMAGE057
。
And S4, predicting the track of each traffic participant by using the trained heterogeneous graph model.
In the embodiment, an interactive behavior modeling method based on risk acceptance is provided, and the risk acceptance of a plurality of traffic participants in an interactive scene is clustered, then re-expressed in a heterogeneous graph model, and finally used for trajectory prediction of the traffic participants. The interactive behavior modeling method considers the individual cognition of different traffic participants to danger, better accords with the real driving process, has good practicability, and can accurately predict the future track of the traffic participants.
Example 2
As shown in fig. 3, the embodiment provides a traffic participant trajectory prediction system based on risk acceptance degree, including:
a track information acquisition and processing module T1 for collectingOf different traffic participants in a target traffic scenariot-MIs at the momenttTrack information of the moment is obtained, and the track information is preprocessed; different said traffic participants including pedestrians, bicycles and motor vehicles;
the trajectory information acquisition and processing module T1 includes:
a target traffic participant determination unit, configured to determine a target traffic participant from different traffic participants in the target traffic scene;
the characteristic extraction unit is used for determining the relative position, the relative speed and the collision estimation time of each non-target traffic participant and the target traffic participant; the pre-processing trajectory information includes the relative position, the relative velocity, and the estimated time of collision.
A clustering module T2 for pairingtClustering the preprocessing track information of the moment, and determining the preprocessing track information according to the clustering resulttA risk acceptance level for each of the transportation participants at a time;
a heterogeneous graph model training module T3 for utilizingt-MIs at the momenttThe preprocessed trajectory information of a time andttraining a heterogeneous graph model according to the risk acceptance degree of each traffic participant at any moment to obtain a trained heterogeneous graph model;
the heterogeneous map model training module T3 includes:
an example node determination unit, configured to take each of the transportation participants as an example node, and take the relative positions of the transportation participants, the risk acceptance degrees and the transportation participant types as features of the example node;
the class node determining unit is used for taking the motion mode of each traffic participant as a class node; the motion mode comprises a walking mode, a riding mode and a driving mode;
a spatial edge and time edge determining unit, configured to use a spatial relationship between the traffic participants as a spatial edge, and use a time correlation of one of the traffic participants at two adjacent moments as a time edge;
the heterogeneous graph model building unit is used for building the heterogeneous graph model according to the instance nodes, the category nodes, the space edges and the time edges;
a heterogeneous graph model training unit for training a heterogeneous graph modelt-MIs at the momenttThe relative position sum of each of the traffic participants at a timetAnd inputting the risk acceptance degree of each traffic participant into the heterogeneous graph model at any moment to obtain the trained heterogeneous graph model.
And the track prediction module T4 is used for predicting the track of each traffic participant by using the trained heterogeneous graph model.
For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims (6)

1. A traffic participant trajectory prediction method based on risk acceptance degree is characterized by comprising the following steps:
for collecting different traffic participants in a target traffic scenet-M time instants totTrack information of the moment, and preprocessing the track information; different said traffic participants including pedestrians, bicycles and motor vehicles; the pre-processing track information comprises a relative position, a relative speed and a collision estimation time;
to pairtClustering the preprocessing track information of the moment, and determining the preprocessing track information according to the clustering resulttA risk acceptance level for each of the transportation participants at a time; the risk acceptance degree is the cognitive condition of the traffic participant in an interactive scene and the judgment condition of the risk;
by usingt-M time instants totThe preprocessed trajectory information of a time andttraining a heterogeneous graph model according to the risk acceptance degree of each traffic participant at any moment to obtain the trained heterogeneous graph model, and specifically comprising the following steps:
taking each traffic participant as an example node, and taking the relative position, the risk acceptance degree and the traffic participant type of the traffic participant as the characteristics of the example node;
taking the motion mode of each traffic participant as a category node; the motion mode comprises a walking mode, a riding mode and a driving mode;
taking the space relation between the traffic participants as a space edge, and taking the time correlation of one traffic participant at two adjacent moments as a time edge;
constructing the heterogeneous graph model according to the instance nodes, the category nodes, the spatial edges and the temporal edges;
will be provided witht-M time instants totThe relative position sum of each of the traffic participants at a timetInputting the risk acceptance degree of each traffic participant into the heterogeneous graph model at any moment to obtain the trained heterogeneous graph model;
and predicting the track of each traffic participant by using the trained heterogeneous graph model.
2. The method according to claim 1, wherein the preprocessing the trajectory information specifically comprises:
determining a target traffic participant from the different traffic participants in the target traffic scene;
a relative position, a relative speed, and a collision estimated time of each non-target traffic participant with the target traffic participant are determined.
3. The method of claim 2, wherein the expression of the relative position is:
Figure DEST_PATH_IMAGE001
(ii) a Wherein the content of the first and second substances,iis shown asiA traffic participant;
Figure DEST_PATH_IMAGE002
and
Figure DEST_PATH_IMAGE003
is shown asiThe horizontal and vertical coordinates of each traffic participant relative to the target traffic participant;
the expression for the relative velocity is:
Figure DEST_PATH_IMAGE004
Figure DEST_PATH_IMAGE005
wherein, the first and the second end of the pipe are connected with each other,
Figure DEST_PATH_IMAGE006
the time difference between two adjacent moments;
the expression for the estimated time to collision is:
Figure DEST_PATH_IMAGE007
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE008
is around the target traffic participantiA non-target traffic participant in
Figure DEST_PATH_IMAGE009
The time of impact at the moment of time,
Figure DEST_PATH_IMAGE010
is the relative velocity
Figure DEST_PATH_IMAGE011
In relative position
Figure DEST_PATH_IMAGE012
Projection of (2).
4. The method of claim 1, wherein the kernel principal component analysis is combined with a K-means clustering algorithm pairtClustering the preprocessing track information at the moment;
the target function expression of the K-means clustering algorithm is as follows:
Figure DEST_PATH_IMAGE013
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE014
representing the clustering result when the latter equation reaches the minimum
Figure DEST_PATH_IMAGE015
；
Figure DEST_PATH_IMAGE016
Is shown askClustering;
Figure DEST_PATH_IMAGE017
representing calculating the square sum in the cluster;
Figure DEST_PATH_IMAGE018
after transformation by kernel principal component analysistPreprocessing track information of the moment;
Figure DEST_PATH_IMAGE019
is the firstkCluster center of individual clusters.
5. System of methods according to any of claims 1 to 4, characterized in that it comprises:
the track information acquisition and processing module is used for acquiring different traffic participants in a target traffic scenet-M time instants totTrack information of the moment is obtained, and the track information is preprocessed; different said traffic participants including pedestrians, bicycles and motor vehicles; the pre-processing trajectory information comprises the relative position, the relative velocity, and the estimated time of collision;
a clustering module for pairingtClustering the preprocessing track information of the moment, and determining the preprocessing track information according to the clustering resulttThe risk acceptance level of each traffic participant at the moment;
a heterogeneous graph model training module for utilizingt-MIs at the momenttThe preprocessed trajectory information of a time andttraining a heterogeneous graph model according to the risk acceptance degree of each traffic participant at any moment to obtain a trained heterogeneous graph model;
the heterogeneous graph model training module comprises:
an example node determination unit, configured to take each of the transportation participants as an example node, and take the relative positions of the transportation participants, the risk acceptance degrees and the transportation participant types as features of the example node;
the class node determining unit is used for taking the motion mode of each traffic participant as a class node; the motion modes comprise a walking mode, a riding mode and a driving mode;
a spatial edge and time edge determining unit, configured to use a spatial relationship between the traffic participants as a spatial edge, and use a time correlation of one of the traffic participants at two adjacent moments as a time edge;
the heterogeneous graph model building unit is used for building the heterogeneous graph model according to the instance nodes, the category nodes, the space edges and the time edges;
the heterogeneous graph model training unit is used for inputting the relative position of each traffic participant from the time t-M to the time t and the risk acceptance degree of each traffic participant at the time t into the heterogeneous graph model to obtain the trained heterogeneous graph model;
and the track prediction module is used for predicting the track of each traffic participant by utilizing the trained heterogeneous graph model.
6. The system of claim 5, wherein the trajectory information acquisition and processing module comprises:
a target traffic participant determination unit, configured to determine a target traffic participant from different traffic participants in the target traffic scene;
and the characteristic extraction unit is used for determining the relative position, the relative speed and the collision estimation time of each non-target traffic participant and the target traffic participant.
CN202210784856.7A 2022-07-06 2022-07-06 Traffic participant trajectory prediction method and system based on risk acceptance degree Active CN114863685B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210784856.7A CN114863685B (en) 2022-07-06 2022-07-06 Traffic participant trajectory prediction method and system based on risk acceptance degree

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210784856.7A CN114863685B (en) 2022-07-06 2022-07-06 Traffic participant trajectory prediction method and system based on risk acceptance degree

Publications (2)

Publication Number Publication Date
CN114863685A CN114863685A (en) 2022-08-05
CN114863685B true CN114863685B (en) 2022-09-27

Family

ID=82626097

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210784856.7A Active CN114863685B (en) 2022-07-06 2022-07-06 Traffic participant trajectory prediction method and system based on risk acceptance degree

Country Status (1)

Country Link
CN (1) CN114863685B (en)

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107610464A (en) * 2017-08-11 2018-01-19 河海大学 A kind of trajectory predictions method based on Gaussian Mixture time series models
WO2018122585A1 (en) * 2016-12-30 2018-07-05 同济大学 Method for urban road traffic incident detecting based on floating-car data
CN109927719A (en) * 2017-12-15 2019-06-25 百度在线网络技术（北京）有限公司 A kind of auxiliary driving method and system based on barrier trajectory predictions
CN111340855A (en) * 2020-03-06 2020-06-26 电子科技大学 Road moving target detection method based on track prediction
CN112364997A (en) * 2020-12-08 2021-02-12 北京三快在线科技有限公司 Method and device for predicting track of obstacle
CN112487905A (en) * 2020-11-23 2021-03-12 北京理工大学 Method and system for predicting danger level of pedestrian around vehicle
WO2021179838A1 (en) * 2020-03-10 2021-09-16 支付宝(杭州)信息技术有限公司 Prediction method and system based on heterogeneous graph neural network model
CN114299607A (en) * 2021-12-13 2022-04-08 南京理工大学 Human-vehicle collision risk degree analysis method based on automatic driving of vehicle

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018122585A1 (en) * 2016-12-30 2018-07-05 同济大学 Method for urban road traffic incident detecting based on floating-car data
CN107610464A (en) * 2017-08-11 2018-01-19 河海大学 A kind of trajectory predictions method based on Gaussian Mixture time series models
CN109927719A (en) * 2017-12-15 2019-06-25 百度在线网络技术（北京）有限公司 A kind of auxiliary driving method and system based on barrier trajectory predictions
CN111340855A (en) * 2020-03-06 2020-06-26 电子科技大学 Road moving target detection method based on track prediction
WO2021179838A1 (en) * 2020-03-10 2021-09-16 支付宝(杭州)信息技术有限公司 Prediction method and system based on heterogeneous graph neural network model
CN112487905A (en) * 2020-11-23 2021-03-12 北京理工大学 Method and system for predicting danger level of pedestrian around vehicle
CN112364997A (en) * 2020-12-08 2021-02-12 北京三快在线科技有限公司 Method and device for predicting track of obstacle
CN114299607A (en) * 2021-12-13 2022-04-08 南京理工大学 Human-vehicle collision risk degree analysis method based on automatic driving of vehicle

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于车辆视角数据的行人轨迹预测与风险等级评定;张哲雨 等;《汽车工程》;20220531;第44卷(第5期);第675-683页 *
张哲雨 等.基于车辆视角数据的行人轨迹预测与风险等级评定.《汽车工程》.2022,第44卷(第5期),第675-683页. *

Also Published As

Publication number Publication date
CN114863685A (en) 2022-08-05

Similar Documents

Publication Publication Date Title
Yang et al. Top-view trajectories: A pedestrian dataset of vehicle-crowd interaction from controlled experiments and crowded campus
CN112700470B (en) Target detection and track extraction method based on traffic video stream
WO2022206942A1 (en) Laser radar point cloud dynamic segmentation and fusion method based on driving safety risk field
Fernando et al. Deep inverse reinforcement learning for behavior prediction in autonomous driving: Accurate forecasts of vehicle motion
CN111860274B (en) Traffic police command gesture recognition method based on head orientation and upper half skeleton characteristics
CN114970321A (en) Scene flow digital twinning method and system based on dynamic trajectory flow
CN111027430B (en) Traffic scene complexity calculation method for intelligent evaluation of unmanned vehicles
Niranjan et al. Deep learning based object detection model for autonomous driving research using carla simulator
Wei et al. Survey of connected automated vehicle perception mode: from autonomy to interaction
CN111178286B (en) Gesture track prediction method and device and electronic equipment
CN115662166B (en) Automatic driving data processing method and automatic driving traffic system
CN115984537A (en) Image processing method and device and related equipment
CN111402632A (en) Risk prediction method for pedestrian movement track at intersection
Del Ser et al. Introduction to the special issue on online learning for big-data driven transportation and mobility
CN114863685B (en) Traffic participant trajectory prediction method and system based on risk acceptance degree
Das et al. Why slammed the brakes on? auto-annotating driving behaviors from adaptive causal modeling
CN114912719B (en) Heterogeneous traffic individual trajectory collaborative prediction method based on graph neural network
Li A scenario-based development framework for autonomous driving
CN116501820A (en) Vehicle track prediction method, device, equipment and storage medium
CN114492544B (en) Model training method and device and traffic incident occurrence probability evaluation method and device
CN116310748A (en) Automatic driving scene recovery and automatic driving prototype testing method and system
Ge et al. Theory and method of data collection for mixed traffic flow based on image processing technology
Yao et al. TL-Detector: Lightweight Based Real-Time Traffic Light Detection Model for Intelligent Vehicles
Oh et al. Towards defensive autonomous driving: Collecting and probing driving demonstrations of mixed qualities
Yuan et al. Real-time long-range road estimation in unknown environments

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant