CN110968101A - Unmanned vehicle behavior decision method based on ontology and Bayesian network - Google Patents
Unmanned vehicle behavior decision method based on ontology and Bayesian network Download PDFInfo
- Publication number
- CN110968101A CN110968101A CN201911317451.7A CN201911317451A CN110968101A CN 110968101 A CN110968101 A CN 110968101A CN 201911317451 A CN201911317451 A CN 201911317451A CN 110968101 A CN110968101 A CN 110968101A
- Authority
- CN
- China
- Prior art keywords
- unmanned vehicle
- class
- ontology
- bayesian network
- behavior decision
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000012545 processing Methods 0.000 claims abstract description 6
- 230000006399 behavior Effects 0.000 claims description 62
- 238000006243 chemical reaction Methods 0.000 claims description 17
- 230000003542 behavioural effect Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 7
- 230000009471 action Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 210000001072 colon Anatomy 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000007476 Maximum Likelihood Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 229920002160 Celluloid Polymers 0.000 description 1
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides an unmanned vehicle behavior decision-making method based on an ontology and a Bayesian network, which comprises the following steps: acquiring a driving scene of an unmanned vehicle; setting a corresponding class set according to a driving scene; constructing corresponding object attributes according to the class set and carrying out instantiation processing; constructing an ontology model corresponding to the driving scene according to the object attributes and the instantiated class set; constructing a Bayesian network model corresponding to the driving scene according to the ontology model; and generating a behavior decision of the unmanned vehicle according to the Bayesian network model. The invention can improve the real-time performance and the accuracy of behavior decision, thereby improving the safety of the unmanned vehicle.
Description
Technical Field
The invention relates to the technical field of automatic driving, in particular to a decision-making method for unmanned vehicle behaviors based on ontology and Bayesian network.
Background
The unmanned vehicle is a comprehensive intelligent body integrating human behaviors such as environment perception, behavior decision, autonomous control and the like, and relates to the research of a plurality of subjects such as computer technology, mode recognition, automatic control and the like. As the 'brain' of the unmanned vehicle, the behavior decision system determines the driving safety and reasonability of the unmanned vehicle, and the improvement of the intelligent level of the behavior decision system is always the key point and the difficulty of the research in the field of unmanned driving.
At present, the behavior decision model of the unmanned system roughly comprises two models, namely a rule-based model and a statistic-based model, and as research goes into, the ontology has also been successfully applied to the field of unmanned driving, for example, Armand et al describe how to apply the ontology to model interaction of spatiotemporal relations between traffic participants and infrastructure; the xylonite provides an ontology-based unmanned vehicle scene evaluation and behavior decision method. Despite the numerous studies, the existing decision-making systems still have problems of low real-time performance and safety due to the higher complexity and uncertainty of urban traffic environment compared to other traffic scenarios.
Disclosure of Invention
The present invention is directed to solving, at least to some extent, one of the technical problems in the art described above. Therefore, the invention aims to provide an unmanned vehicle behavior decision method based on an ontology and a Bayesian network, which can improve the real-time performance and accuracy of behavior decision, and thus can improve the safety of an unmanned vehicle.
In order to achieve the above object, an embodiment of the present invention provides an unmanned vehicle behavior decision method based on ontology and bayesian network, including the following steps: acquiring a driving scene of an unmanned vehicle; setting a corresponding class set according to the driving scene; constructing corresponding object attributes according to the class set and carrying out instantiation processing; constructing an ontology model corresponding to the driving scene according to the object attributes and the instantiated class set; constructing a Bayesian network model corresponding to the driving scene according to the ontology model; and generating a behavior decision of the unmanned vehicle according to the Bayesian network model.
According to the unmanned vehicle behavior decision method based on the ontology and the Bayesian network, firstly, the driving scene of the unmanned vehicle is obtained, the corresponding class set is set according to the driving scene, then the corresponding object attribute is constructed according to the class set and the instantiation processing is carried out, the ontology model corresponding to the driving scene is constructed according to the object attribute and the instantiated class set, then the Bayesian network model corresponding to the driving scene is constructed according to the ontology model, and finally the behavior decision of the unmanned vehicle is generated according to the Bayesian network model, so that the real-time performance and the accuracy of the behavior decision can be improved, and the safety of the unmanned vehicle can be improved.
In addition, the unmanned vehicle behavior decision method based on ontology and bayesian network proposed according to the above embodiment of the present invention may further have the following additional technical features:
according to one embodiment of the invention, the set of classes comprises: a self vehicle class that represents a condition of the unmanned vehicle itself; a behavior class representing a set of driving behaviors of the unmanned vehicle; an obstacle class representing a set of obstacle entities encountered by the unmanned vehicle during travel; a road class indicating a condition of a road on which the unmanned vehicle travels; a probability class representing a set of probabilities associated with a behavioral decision.
According to an embodiment of the present invention, the object attribute represents a relationship between classes in the class set, and constrains the relationship between classes in the class set by defining a domain and a value domain.
According to one embodiment of the invention, instantiating the process according to the collection includes: and instantiating the self vehicle class, the behavior class, the obstacle class and the road class to obtain corresponding instance information.
Further, the ontology model is used for structurally expressing state and semantic relations between class and instance information in the driving scene.
Further, constructing a bayesian network model corresponding to the driving scene according to the ontology model comprises: giving corresponding probability information to the ontology model; analyzing the ontology model endowed with the probability information and generating a corresponding file; carrying out format conversion on the file; and constructing the Bayesian network model corresponding to the driving scene according to the file after format conversion.
Further, the bayesian network model comprises: generating nodes, wherein the node information of the generating nodes is the information in the file after format conversion; generating an edge, wherein the generating edge is constructed according to the information in the file after format conversion; and generating a conditional probability table for indicating the probability of occurrence of a child node under the condition that a parent node occurs.
Further, the generation conditional probability table represents the probability of occurrence of a child node under the condition that a parent node occurs by the following formula:
P(Xi|Xi-1,……，X1)＝P(Xi|Pa(Xi))
wherein pa (x)i) Is node xiIs selected.
Further, the bayesian network model generates conditional probabilities of the unmanned vehicle behavior decisions by the following formula:
where E is a known behavioral decision.
Drawings
FIG. 1 is a flow chart of an ontology and Bayesian network based unmanned vehicle behavior decision method in accordance with an embodiment of the present invention;
FIG. 2 is a schematic diagram illustrating a hierarchical relationship of a set of ontologies corresponding to a driving scenario setting according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a tree structure of an ontology model corresponding to a driving scene according to an embodiment of the present invention;
FIG. 4 is a flow diagram of a method of constructing a Bayesian network model of one embodiment of the present invention;
FIG. 5 is a diagram of an initial Bayesian network model in accordance with an embodiment of the present invention;
FIG. 6 is a diagram of an updated Bayesian network model in accordance with one embodiment of the present invention;
FIG. 7 is a diagram of an updated Bayesian network model in accordance with another embodiment 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.
Fig. 1 is a flowchart of an unmanned vehicle behavior decision method based on ontology and bayesian network according to an embodiment of the present invention.
As shown in fig. 1, the unmanned vehicle behavior decision method based on ontology and bayesian network in the embodiment of the present invention includes the following steps:
and S1, acquiring the driving scene of the unmanned vehicle.
Specifically, the driving scene of the unmanned vehicle, such as city road conditions, surrounding vehicle and pedestrian conditions, traffic light information, and vehicle driving conditions, may be acquired through the sensor.
And S2, setting a corresponding class set according to the driving scene.
Specifically, as shown in fig. 2, a class set corresponding to an ontology may be set according to a driving scenario, that is, owl: thing, for example, may set the own vehicle class, EgoVehicle, Behavior class, Behavior, Obstacle class, Obstacle, Road class, Road, Probability class, such as Probasic class, State class, and Variable class.
Wherein the self vehicle class may represent the condition of the unmanned vehicle itself, such as the unmanned vehicle's own speed, direction, relative speed and relative position with other vehicles; the behavior class may represent a set of driveability behaviors of the unmanned vehicle, such as driving operations of the unmanned vehicle for straight ahead, steering, and backward; the obstacle class may represent a set of obstacle entities encountered by the unmanned Vehicle during travel, such as pedestrians, i.e., pedestrians, and vehicles, i.e., vehicles; the road class may represent the condition of a road on which the unmanned vehicle travels, such as an intersection, i.e., Junction, a Lane, i.e., Lane, a crosswalk, i.e., silk walk, a pavement marker, i.e., LaneMarker, a traffic light, i.e., TrafficLight, and a traffic sign, i.e., TrafficSign; the Probability class may represent a set of probabilities associated with the behavior decision, and the Proavailability class may further include a PriorProavailability class and a CondProavailability class, which represent a prior Probability and a conditional Probability, respectively.
And S3, constructing corresponding object attributes according to the class set and performing instantiation processing.
In an embodiment of the present invention, the object attribute may represent a relationship between classes in the class set, and may constrain the relationship between the classes in the class set through a definition Domain, i.e., Domain, and a value Domain, i.e., Range, and may input the object attribute and its corresponding definition Domain and value Domain into table 1 in a unified manner.
TABLE 1
For example, as shown in table 1, the relationship between the host vehicle class/obstacle class and the behavior class is that the driverless vehicle can perform driving behavior, so that there is a decision, i.e., a decision object attribute, between the host vehicle class and the behavior class, and the relationship between the host vehicle class/obstacle class and the behavior class can be constrained by using the host vehicle class/obstacle class as a definition domain and the behavior class as a value domain; the relationship between the self vehicle/obstacle and the road is that the driveway of the unmanned vehicle running on the road can be specified to a certain driveway, so that the driveway between the unmanned vehicle and the driveway has the property of ison lane, the relationship between the self vehicle/obstacle and the road can be restrained by taking the self vehicle/obstacle as a definition domain and taking the road as a value domain, in addition, a left driveway, namely hasLeftLane, has the property of right driveway, namely hasRightLane, and a left driveway line, namely hasLeftLine, has the property of right driveway, namely hasRightLine; the relation between the vehicle and the obstacle is that the vehicle and the obstacle have relative positions, the obstacle affects the driving action of the vehicle, and the relation between the vehicle and the obstacle can be restrained by using the obstacle as a definition domain and the vehicle as a value domain.
In one embodiment of the invention, the self vehicle class, the behavior class, the obstacle class and the road class can be instantiated to obtain corresponding instance information.
Specifically, as shown in fig. 3, instantiating the behavior class may obtain driving behavior information such as acceleration, i.e., Accelerate, deceleration, i.e., Decelerate, holding speed, i.e., Keep, parking, i.e., Stop, right lane change, i.e., ChangeToRight, and left lane change, i.e., changetolleft; instantiating the obstacle class to obtain the obstacle information of the driverless vehicle, namely a front vehicle of a lane, namely a front vehicle of a right lane, namely a rightFrontCar, and a rear vehicle of the right lane, namely a rightBackCar; instantiating the road class can obtain the condition information of the driving road, wherein the unmanned vehicle drives in a left lane, namely LeftLane, and the right lane is a right lane, namely RightLane, and the road surface marks comprise straight road surface marks, namely SMarkers, right-turn road surface marks, namely RtnArker, straight right-turn road surface marks, namely SRTurnMarkers, and traffic signs comprise speed limit marks, for example, marks of 40km/h and the like.
And S4, constructing an ontology model corresponding to the driving scene according to the object attributes and the instantiated class set.
Specifically, an ontology model corresponding to the driving scene can be constructed according to the object attributes and the instantiated class set through the assertion tool box and the term tool box.
The term toolbox can store the set of classes and object attributes set corresponding to the driving scene as a priori knowledge, the assertion toolbox can instantiate and represent the set of classes in the term toolbox and store the set of classes as scene knowledge, and then the stored priori knowledge and the stored scene knowledge are combined to form a complete ontology knowledge base so as to complete the construction of the ontology model. The state and semantic relation between the class information and the instance information in the driving scene can be expressed in a structuralized mode through the ontology model, and therefore the problems that in different driving environments, multisource heterogeneous information is not sufficiently expressed, and priori driving experience cannot be effectively utilized, and driving behavior decision real-time performance and accuracy are low are solved.
And S5, constructing a Bayesian network model corresponding to the driving scene according to the ontology model.
Specifically, step S5 includes: giving corresponding probability information to the ontology model; analyzing the ontology model endowed with the probability information and generating a corresponding file; carrying out format conversion on the file; and constructing a Bayesian network model according to the converted file.
More specifically, as shown in fig. 4, corresponding prior probability and conditional probability, that is, probability extension ontology file, can be given to the ontology model, where the representation format of the prior probability is:
prior probability p (leftlane) ═ 0.5
<！—PriorProbability for being Lane Lane1-->
<Variable rdf:ID＝“LeftLane”>
<hasClass>Lane</hasClass>
<has State>True</has State>
</Variable>
<PriorProb rdf:ID＝"P(LeftLane)">
<has Variable>LeftLane</has Variable>
<has ProbValue>0.5</has ProbValue>
</PriorProb>。
The expression format of the conditional probability is as follows:
conditional probability P (Keep | EgoCar) ═ 0.202
<！—Condition Probability for being Keep in Behavior Keep-->
<Variable rdf:ID＝"Keep">
<has Class>Behavior</has Class>
<has State>True</has State>
</Variable>
<Variable rdf:ID＝"EgoCar">
<has Class>EgoVehicle</has Class>
<has State>True</has State>
</Variable>
<CondProbability rdf:ID＝"P(Keep|EgoCar)">
<hasCondition>EgoCar</hasCondition>
<hasVariable>Keep</hasVariable>
<has Prob Value>0.202</has Prob Value>
</Cond Prob>。
Further, as shown in fig. 4, the ontology model to which the probability information is assigned may be analyzed to analyze prior probability, conditional probability, and instance information therein, and three files are correspondingly generated to respectively represent class-instance information, prior probability, and conditional probability in the ontology model, where a representation format of the class-instance information file is:
Class-Instance document:
<Lane,LeftLane>
<Obstacle,FrontCar>
<Behavior,Keep>。
the representation format of the prior probability file is as follows:
PriorProbability file:
P(LeftLane)＝0.5
the representation format of the conditional probability file is as follows:
ConditionProavailability File:
P(Keep|EgoCar)＝0.202。
further, as shown in fig. 4, the class-instance information and the conditional probability obtained after analyzing the ontology model to which the probability information is given may be subjected to format conversion, where the representation format after the class-instance information is subjected to format conversion is:
in the Class-Instance document:
<Lane：LeftLane，RightLane>
<Obstacle:FrontCar,RightFrontCar,RightBackCar>
where the colon precedes the class and the colon succeeds the colon is the instance that the class contains.
The representation format after analyzing the ontology model endowed with probability information and obtaining prior probability and carrying out format conversion is as follows:
in Link file
<Lane,EgoVehicle>
<Lane,Obstacle>
<Obstacle,EgoVehicle>
<EgoVehicle,Behavior>
Where each angle bracket has a dependency relationship to two classes within the bracket, the comma preceded by a parent node and followed by a child node.
Further, as shown in fig. 4, node information of the generation node, that is, node information of a node of the bayesian network model, may be formed according to Class-Instance information after format conversion, that is, a Class-Instance file, so that the generation node may embody both Class information and instances included in the Class and prior probability distribution corresponding to the instances.
Further, as shown in fig. 4, an edge, that is, an edge of the bayesian network model, may be generated according to the conditional probability after format conversion, that is, the Link file structure, thereby ensuring that the edges of the bayesian network model have dependency relationships and ensuring that only nodes having dependency relationships are connected to each other.
Further, as shown in fig. 4, a CPT (conditional Probability Table) may be generated according to the prior Probability and the conditional Probability after format conversion, and the conditional Probability Table may be generated to indicate the possibility of the child node looking under the Condition of the parent node, that is, the influence degree of the parent node on the child node. If the node of the Bayesian network model has no father node, the conditional probability obtained by analyzing the ontology model given with the probability information is used as the conditional probability of the Bayesian network model; and if the nodes of the Bayesian network model have father nodes, taking the conditional probability after the format conversion as the conditional probability of the Bayesian network model.
Wherein the conditional probability table is generated to represent the probability of occurrence of a child node under the condition that the parent node occurs by the following formula:
P(Xi|Xi-1,……，X1)＝P(Xi|Pa(Xi))
wherein pa (x)i) Is node xiIs selected.
In summary, a bayesian network model corresponding to the driving scenario can be constructed according to the generation nodes, the generation edges and the generation condition probability table.
It should be noted that the class-instance information, the prior probability, and the conditional probability obtained after the ontology model given with the probability information is analyzed are in the form of a numerical value pair, and are difficult to be directly used for constructing a bayesian network model, and secondly, the class-instance information obtained by the analysis only represents the one-to-one correspondence between a single instance and a single class, and is difficult to be used for constructing a bayesian network model that needs to contain all instances in each class, so that the class-instance information, the prior probability, and the conditional probability obtained after the ontology model given with the probability information is analyzed need to be subjected to format conversion.
And S6, generating a behavior decision of the unmanned vehicle according to the Bayesian network model.
Specifically, Netica can be used as a visual inference tool of the Bayesian network model to make a behavior decision for visually inferring the unmanned vehicle.
More specifically, the conditional probability of an unmanned vehicle behavior decision may be generated by the following formula:
wherein E is a known evidence variable.
The practical applicability of the unmanned vehicle behavior decision method based on the ontology and the bayesian network of the present invention will be explained below by taking as an example that the unmanned vehicle performs driving operation in the T-shaped intersection region according to the unmanned vehicle behavior decision method based on the ontology and the bayesian network of the present invention.
In an embodiment of the present invention, the initial bayesian network model shown in fig. 5 may be updated by the above method for constructing the bayesian network model to obtain the updated bayesian network model shown in fig. 6 and 7. The updating of the initial bayesian network model shown in fig. 5 by the method for constructing the bayesian network model includes updating a structure and parameters of the initial bayesian network model, that is, updating a graph model and a conditional probability table of the initial bayesian network model, and specifically includes: when the structure of the initial Bayesian network model is known and the sample data is complete, the initial Bayesian network model can be updated by adopting maximum posterior probability and maximum likelihood estimation; when the structure is known but the data sample is missing, EM (Expectation Maximization algorithm) can be adopted to update the initial Bayesian network model; when the initial Bayesian network model structure is unknown but the sample data is complete, the initial Bayesian network model can be updated by adopting the maximum posterior probability and the maximum likelihood; when the structure of the initial Bayesian network model is unknown and the sample data is incomplete, the initial Bayesian network model can be updated by adopting an EM algorithm.
Further, the posterior probability problem, the maximum possible interpretation problem and the maximum posterior hypothesis problem can be calculated after the evidence variable is given by using the updated structure of the initial bayesian network model and the conditional probability table thereof, which specifically includes: determining initial conditional probability distribution among adjacent nodes; taking a value of an evidence node; and selecting a proper reasoning algorithm, updating the conditional probability distribution of each node, and finally obtaining a reasoning result. In other words, the probabilities P (X) of certain nodes of the initial Bayesian network model that can be updated after an update is knowni) Or a known evidence variable ETime-resolved unknown node XjTaking the probability of a certain value a, namely calculating the conditional probability:
wherein E is a known evidence variable.
When the updated initial Bayesian network model is used for accurate reasoning, a message transmission algorithm, a junction tree algorithm and a variable elimination method can be selected; when the updated initial Bayesian network model is used for approximate reasoning, a variational reasoning method, a Monte Carlo method and a circular message passing algorithm can be selected.
In an embodiment of the present invention, when the unmanned vehicle is located in the T-junction area, the driving action can be determined according to obstacles, such as FrontCar, RightFrontCar, and RightBackCar, that is, the driving action of the unmanned vehicle can be determined by determining the relative speed with the vehicle in front of the own lane and the relative distance between the front vehicle and the rear vehicle in the right lane.
More specifically, when the FrontCar decelerates, the RightFrontCar decelerates, and the RightBackCar maintains the speed, it can be seen from the updated bayes network model shown in fig. 6 that the probability that the unmanned vehicle takes the right lane change is greater; when the FrontCar decelerates, the rightwardfront turns, and the rightwardback accelerates, it can be seen from the updated bayesian network model shown in fig. 7 that the probability that the unmanned vehicle will take on deceleration is greater.
Note that, in fig. 5, 6, and 7, each block diagram includes class-instance information and corresponding probability information, for example, in fig. 5, FrontCar indicates a vehicle ahead of the unmanned vehicle in the same lane, Keep indicates a holding speed, and 46.3 indicates a probability of the holding speed.
According to the unmanned vehicle behavior decision method based on the ontology and the Bayesian network, provided by the embodiment of the invention, the driving scene of the unmanned vehicle is obtained, the corresponding class set is set according to the driving scene, the corresponding object attribute is constructed according to the class set and the instantiation processing is carried out, the ontology model corresponding to the driving scene is constructed according to the object attribute and the instantiated class set, the Bayesian network model corresponding to the driving scene is constructed according to the ontology model, and the behavior decision of the unmanned vehicle is generated according to the Bayesian network model.
In the present invention, unless otherwise expressly specified or limited, the term "coupled" is to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral part; can be mechanically or electrically connected; either directly or indirectly through intervening media, either internally or in any other relationship. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (9)
1. An unmanned vehicle behavior decision method based on ontology and Bayesian network is characterized by comprising the following steps:
acquiring a driving scene of an unmanned vehicle;
setting a corresponding class set according to the driving scene;
constructing corresponding object attributes according to the class set and carrying out instantiation processing;
constructing an ontology model corresponding to the driving scene according to the object attributes and the instantiated class set;
constructing a Bayesian network model corresponding to the driving scene according to the ontology model;
and generating a behavior decision of the unmanned vehicle according to the Bayesian network model.
2. The ontology and bayesian network based unmanned vehicle behavior decision method of claim 1, wherein the set of classes comprises:
a self vehicle class that represents a condition of the unmanned vehicle itself;
a behavior class representing a set of driving behaviors of the unmanned vehicle;
an obstacle class representing a set of obstacle entities encountered by the unmanned vehicle during travel;
a road class indicating a condition of a road on which the unmanned vehicle travels;
a probability class representing a set of probabilities associated with a behavioral decision.
3. The ontology and bayesian network based unmanned vehicle behavior decision method of claim 2, wherein the object attributes represent and constrain the relationships between classes in the set of classes by defining a domain and a value domain.
4. The ontology and bayesian network based unmanned vehicle behavior decision method of claim 3, wherein instantiating a process according to the class set comprises: and instantiating the self vehicle class, the behavior class, the obstacle class and the road class to obtain corresponding instance information.
5. The ontology and bayesian network-based unmanned vehicle behavior decision method of claim 4, wherein the ontology model is used to structurally express state and semantic relationships between class and instance information in the driving scenario.
6. The ontology and bayesian network based unmanned vehicle behavior decision method of claim 5, wherein constructing a bayesian network model corresponding to the driving scenario from the ontology model comprises:
giving corresponding probability information to the ontology model;
analyzing the ontology model endowed with the probability information and generating a corresponding file;
carrying out format conversion on the file;
and constructing the Bayesian network model corresponding to the driving scene according to the file after format conversion.
7. The ontology and bayesian network based unmanned vehicle behavior decision method of claim 6, wherein the bayesian network model comprises:
generating nodes, wherein the node information of the generating nodes is the information in the file after format conversion;
generating an edge, wherein the generating edge is constructed according to the information in the file after format conversion;
and generating a conditional probability table for indicating the probability of occurrence of a child node under the condition that a parent node occurs.
8. The ontology-and-bayesian-network-based unmanned vehicle behavior decision method of claim 7, wherein the generating the conditional probability table represents a probability of occurrence of a child node on a condition that a parent node occurs by the following formula:
P(Xi|Xi-1,……，X1)＝P(Xi|Pa(Xi))
wherein pa (x)i) Is node xiIs selected.
9. The ontology-and-bayesian-network-based unmanned vehicle behavior decision method of claim 8, wherein the bayesian network model generates the conditional probability of the unmanned vehicle behavior decision by the formula:
wherein E is a known evidence variable.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911317451.7A CN110968101A (en) | 2019-12-19 | 2019-12-19 | Unmanned vehicle behavior decision method based on ontology and Bayesian network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911317451.7A CN110968101A (en) | 2019-12-19 | 2019-12-19 | Unmanned vehicle behavior decision method based on ontology and Bayesian network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110968101A true CN110968101A (en) | 2020-04-07 |
Family
ID=70035228
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911317451.7A Pending CN110968101A (en) | 2019-12-19 | 2019-12-19 | Unmanned vehicle behavior decision method based on ontology and Bayesian network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110968101A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114051220A (en) * | 2021-10-27 | 2022-02-15 | 上海智能网联汽车技术中心有限公司 | Ontology-based dynamic Bayesian attack graph generation method and system for Internet of vehicles |
| CN115034281A (en) * | 2022-03-23 | 2022-09-09 | 华东师范大学 | Robosim-based construction method of automatic driving behavior decision model |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20080029417A (en) * | 2006-09-29 | 2008-04-03 | 김홍기 | Automatic semantic information extraction from web documents for semantic web annotation |
| CN104850399A (en) * | 2015-04-30 | 2015-08-19 | 昆明理工大学 | Retrospective analysis method and retrospective analysis system capable of generating mapping table between component interface and component implementation |
| CN105447504A (en) * | 2015-11-06 | 2016-03-30 | 中国科学院计算技术研究所 | Traffic mode behavior recognition method and corresponding recognition model construction method |
| CN105631414A (en) * | 2015-12-23 | 2016-06-01 | 上海理工大学 | Vehicle-borne multi-obstacle classification device and method based on Bayes classifier |
| CN106950956A (en) * | 2017-03-22 | 2017-07-14 | 合肥工业大学 | The wheelpath forecasting system of fusional movement model and behavior cognitive model |
| CN109919181A (en) * | 2019-01-24 | 2019-06-21 | 南京航空航天大学 | Dynamic fault tree quantitative analysis method based on probabilistic model checking |
| US20190325325A1 (en) * | 2018-04-19 | 2019-10-24 | International Business Machines Corporation | Vehicular driving actions in the presence of non-recurrent events |
-
2019
- 2019-12-19 CN CN201911317451.7A patent/CN110968101A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20080029417A (en) * | 2006-09-29 | 2008-04-03 | 김홍기 | Automatic semantic information extraction from web documents for semantic web annotation |
| CN104850399A (en) * | 2015-04-30 | 2015-08-19 | 昆明理工大学 | Retrospective analysis method and retrospective analysis system capable of generating mapping table between component interface and component implementation |
| CN105447504A (en) * | 2015-11-06 | 2016-03-30 | 中国科学院计算技术研究所 | Traffic mode behavior recognition method and corresponding recognition model construction method |
| CN105631414A (en) * | 2015-12-23 | 2016-06-01 | 上海理工大学 | Vehicle-borne multi-obstacle classification device and method based on Bayes classifier |
| CN106950956A (en) * | 2017-03-22 | 2017-07-14 | 合肥工业大学 | The wheelpath forecasting system of fusional movement model and behavior cognitive model |
| US20190325325A1 (en) * | 2018-04-19 | 2019-10-24 | International Business Machines Corporation | Vehicular driving actions in the presence of non-recurrent events |
| CN109919181A (en) * | 2019-01-24 | 2019-06-21 | 南京航空航天大学 | Dynamic fault tree quantitative analysis method based on probabilistic model checking |
Non-Patent Citations (4)
| Title |
|---|
| 何艳侠，等: ""基于环境态势评估的智能车自主变道决策机制"", 《汽车工程》 * |
| 曹雪亚: ""基于贝叶斯网络的本体不确定性推理研究"", 《中国优秀硕士学位论文全文数据库》 * |
| 高明: "《知识协同工作流建模、服务规划与服务组合》", 31 March 2015 * |
| 黄璐: ""基于本体论的无人驾驶车辆场景评估与行为决策方法研究"", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114051220A (en) * | 2021-10-27 | 2022-02-15 | 上海智能网联汽车技术中心有限公司 | Ontology-based dynamic Bayesian attack graph generation method and system for Internet of vehicles |
| CN114051220B (en) * | 2021-10-27 | 2024-04-16 | 上海智能网联汽车技术中心有限公司 | Ontology-based vehicle networking dynamic Bayesian attack graph generation method and system |
| CN115034281A (en) * | 2022-03-23 | 2022-09-09 | 华东师范大学 | Robosim-based construction method of automatic driving behavior decision model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP4009300A1 (en) | Vehicle automatic control method and lane change intention prediction network training method | |
| CN110304075B (en) | Vehicle track prediction method based on hybrid dynamic Bayesian network and Gaussian process | |
| Forbes et al. | The batmobile: Towards a bayesian automated taxi | |
| CN107229973B (en) | Method and device for generating strategy network model for automatic vehicle driving | |
| WO2020034903A1 (en) | Smart navigation method and system based on topological map | |
| Huang et al. | A probabilistic risk assessment framework considering lane-changing behavior interaction | |
| el abidine Kherroubi et al. | Novel decision-making strategy for connected and autonomous vehicles in highway on-ramp merging | |
| Zhao et al. | An ontology-based intelligent speed adaptation system for autonomous cars | |
| Scheel et al. | Situation assessment for planning lane changes: Combining recurrent models and prediction | |
| Lefèvre et al. | Context-based estimation of driver intent at road intersections | |
| CN110968101A (en) | Unmanned vehicle behavior decision method based on ontology and Bayesian network | |
| Wheeler et al. | Factor graph scene distributions for automotive safety analysis | |
| Kolekar et al. | Behavior prediction of traffic actors for intelligent vehicle using artificial intelligence techniques: A review | |
| Sequeira et al. | A lane merge coordination model for a V2X scenario | |
| Wheeler et al. | A probabilistic framework for microscopic traffic propagation | |
| Chen et al. | Advanced driver assistance strategies for a single-vehicle overtaking a platoon on the two-lane two-way road | |
| Chowdhury et al. | Applications of artificial intelligence paradigms to decision support in real-time traffic management | |
| Sarkar et al. | Solution concepts in hierarchical games under bounded rationality with applications to autonomous driving | |
| US20230391357A1 (en) | Methods and apparatus for natural language based scenario discovery to train a machine learning model for a driving system | |
| CN115981302A (en) | Vehicle following lane change behavior decision-making method and device and electronic equipment | |
| Wang et al. | Cooperative decision making for connected automated vehicles in multiple driving scenarios | |
| CN115034281A (en) | Robosim-based construction method of automatic driving behavior decision model | |
| Withanawasam et al. | Road scene risk perception for intelligent vehicles using end-to-end affordance learning and visual reasoning | |
| CN108898284B (en) | Internet of vehicles management control strategy evaluation method and system | |
| Raiyn et al. | Improving Autonomous Vehicle Reasoning with Non-Monotonic Logic: Advancing Safety and Performance in Complex 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200407 |