CN109948830B

CN109948830B - Bicycle track prediction method, equipment and medium oriented to self-mixed environment

Info

Publication number: CN109948830B
Application number: CN201910084231.8A
Authority: CN
Inventors: 王晓原; 刘亚奇; 夏媛媛; 郭永青; 韩俊彦; 刘士杰; 刘善良
Original assignee: Qingdao University of Science and Technology
Current assignee: Qingdao University of Science and Technology
Priority date: 2019-01-29
Filing date: 2019-01-29
Publication date: 2021-03-02
Anticipated expiration: 2039-01-29
Also published as: CN109948830A

Abstract

The invention relates to a bicycle track prediction method, equipment and medium for a human-self hybrid environment. The method includes acquiring first data of a target bicycle; determining second data between the target bicycle and other traffic entities in the interest sensing area; determining a traffic phase state between the target bicycle and other traffic entities in the interest sensing area; and predicting the running track of the target bicycle according to the first data, the second data, the traffic phase state and a bicycle track prediction model established in advance. The method is based on the data of the target bicycle; the method comprises the steps of predicting the running track of a target bicycle according to data between the target bicycle and other traffic entities in an interest sensing area, traffic phases between the target bicycle and other traffic entities in the interest sensing area and a bicycle track prediction model established in advance, and achieving track prediction of the target bicycle in a complex self-mixing environment.

Description

Bicycle track prediction method, equipment and medium oriented to self-mixed environment

Technical Field

The invention relates to the technical field of intelligent transportation, in particular to a bicycle track prediction method, equipment and medium for a self-mixed environment.

Background

With the rapid increase of automobile holding capacity, roads are rapidly increased, more and more motor vehicle lanes are continuously widened, the running space of pedestrians and non-motor vehicles is seriously occupied, and the conflict between bicycles and pedestrians is increasingly serious.

The motorized and flexible bicycle is an important factor influencing the safety of the man-self communication. The development of technologies such as wide application, big data analysis and high-speed cloud computing of GPS mobile sensing equipment (such as smart phones) provides powerful guarantee for researching timely safety early warning and improving road safety under networking conditions.

The real-time prediction of the running track of a target bicycle in a human-self hybrid environment is a core scientific problem which needs to be solved urgently.

Disclosure of Invention

Technical problem to be solved

In order to predict bicycle tracks in a self-mixing environment, the invention provides a bicycle track prediction method, equipment and a medium facing the self-mixing environment.

(II) technical scheme

In order to achieve the purpose, the invention adopts the main technical scheme that:

a human-self-confounding environment oriented bicycle trajectory prediction method, the method comprising:

s101, collecting first data of a target bicycle;

s102, determining second data between the target bicycle and other traffic entities in the interest sensing area;

s103, determining a traffic phase state between the target bicycle and other traffic entities in the interest induction area;

and S104, predicting the running track of the target bicycle according to the first data, the second data, the traffic phase and a bicycle track prediction model established in advance.

Optionally, the first data comprises: target bicycle rider type, target bicycle rider grip, target bicycle rider braking force, target bicycle rider pedaling frequency, target bicycle front wheel turning angle, target bicycle lateral displacement, longitudinal displacement and running speed;

the second data includes: the type of other traffic entities in the interest sensing area, the relative transverse distance between the target bicycle and the other traffic entities in the interest sensing area, the relative longitudinal distance between the target bicycle and the other traffic entities in the interest sensing area, and the relative speed between the target bicycle and the other traffic entities in the interest sensing area;

the target bicycle rider type is: conservative, or robust, or aggressive;

the turning angle of the front wheel of the target bicycle is large, or the turning angle is medium, or the turning angle is small;

the type of other traffic entities in the interest induction area is bicycles or pedestrians;

the braking force of the target bicycle rider is high, or moderate, or low;

the pedaling frequency of the target bicycle rider is high, or the pedaling frequency is medium, or the pedaling frequency is low;

the relative transverse distance between the target bicycle and other traffic entities in the interest sensing area is far relative transverse distance, or in the relative transverse distance, or close to the relative transverse distance;

the relative longitudinal distance between the target bicycle and other traffic entities in the interest sensing area is far relative longitudinal distance, or the relative longitudinal distance is medium relative longitudinal distance, or the relative longitudinal distance is close.

Optionally, the method for determining the sensing region of interest is as follows:

dividing the road section into a left area, a middle area and a right area with the widths of 1 m according to the maximum parking density and the maximum outline size of the target bicycle rider during riding;

based on the left, middle and right regions, the front right sub-region, the rear right sub-region, the secondary right sub-region, the front sub-region, the rear sub-region, the front sub-region, the rear sub-region, the front left sub-region, the rear left sub-region and the secondary left sub-region with the position of the front wheel axle of the target bicycle as a base point.

Optionally, the S103 includes:

s103-1, based on the types of other traffic entities in the interest induction area,The relative speed of the target bicycle and other traffic entities in the interest sensing area, the relative transverse distance of the target bicycle and other traffic entities in the interest sensing area, and the relative longitudinal distance of the target bicycle and other traffic entities in the interest sensing area are calculated by using a fuzzy inference rule, and the action granularity of other traffic entity types in the interest sensing area on the target bicycle is calculated by using a fuzzy inference rule

S103-2, according to

Determining the traffic phase.

Wherein the content of the first and second substances,

i is the mark of the road section virtual area where the target bicycle is located, i is the left area, or the middle area, or the right area, n is the mark of other traffic entities in the interest induction area,

the granularity of the action of the left front area on the target bicycle located in the i area,

the granularity of the action of the left rear area on the target bicycle located in the i area,

the grain size of the left-time rear-side region acting on the target bicycle located in the i region,

the granularity of the effect of the front area on the target bicycle located in the i area,

for the rear zone pair in zone iThe function granularity of the target bicycle is set,

the granularity of the action of the right front area on the target bicycle located in the i area,

is the granularity of the action of the right rear side region on the target bicycle located in the i region,

the granularity of the effect of the right next rear area on the target bicycle located in the i area.

Optionally, the S103-2 includes:

s103-2-1, according to

Determining the field intensity type of the traffic phase state;

s103-2-2, determining a traffic phase state according to the field intensity type;

the field intensity types are divided into strong repulsion field intensity, middle repulsion field intensity, weak repulsion field intensity, zero field intensity, weak attraction field intensity, middle attraction field intensity and strong attraction field intensity;

the action granularity of the j area on the target bicycles positioned in the i area is in an interval of [ -1, -0.8 ], the field intensity type is strong repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in an interval of [ -0.8, -0.4 ], the field intensity type is medium repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in an interval of [ -0.4,0 ], the field intensity type is weak repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is 0, the field intensity type is zero field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in (0,0.3], the field intensity type is weak attraction field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in (0.3,0.7, the field intensity type is a medium attraction field intensity, the action granularity of the j area on the target bicycle positioned in the i area is in (0.7, 1), the field intensity type is a strong attraction field intensity, j is the interest induction area identification, and the j area is a right front side sub-area, a right rear side sub-area, a front side sub-area, a rear side sub-area, a left front side sub-area, a left rear side sub-area, or a left rear side sub-area.

Optionally, the S103-2-2 includes:

determining to form a repulsive field if the field intensity types are strong repulsive field intensity, medium repulsive field intensity and weak repulsive field intensity;

if the field intensity types are weak attraction field intensity, medium attraction field intensity and strong attraction field intensity, determining to form an attraction field;

the traffic phase state is as follows:

the i area is a middle area, and the front sub-area, the left front sub-area and the right front sub-area form an attraction field; alternatively, the first and second electrodes may be,

the i area is a middle area, the front sub-area and the left front sub-area form an attraction field, and the right front sub-area forms a repulsion field; alternatively, the first and second electrodes may be,

the i area is a middle area, the front side sub-area and the right front side sub-area form an attraction field, and the left front side sub-area forms a repulsion field; alternatively, the first and second electrodes may be,

the i area is a middle area, the front side sub-area forms an attraction field, and the left front side sub-area and the right front side sub-area form a repulsion field; alternatively, the first and second electrodes may be,

the i area is a middle area, the left front sub-area and the right front sub-area form an attraction field, and the front sub-area forms a repulsion field; alternatively, the first and second electrodes may be,

the i area is a middle area, the left front sub-area forms an attraction field, and the front sub-area and the right front sub-area form a repulsion field; alternatively, the first and second electrodes may be,

the i area is a middle area, the right front sub-area forms an attraction field, and the front sub-area and the left front sub-area form a repulsion field; alternatively, the first and second electrodes may be,

the i area is a middle area, and the front side sub-area, the left front side sub-area and the right front side sub-area form a repulsive field; alternatively, the first and second electrodes may be,

the i area is a left area, and the front sub-area and the right front sub-area form an attraction field; alternatively, the first and second electrodes may be,

the i area is a left area, the front sub-area forms an attraction field, and the right front sub-area forms a repulsion field; alternatively, the first and second electrodes may be,

the i area is a left area, the front sub-area forms a repulsive field, and the right front sub-area forms an attractive field; alternatively, the first and second electrodes may be,

the i area is a left area, and the front sub-area and the right front sub-area both form a repulsive field; alternatively, the first and second electrodes may be,

the i area is a right area, and the front sub-area and the left front sub-area both form an attraction field; alternatively, the first and second electrodes may be,

the i area is a right area, the front sub-area forms an attraction field, and the left front sub-area forms a repulsion field; alternatively, the first and second electrodes may be,

the i area is a right area, the front sub-area forms a repulsive field, and the left front sub-area forms an attractive field; alternatively, the first and second electrodes may be,

the i-region is the right-side region, and the front-side sub-region and the left-front-side sub-region both form a repulsive field.

Optionally, the bicycle track prediction model is established as follows:

s100-1, collecting sample data once every preset time to form a sample data set;

s100-2, according to the sample data set, adopting a dynamic Bayesian network to carry out inference to calculate bicycle transverse displacement probability, longitudinal displacement probability and speed change probability corresponding to the sample data;

s100-3, obtaining bicycle track prediction corresponding to sample data according to the bicycle transverse displacement probability, the longitudinal displacement probability and the speed change probability corresponding to the sample data, and establishing a bicycle track prediction model;

any sample data comprises the third data, the fourth data and a sample traffic phase;

the third data includes: the bicycle rider type corresponding to the sample data, the bicycle rider gripping power corresponding to the sample data, the bicycle rider braking force corresponding to the sample data, the bicycle rider pedaling frequency corresponding to the sample data, the bicycle front wheel rotation angle corresponding to the sample data, and the bicycle transverse displacement, the bicycle longitudinal displacement and the bicycle running speed corresponding to the sample data;

the fourth data includes: the type of other traffic entities corresponding to the sample data, the relative transverse distance between the bicycle corresponding to the sample data and the other traffic entities corresponding to the sample data, the relative longitudinal distance between the bicycle corresponding to the sample data and the other traffic entities corresponding to the sample data, and the relative speed between the bicycle corresponding to the sample data and the other traffic entities corresponding to the sample data;

the bicycle rider type corresponding to the sample data is as follows: conservative, or robust, or aggressive;

the corner of the front wheel of the bicycle corresponding to the sample data is large, or in the corner, or small;

the other traffic entity type corresponding to the sample data is a bicycle or a pedestrian;

the bicycle rider brake force corresponding to the sample data is strong, or moderate, or small;

the pedaling frequency of the bicycle rider corresponding to the sample data is high, or medium, or low;

the bicycle corresponding to the sample data and other traffic entities corresponding to the sample data are far from each other in relative transverse distance, or close to each other in relative transverse distance;

the bicycle corresponding to the sample data and other traffic entities corresponding to the sample data are far from each other in relative longitudinal distance, or close to each other in relative longitudinal distance.

Optionally, the dynamic bayesian network inference process is:

wherein T is the total times of collecting sample data, T is more than or equal to 1 and less than or equal to TM is the number of observation nodes in the dynamic Bayesian network, M is more than or equal to 1 and less than or equal to M, K is the number of hidden nodes in the dynamic Bayesian network, K is more than or equal to 1 and less than or equal to K, x_tkIs X_tkIs a value state of X_tkThe value y of the hidden node k in the sample data acquired at the t time_tmFor observing variable Y_tmValue of (A), Y_tmIs an observed variable, y, of the observed node m at the t-th acquisition_tm0Is Y_tm0Value of (A), Y_tm0Is the observed value of the observed node m at the t-th acquisition, pi (Y)_tm) Is Y_tmFather node, pi (X)_tk) Is X_tkParent node, P (Y)_tm0＝y_tm) Is Y_tmBelongs to the continuous observed value of y_tmDegree of membership, P (x)_tk|π(X_tk) Is x)_tkAt the father node pi (X)_tk) Conditional probability of P (y)_tm|π(Y_tm) Is y)_tmAt the father node pi (Y)_tm) A conditional probability of;

degree of membership

y_tm,minFor y in all sample data_tmMinimum value of, y_tm,maxFor y in all sample data_tmThe maximum value of (a) is,

for y in all sample data_tmIs measured.

In order to achieve the above purpose, the main technical solution adopted by the present invention further comprises:

an electronic device comprising a memory, a processor, a bus and a computer program stored on the memory and executable on the processor, the processor implementing the steps of any of the above methods when executing the program.

a computer storage medium having stored thereon a computer program which, when executed by a processor, carries out the steps of any of the methods as described above.

(III) advantageous effects

The invention has the beneficial effects that: according to the data of the target bicycle; the method comprises the steps of predicting the running track of a target bicycle according to data between the target bicycle and other traffic entities in an interest sensing area, traffic phases between the target bicycle and other traffic entities in the interest sensing area and a bicycle track prediction model established in advance, and achieving track prediction of the target bicycle in a complex self-mixing environment.

Drawings

Fig. 1 is a schematic structural diagram of a dynamic bayesian network according to an embodiment of the present invention;

FIG. 2 is a schematic flow chart of a method for predicting a bicycle track facing a self-mixing environment according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a sensing region of interest according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of a traffic phase type according to an embodiment of the present invention;

FIG. 5 is a schematic diagram illustrating a comparison of predicted values and actual values of a robust rider's trajectory in accordance with an embodiment of the present invention;

FIG. 6 is a schematic diagram illustrating a comparison between a predicted value and an actual value of a conservative rider trajectory according to an embodiment of the present invention;

FIG. 7 is a schematic diagram illustrating a comparison of predicted and actual values of an aggressive rider trajectory in accordance with an embodiment of the present invention;

fig. 8 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

In order to predict bicycle tracks in a self-mixing environment, the proposal provides a bicycle track prediction method facing the self-mixing environment, according to the data of target bicycles; the method comprises the steps of predicting the running track of a target bicycle according to data between the target bicycle and other traffic entities in an interest sensing area, traffic phases between the target bicycle and other traffic entities in the interest sensing area and a bicycle track prediction model established in advance, and achieving track prediction of the target bicycle in a complex self-mixing environment.

The implementation flow of the human-self hybrid environment-oriented bicycle track prediction method provided by the embodiment comprises 2 major parts: in a first section, a bicycle track prediction model is established. And a second part for predicting the target bicycle track based on the built bicycle track prediction model. The following describes the implementation flow of part 2.

In a first section, a bicycle track prediction model is established.

S100-1, collecting sample data once every preset time to form a sample data set.

For example, sample data is acquired every Δ t, and a sample data set is formed.

In this embodiment, the value of Δ t is not limited, and Δ t is, for example, 1 second.

Additionally, the acquisition methods include, but are not limited to: the method comprises the steps of collecting corresponding numerical values of all variables at the same time at an observation place of an urban road, wherein the value range of each parameter variable must be composed of a series of discrete values. If the current time is t, the time before t is represented by t-1, t-2, etc., and the time after t is represented by t +1, t +2, etc.

Any sample data comprises third data, fourth data and a sample traffic phase.

The third data includes: the bicycle rider type corresponding to the sample data, the bicycle rider gripping power corresponding to the sample data, the bicycle rider braking force corresponding to the sample data, the bicycle rider pedaling frequency corresponding to the sample data, the bicycle front wheel rotating angle corresponding to the sample data, and the bicycle transverse displacement, the bicycle longitudinal displacement and the bicycle running speed corresponding to the sample data.

The fourth data includes: the bicycle corresponding to the sample data is in a corresponding mode, and the bicycle corresponding to the sample data is in a corresponding mode.

The bicycle rider type corresponding to the sample data is as follows: conservative, or robust, or aggressive.

The turning angle of the front wheel of the bicycle corresponding to the sample data is large, or in the turning angle, or small.

The other traffic entity type corresponding to the sample data is a bicycle or a pedestrian.

The bicycle rider's brake dynamics that sample data correspond is the dynamics big, or in the dynamics, or the dynamics is little.

The pedaling frequency of the bicycle rider corresponding to the sample data is high, or medium, or low.

The bicycle corresponding to the sample data and other traffic entities corresponding to the sample data are far from the corresponding transverse distance, or in the corresponding transverse distance, or close to the corresponding transverse distance.

The bicycle corresponding to the sample data and other traffic entities corresponding to the sample data are far from the longitudinal distance, or in the longitudinal distance, or close to the longitudinal distance.

Where each term is taken as a variable.

And S100-2, adopting a dynamic Bayesian network to carry out inference calculation on the bicycle transverse displacement probability, the longitudinal displacement probability and the speed change probability corresponding to the sample data according to the sample data set.

In the step, a bicycle transverse displacement probability, a longitudinal displacement probability and a speed change probability corresponding to the sample data are calculated by adopting dynamic Bayesian network reasoning.

A Dynamic Bayesian Network (DBN) can be defined as (B)₀,B_→) In which B is₀Representing a priori distributions of static Bayesian Networks (BN), i.e. an initial Network of DBNs, B_→Graph showing BN composition for more than 2 time slices.

For example, the reasoning process of the dynamic bayesian network of the bicycle running track state change corresponding to the sample data shown in fig. 1 shows the causal relationship among the levels and the variables. The set of basic variable states in the dynamic bayesian network model is: the bicycle front wheel rotation angle FA corresponding to the sample data, the bicycle rider gripping strength CL corresponding to the sample data, the bicycle rider pedaling frequency CP corresponding to the sample data, the traffic entity type ET (pedestrian and bicycle) corresponding to the sample data, the longitudinal distance RD (far, medium and near) between the traffic entity corresponding to the sample data and the bicycle corresponding to the sample data, the transverse distance RS (far, medium and near) between the traffic entity corresponding to the sample data and the bicycle corresponding to the sample data, and the relative speed RV (fast, medium and slow) between the traffic entity corresponding to the sample data and the bicycle corresponding to the sample data.

The time horizon of a dynamic bayesian network can be any segment of [0, + ∞ ], but in practical applications one typically only needs to look at a limited time segment [1,2, …, T ].

The static Bayesian network with K hidden nodes and M observation nodes comprises a dynamic Bayesian network reasoning process consisting of T time slices as follows:

wherein T is the total times of acquiring sample data, T is more than or equal to 1 and less than or equal to T, M is the number of observation nodes in the dynamic Bayesian network, M is more than or equal to 1 and less than or equal to M, K is the number of hidden nodes in the dynamic Bayesian network, K is more than or equal to 1 and less than or equal to K, and x_tkIs X_tkIs a value state of X_tkThe value y of the hidden node k in the sample data acquired at the t time_tmFor observing variable Y_tmValue of (A), Y_tmIs an observed variable, y, of the observed node m at the t-th acquisition_tm0Is Y_tm0Value of (A), Y_tm0Is the observed value of the observed node m at the t-th acquisition, pi (Y)_tm) Is Y_tmFather node, pi (X)_tk) Is X_tkParent node, P (Y)_tm0＝y_tm) Is Y_tmBelongs to the continuous observed value of y_tmDegree of membership, P (x)_tk|π(X_tk) Is x)_tkAt the father node pi (X)_tk) Conditional probability of P (y)_tm|π(Y_tm) Is y)_tmAt the father node pi (Y)_tm) Conditional probability of (c).

P(Y_tm0＝y_tm) The probability of a characteristic variable is expressed by the membership degree, expressed by the numerical value of a membership function and P_uIndicating Y in the sample data_tmMembership to feature y_tmThe possibility of (a). The states of the feature vectors selected in this embodiment are all three types, and the membership degree (probability) calculation formula is:

degree of membership

y_tm,minFor y in all sample data_tmMinimum value of, y_tm,maxFor y in all sample data_tmMaximum value of, y_tmFor y in all sample data_tmIs measured.

And S100-3, obtaining bicycle track prediction corresponding to the sample data according to the bicycle transverse displacement probability, the longitudinal displacement probability and the speed change probability corresponding to the sample data, and establishing a bicycle track prediction model.

The method comprises the steps of predicting the running track of a target bicycle in the obtained human self-mixing environment based on a dynamic Bayesian network, and establishing a bicycle track prediction model, wherein the bicycle track prediction model mainly comprises a training stage and a prediction stage. The training stage can be carried out in an off-line state, and mainly learns and excavates historical random data, determines variables representing the bicycle running track characteristics, and constructs a track prediction model; and in the prediction stage, the target bicycle track is analyzed and predicted on line based on the trained model.

When a bicycle track prediction model is established, the embodiment takes variables which can represent the running track of the bicycle, such as: and the sample data corresponds to the transverse displacement, the longitudinal displacement and the speed of the bicycle in the running process. And comprehensively analyzing information such as human-vehicle-environment and the like on the basis to screen out basic characteristic quantities, and further calculating the bicycle transverse displacement probability, the longitudinal displacement probability and the speed change probability corresponding to the sample data according to dynamic Bayesian network reasoning, so as to determine the bicycle running track corresponding to the sample data, and further establish a bicycle track prediction model.

After the bicycle track prediction model is established through the first part, the second part is entered to predict the target bicycle track based on the established bicycle track prediction model, and the implementation process is shown in fig. 2.

And a second part for predicting the target bicycle track based on the built bicycle track prediction model.

S101, collecting first data of the target bicycle.

Wherein the first data includes: target bicycle rider type, target bicycle rider grip, target bicycle rider braking effort, target bicycle rider pedaling frequency, target bicycle front wheel turn, target bicycle lateral displacement, longitudinal displacement, and running speed.

The second data includes: the type of other traffic entities in the interest sensing area, the relative lateral distance between the target bicycle and the other traffic entities in the interest sensing area, the relative longitudinal distance between the target bicycle and the other traffic entities in the interest sensing area, and the relative speed between the target bicycle and the other traffic entities in the interest sensing area.

The target bicycle rider type is: conservative, or robust, or aggressive.

The turning angle of the front wheel of the target bicycle is large, or the turning angle is middle, or the turning angle is small.

The other traffic entity types in the interest sensing area are bicycles or pedestrians.

The braking force of the target bicycle rider is high, or medium, or low.

The pedaling frequency of the target bicycle rider is high, or the pedaling frequency is medium, or the pedaling frequency is low.

The target bicycle is relatively far away from other traffic entities in the interest sensing area in the transverse direction, or is relatively middle in the transverse direction, or is relatively close to the transverse direction.

The target bicycle is relatively far from other traffic entities in the interest sensing area in the longitudinal distance, or is relatively medium in the longitudinal distance, or is relatively close to the longitudinal distance.

And S102, determining second data between the target bicycle and other traffic entities in the interest sensing area.

The interest sensing area is an area which is sensed by a perception system of a rider and has influence on riding, and the determination method comprises the following steps:

and dividing the road virtual region into a left region, a middle region and a right region with the widths of 1 m according to the maximum parking density and the maximum outline size of the target bicycle rider during riding.

For example, as shown in fig. 3, the human-self mixed road section is virtually divided into three areas of left, middle and right sides with widths of 1 meter according to the maximum parking density and the maximum overall dimension when the rider rides the bicycle, and the interest sensing area of the rider is divided into 8 sub-areas of the right front side, the right rear side, the front side, the rear side, the left front side, the left rear side and the left rear side by taking the position of the front wheel axle of the target bicycle as a base point.

And S103, determining the traffic phase state between the target bicycle and other traffic entities in the interest sensing area.

The implementation process of the step comprises 1) traffic phase mathematical expression and 2) traffic phase simplification, and the implementation process comprises the following steps:

1) mathematical expression of traffic phase

S103-1, calculating action granularity of other traffic entity types in the interest induction area on the target bicycle by using a fuzzy inference rule based on other traffic entity types in the interest induction area, relative speeds of the target bicycle and other traffic entities in the interest induction area, relative transverse distances of the target bicycle and other traffic entities in the interest induction area, and relative longitudinal distances of the target bicycle and other traffic entities in the interest induction area

Wherein the content of the first and second substances,

the action granularity of the left-time rear-side region on the target bicycle located in the i region,

the granularity of the effect of the rear area on the target bicycle located in the i area,

the granularity of the effect of the right front region on the target bicycle located in the i region,

the granularity of the action of the right rear side region on the target bicycle located in the i region,

the granularity of the effect of the right-next-rear-side region on the target bicycle located in the i region.

S103-2, according to

And determining the traffic phase state.

The method comprises the following steps:

s103-2-1, according to

The field strength type of the traffic phase is determined.

The field intensity types are divided into strong repulsion field intensity, middle repulsion field intensity, weak repulsion field intensity, zero field intensity, weak attraction field intensity, middle attraction field intensity and strong attraction field intensity.

The action granularity of the j area on the target bicycles positioned in the i area is in an interval < -1 > -0.8), the field intensity type is strong repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in an interval < -0.8 > -0.4), the field intensity type is medium repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in an interval < -0.4,0, the field intensity type is weak repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is 0, the field intensity type is zero field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in (0, 0.3), the field intensity type is weak attraction field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in (0.3, 0.7), the field intensity type is medium attraction field intensity, the action granularity of the j area on the target bicycle in the i area is (0.7, 1), the field intensity type is strong absorption field intensity, j is the interest induction area identification, and the j area is a right front sub-area, or a right rear sub-area, or a front sub-area, or a rear sub-area, or a left front sub-area, or a left rear sub-area.

Executing the steps, and finishing the process of 1) traffic phase mathematical expression. In the traffic phase mathematical expression process, a fuzzy logic is applied, a language variable is adopted to carry out approximate reasoning, and a subjective judgment process established on the basis of knowledge and experience of a rider is depicted.

And obtaining the field intensity of a field formed by mutual attraction or repulsion between the traffic entity and the target bicycle and the rider thereof in the interest induction area of the target bicycle rider by using a fuzzy calculation method. The field intensity is described by the granularity, the action granularity of the field intensity is divided into six types by intervals,

the range of action size for strong repelling field strength [ -1, -0.8),

the range of the action particle size for moderate repulsive field strengths [ -0.8, -0.4),

the range of action size for weak repulsive field strength [ -0.4,0 ],

● the action strength is 0 at zero field strength,

● the range (0, 0.3) for the particle size to be acted upon when the field strength is weak,

the region for particle size (0.3, 0.7) at the medium-field-attracting-field strength,

the range for the action particle size at a strong absorption field strength (0.7, 1%).

Taking the calculation of the action granularity of the traffic entity in the interest induction area on the left front side of the target bicycle positioned in the middle of the human-self hybrid road section on the target bicycle and the rider thereof as an example, when the traffic entity in the left front side area is the bicycle or the pedestrian, the fuzzy inference rules of the action granularity on the target bicycle and the rider thereof are different due to the significant difference between the speeds of the bicycle and the pedestrian, and the following two typical language fuzzy rules are listed:

1) if the traffic entity in the front left is a pedestrian and the relative speed between the pedestrian and the target bicycle is negative and the relative distance between the pedestrian and the target bicycle is short, the action granularity of the pedestrian in the front left on the target bicycle and the rider is-1, and the pedestrian belongs to strong repelling field intensity.

2) If the traffic entity in the front left is a bicycle and the relative speed between the traffic entity and the target bicycle is positive and the relative distance between the traffic entity and the target bicycle is long, the action granularity of the bicycle in the front left on the target bicycle and a rider is 1, and the bicycle belongs to strong attraction field intensity.

As mentioned above, the traffic entity type, relative speed and relative distance represent three different input variables, and the granularity of action of the magnitude of the field strength is the output variable. Calculating a fuzzy set and membership degree of relative speed and relative distance between the target vehicle and the rider thereof and a traffic entity in the interest induction area to obtain: for target bicycles located in the middle area of the man-self mixed road section, n in fig. 2₀Can use

Representing a target vehicle and its rider n₀The human-self-interacting phase is in. When the target bicycle is positioned in the left or right area of the man-self-mixing road section, the target bicycle and the rider thereof are in the self-human-communication phase state derivation process, and the target bicycle and the rider thereof are in the middle area of the man-self-mixing road section₀Similarly.

After the traffic phase mathematical expression is obtained, 2) a brief 103-2-2 step of traffic phase is performed.

S103-2-2, determining the traffic phase state according to the field intensity type.

And determining to form a repulsive field if the field intensity types are strong repulsive field intensity, medium repulsive field intensity and weak repulsive field intensity.

And if the field intensity types are weak attraction field intensity, medium attraction field intensity and strong attraction field intensity, determining to form an attraction field.

The traffic phase is:

the i-region is a middle region, and the front sub-region, the left front sub-region and the right front sub-region form an attraction field. Alternatively, the first and second electrodes may be,

the i area is a middle area, the front sub-area and the left front sub-area form an attraction field, and the right front sub-area forms a repulsion field. Alternatively, the first and second electrodes may be,

the i area is a middle area, the front sub-area and the right front sub-area form an attraction field, and the left front sub-area forms a repulsion field. Alternatively, the first and second electrodes may be,

the i area is a middle area, the front sub-area forms an attraction field, and the left front sub-area and the right front sub-area form a repulsion field. Alternatively, the first and second electrodes may be,

the i area is a middle area, the left front sub-area and the right front sub-area form an attraction field, and the front sub-area forms a repulsion field. Alternatively, the first and second electrodes may be,

the i area is a middle area, the left front sub-area forms an attraction field, and the front sub-area and the right front sub-area form a repulsion field. Alternatively, the first and second electrodes may be,

the i area is a middle area, the right front sub-area forms an attraction field, and the front sub-area and the left front sub-area form a repulsion field. Alternatively, the first and second electrodes may be,

the i area is a middle area, and the front sub-area, the left front sub-area and the right front sub-area form a repulsive field. Alternatively, the first and second electrodes may be,

the i-region is the left-side region, and the front-side sub-region and the right-front-side sub-region both form an attraction field. Alternatively, the first and second electrodes may be,

the i-region is the left-hand region, the front sub-region forms the attracting field and the right-hand sub-region forms the repelling field. Alternatively, the first and second electrodes may be,

the i-region is the left-hand region, the front sub-region forms the repulsive field, and the right-hand front sub-region forms the attractive field. Alternatively, the first and second electrodes may be,

the i area is the left area, and the front sub-area and the right front sub-area both form a repulsive field. Alternatively, the first and second electrodes may be,

the i-region is the right-side region, and the front-side sub-region and the left-front-side sub-region both form an attraction field. Alternatively, the first and second electrodes may be,

the i-region is the right-hand region, the front sub-region forms the attracting field and the left-hand front sub-region forms the repelling field. Alternatively, the first and second electrodes may be,

the i-region is the right-hand region, the front sub-region forms the repulsive field and the left-hand front sub-region forms the attractive field. Alternatively, the first and second electrodes may be,

And S103-2-2, judging the field intensity of the action fields of the traffic entities, the target bicycles and the riders thereof in each sub-area according to the action granularity, and utilizing a fuzzy logic method to form repulsion fields and attraction fields of the traffic entities, the target bicycles and the riders thereof in each sub-area.

For example, the field strengths of the repulsive field and the attractive field are denoted by "-" and "+", respectively. And obtaining the field intensity of the interaction field of the area where the target bicycle is located, the target bicycle and the rider thereof according to the field intensities of the interaction fields of the traffic entities of the front side and the rear side of the target bicycle and the rider thereof. The field intensity of the interaction field between the target bicycle and the interaction field between the target bicycle and the rider can be obtained by applying a derivation process similar to the field intensity of the target bicycle in the region where the target bicycle is located.

When the target bicycle is located in the left or right area of the self-passing phase of the person, the influence of one area on the target bicycle and the rider thereof mainly comprises the combined influence of the adjacent area and the separated area. The influence of the separated area on the target vehicle and the rider is mainly reflected in that when the traffic entity in the separated area generates the intentions of steering or changing the running track and the like, the traffic entity in the adjacent area of the target vehicle and the rider can be influenced, so that the intention of the target vehicle and the rider is influenced. Calculating the field intensity of the target bicycle side area, the target bicycle and the action field of the rider by using fuzzy logic, and listing the following two fuzzy inference rules:

1) if the interaction fields of the traffic entity, the target vehicle and the rider in the adjacent area and the separated area are respectively repulsive fields, the field intensity of the interaction field of the one side area, the target vehicle and the rider is repulsive field intensity.

2) If the interaction fields of the traffic entity, the target vehicle and the rider in the adjacent area and the separated area are respectively a repulsive field and an attractive field, the field intensity of the interaction field of the one side area, the target vehicle and the rider is zero.

Finally, 16 human-self-interacting phases are obtained, as shown in fig. 4, which are respectively:

the target bicycle is positioned in the middle area, and the field intensities of the front side, the left front side and the right front side are all plus;

the target bicycle is located in the middle area, the field intensity of the front side and the left front side is "+", and the field intensity of the right front side is "-";

the target bicycle is located in the middle area, the field intensity of the front side and the right front side is "+", and the field intensity of the left front side is "-";

the target bicycle is located in the middle area, the field intensity on the front side is "+", and the field intensity on the left front side and the right front side is "-";

the target bicycle is located in the middle area, the field intensity of the right front side and the left front side is "+", and the field intensity of the front side is "-";

the target bicycle is located in the middle area, the field intensity on the left front side is "+", and the field intensity on the right front side and the front side is "-";

the target bicycle is located in the middle area, the field intensity on the right front side is "+", and the field intensity on the front side and the left front side is "-";

the target bicycle is located in the middle area, and the field intensities of the front side, the left front side and the right front side are all "-";

the target bicycle is positioned in the left area, and the field intensities of the front side and the right front side are both "+";

the target bicycle is located in the left area, the front side is "+" and the right front side field strength is "-";

the target bicycle is located in the left area, the front side is "-", and the right front side field strength is "+";

the target bicycle is located in the left area, and the field intensities of the front side and the right front side are both "-";

the target bicycle is positioned in the right side area, and the field intensities of the front side and the left front side are both "+";

the target bicycle is located in the right area, the front side is "+", and the left front side field strength is "-";

the target bicycle is located in the right area, the front side is "-", and the left front side field strength is "+";

the target bicycle is located in the right area, and the field strengths of the front side and the left front side are both "-".

And S104, predicting the running track of the target bicycle according to the first data, the second data, the traffic phase state and the pre-established bicycle track prediction model.

For example, fig. 5-7 show graphs comparing predicted values and actual values of robust, conservative, and aggressive rider trajectories. And comparing the characteristic data of different types of riders collected and collated through experiments, predicting the running track of the target bicycle in a certain time period in real time, comparing and verifying the running track with the real track data to determine the effectiveness and reliability of the model prediction result, and correcting the model parameters according to the effectiveness and reliability to form a bicycle running track prediction model based on the dynamic Bayesian network.

The method provided by the embodiment is based on the self-mixing environment, the data of the bicycle rider is acquired by utilizing the human factor sensor, the GPS and other wireless sensing equipment, and the data is analyzed by utilizing the dynamic Bayesian network according to the wireless sensing equipment or a large amount of acquired data, so that the trajectory prediction of the target bicycle in the complex self-mixing environment is realized.

The method provided by the embodiment comprehensively considers the influence degree of 8 basic variables such as the target bicycle rider type, the target bicycle rider braking force, the target bicycle rider pedaling frequency, the target bicycle front wheel rotation angle, other traffic entity types, the relative transverse distance, the relative longitudinal distance and the relative speed on the transverse displacement, the longitudinal displacement and the running speed of the target bicycle in different person-self traffic phases, and establishes the standard for the value range of the transverse displacement, the longitudinal displacement and the running speed of the target bicycle. And determining a conditional probability matrix of the main component part of the running track of the target bicycle and a conditional probability matrix of the state change of the running track. And finally, obtaining the bicycle motion state transition probability to realize the prediction of the bicycle running track.

It should be noted that "first", "second", "third", and "fourth" in this embodiment and the following embodiments only play a role in identifying different data, and have no practical meaning.

The method provided by the invention comprises the steps of according to data of a target bicycle; the method comprises the steps of predicting the running track of a target bicycle according to data between the target bicycle and other traffic entities in an interest sensing area, traffic phases between the target bicycle and other traffic entities in the interest sensing area and a bicycle track prediction model established in advance, and achieving track prediction of the target bicycle in a complex self-mixing environment.

Referring to fig. 8, the present embodiment provides an electronic apparatus including: memory 801, processor 802, bus 803, and computer programs stored on memory 801 and executable on processor 802.

When the processor 802 executes the program, the following method is implemented:

s101, collecting first data of a target bicycle;

the target bicycle rider type is: conservative, or robust, or aggressive;

the other traffic entity types in the interest induction area are bicycles or pedestrians;

the braking force of the target bicycle rider is high, or moderate, or low;

the relative transverse distance between the target bicycle and other traffic entities in the interest sensing area is a relative transverse distance, or the relative transverse distance is in the middle, or the relative transverse distance is close;

dividing the road section into a left area, a middle area and a right area with the widths of 1 m respectively according to the maximum parking density and the maximum outline size of a target bicycle rider during riding;

Optionally, S103 includes:

S103-2, according to

And determining the traffic phase state.

Wherein the content of the first and second substances,

for the right back region to the target located in the i regionThe function granularity of the bicycle is determined by the following steps,

Optionally, S103-2 comprises:

s103-2-1, according to

Determining the field intensity type of the traffic phase state;

wherein the field intensity types are divided into strong repulsion field intensity, middle repulsion field intensity, weak repulsion field intensity, zero field intensity, weak attraction field intensity, middle attraction field intensity and strong attraction field intensity;

the action granularity of the j area on the target bicycles positioned in the i area is in an interval < -1 > -0.8), the field intensity type is strong repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in an interval < -0.8 > -0.4), the field intensity type is medium repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in an interval < -0.4,0, the field intensity type is weak repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is 0, the field intensity type is zero field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in (0, 0.3), the field intensity type is weak attraction field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in (0.3, 0.7), the field intensity type is medium attraction field intensity, the action granularity of the j area on the target bicycle in the i area is (0.7, 1), the field intensity type is strong absorption field intensity, j is the interest induction area identification, the j area is a right front side sub-area, or a right rear side sub-area, or a front side sub-area, or a rear side sub-area, or a left front side sub-area, or a left rear side sub-area.

Optionally, S103-2-2 comprises:

the traffic phase is:

Optionally, the bicycle track prediction model is established as follows:

s100-2, calculating the bicycle transverse displacement probability, the longitudinal displacement probability and the speed change probability corresponding to the sample data by adopting dynamic Bayesian network reasoning according to the sample data set;

any sample data comprises third data, fourth data and a sample traffic phase;

the bicycle rider brake force corresponding to the sample data is large, or medium, or small;

the pedaling frequency of a bicycle rider corresponding to the sample data is high, or medium, or low;

the bicycle corresponding to the sample data and other traffic entities corresponding to the sample data are far from the corresponding transverse distance, or in the corresponding transverse distance, or close to the corresponding transverse distance;

Optionally, the dynamic bayesian network inference process is:

wherein T is the total times of acquiring sample data, T is more than or equal to 1 and less than or equal to T, M is the number of observation nodes in the dynamic Bayesian network, M is more than or equal to 1 and less than or equal to M, K is the number of hidden nodes in the dynamic Bayesian network, K is more than or equal to 1 and less than or equal to K, and x_tkIs X_tkIs a value state of X_tkThe value y of the hidden node k in the sample data acquired at the t time_tmFor observing variable Y_tmValue of (A), Y_tmIs an observed variable, y, of the observed node m at the t-th acquisition_tm0Is Y_tm0Value of (A), Y_tm0Is the observed value of the observed node m at the t-th acquisition, pi (Y)_tm) Is Y_tmFather node, pi (X)_tk) Is X_tkThe parent node is a node of the network,P(Y_tm0＝y_tm) Is Y_tmBelongs to the continuous observed value of y_tmDegree of membership, P (x)_tk|π(X_tk) Is x)_tkAt the father node pi (X)_tk) Conditional probability of P (y)_tm|π(Y_tm) Is y)_tmAt the father node pi (Y)_tm) A conditional probability of;

degree of membership

The electronic equipment provided by the embodiment is used for processing the data of the target bicycle; the method comprises the steps of predicting the running track of a target bicycle according to data between the target bicycle and other traffic entities in an interest sensing area, traffic phases between the target bicycle and other traffic entities in the interest sensing area and a bicycle track prediction model established in advance, and achieving track prediction of the target bicycle in a complex self-mixing environment.

The present embodiments provide a computer storage medium that may be located on the robot or separate from the robot. The computer storage medium may be connected to the robot via a bus, may be connected to the robot via a wireless link, or may be connected to the robot via another link.

The computer storage medium performs the following operations:

s101, collecting first data of a target bicycle;

the target bicycle rider type is: conservative, or robust, or aggressive;

the braking force of the target bicycle rider is high, or moderate, or low;

Optionally, S103 includes:

S103-2, according to

And determining the traffic phase state.

Wherein the content of the first and second substances,

Optionally, S103-2 comprises:

s103-2-1, according to

Determining the field intensity type of the traffic phase state;

Optionally, S103-2-2 comprises:

the traffic phase is:

Optionally, the bicycle track prediction model is established as follows:

any sample data comprises third data, fourth data and a sample traffic phase;

Optionally, the dynamic bayesian network inference process is:

wherein T is the total times of acquiring sample data, T is more than or equal to 1 and less than or equal to T, M is the number of observation nodes in the dynamic Bayesian network, M is more than or equal to 1 and less than or equal to M, K is the number of hidden nodes in the dynamic Bayesian network, K is more than or equal to 1 and less than or equal to K, and x_tkIs X_tkIs a value state of X_tkThe value y of the hidden node k in the sample data acquired at the t time_tmFor observing variable Y_tmValue of (A), Y_tmIs an observed variable, y, of the observed node m at the t-th acquisition_tm0Is Y_tm0Value of (A), Y_tm0Is the observed value of the observed node m at the t-th acquisition, pi (Y)_tm) Is Y_tmFather node, pi (X)_tk) Is X_tkParent node, P (Y)_tm0＝y_tm) Is Y_tmBelongs to the continuous observed value of y_tmDegree of membership, P (x)_tk|π(X_tk) Is x)_tkAt the father node pi (X)_tk) Conditional probability of P (y)_tm|π(Y_tm) Is y)_tmAt the father node pi (Y)_tm) A conditional probability of;

degree of membership

for y in all sample data_tmIs measured.

The computer storage medium provided by the embodiment is used for storing the data of the target bicycle; the method comprises the steps of predicting the running track of a target bicycle according to data between the target bicycle and other traffic entities in an interest sensing area, traffic phases between the target bicycle and other traffic entities in the interest sensing area and a bicycle track prediction model established in advance, and achieving track prediction of the target bicycle in a complex self-mixing environment.

It is to be understood that the invention is not limited to the specific arrangements and instrumentality described above and shown in the drawings. A detailed description of known methods is omitted herein for the sake of brevity. In the above embodiments, several specific steps are described and shown as examples. However, the method processes of the present invention are not limited to the specific steps described and illustrated, and those skilled in the art can make various changes, modifications and additions or change the order between the steps after comprehending the spirit of the present invention.

It should also be noted that the exemplary embodiments mentioned in this patent describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above-described steps, that is, the steps may be performed in the order mentioned in the embodiments, may be performed in an order different from the order in the embodiments, or may be performed simultaneously.

Finally, it should be noted that: the above-mentioned embodiments are only used for illustrating the technical solution of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A bicycle track prediction method facing a self-mixed environment, which is characterized by comprising the following steps:

s101, collecting first data of a target bicycle;

s104, predicting the running track of the target bicycle according to the first data, the second data, the traffic phase and a bicycle track prediction model established in advance;

the first data includes: target bicycle rider type, target bicycle rider grip, target bicycle rider braking force, target bicycle rider pedaling frequency, target bicycle front wheel turning angle, target bicycle lateral displacement, longitudinal displacement and running speed;

the target bicycle rider type is: conservative, or robust, or aggressive;

the braking force of the target bicycle rider is high, or moderate, or low;

the relative longitudinal distance between the target bicycle and other traffic entities in the interest sensing area is far relative longitudinal distance, or in the relative longitudinal distance, or close relative longitudinal distance;

the method for determining the interest sensing area comprises the following steps:

based on the left, middle and right three areas, dividing the three areas into a right front sub-area, a right rear sub-area, a right secondary rear sub-area, a front sub-area, a rear sub-area, a left front sub-area, a left rear sub-area and a left secondary rear sub-area by taking the position of the target bicycle front wheel axle as a base point;

the S103 includes:

s103-1, calculating the action granularity of other traffic entity types in the interest induction area on the target bicycle by using a fuzzy inference rule based on other traffic entity types in the interest induction area, the relative speed of the target bicycle and other traffic entities in the interest induction area, the relative transverse distance between the target bicycle and other traffic entities in the interest induction area and the relative longitudinal distance between the target bicycle and other traffic entities in the interest induction area

S103-2, according to

Determining the phase of the traffic phase, and determining the phase of the traffic phase,

wherein the content of the first and second substances,

the granularity of the action of the rear area on the target bicycle located in the i area,

the action granularity of the right next rear area on the target bicycle located in the i area;

the S103-2 comprises:

s103-2-1, according to

Determining the field intensity type of the traffic phase state;

the action granularity of the j area on the target bicycles positioned in the i area is in an interval of [ -1, -0.8 ], the field intensity type is strong repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in an interval of [ -0.8, -0.4 ], the field intensity type is medium repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in an interval of [ -0.4,0 ], the field intensity type is weak repulsion field intensity, the action granularity of the j area on the target bicycles positioned in the i area is 0, the field intensity type is zero field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in (0,0.3], the field intensity type is weak attraction field intensity, the action granularity of the j area on the target bicycles positioned in the i area is in (0.3,0.7, the field intensity type is a medium attraction field intensity, the action granularity of the j area on the target bicycle positioned in the i area is in (0.7, 1), the field intensity type is a strong attraction field intensity, j is an interest induction area mark, and the j area is a right front side sub-area, a right rear side sub-area, a front side sub-area, a rear side sub-area, a left front side sub-area, a left rear side sub-area, or a left rear side sub-area;

if the field intensity types are strong repulsion field intensity, middle repulsion field intensity and weak repulsion field intensity, determining to form a repulsion field;

the traffic phase state is as follows:

the i area is a right area, and a front sub-area and a left front sub-area both form a repulsive field;

the method for establishing the bicycle track prediction model comprises the following steps:

any sample data comprises third data, fourth data and a sample traffic phase;

the bicycle corresponding to the sample data and other traffic entities corresponding to the sample data are far from each other in relative longitudinal distance, or close to each other in relative longitudinal distance;

the dynamic Bayesian network reasoning process comprises the following steps:

degree of membership

for y in all sample data_tmIs measured.

2. An electronic device comprising a memory, a processor, a bus and a computer program stored on the memory and executable on the processor, the processor implementing the steps of any of claim 1 when executing the program.

3. A computer storage medium having a computer program stored thereon, characterized in that: the program when executed by a processor implementing the steps of any of claim 1.