CN118980372A

CN118980372A - Vehicle positioning method, vehicle positioning device and system based on driving behavior and inertial navigation under GNSS short-time failure

Info

Publication number: CN118980372A
Application number: CN202411079196.8A
Authority: CN
Inventors: 李志强; 田建; 姜一洲; 董龙威; 周京; 彭建华; 姜瑶; 冯雯雯; 李席宇; 闫磊; 孟琳; 牟凯; 席建锋; 丁同强; 郑黎黎
Original assignee: China Academy of Transportation Sciences
Current assignee: China Academy of Transportation Sciences
Filing date: 2024-08-07
Publication date: 2024-11-19

Abstract

The application relates to the technical field of vehicle positioning, and discloses a vehicle positioning method under GNSS short-time failure based on driving behavior and inertial navigation. The vehicle positioning method comprises the following steps: obtaining and determining driving state information according to driving behavior information; predicting the occurrence probability of each preset motion constraint model according to the driving behavior information and the driving state information; determining a mixed probability matrix according to occurrence probabilities corresponding to all preset motion constraint models; acquiring inertial positioning data, and calculating the overall estimation value and the overall covariance of the filtering result of each preset motion constraint model according to the inertial positioning data and the mixing probability; and correcting the inertial positioning data according to the overall estimated value and the overall covariance, and determining the position information of the vehicle according to the corrected inertial positioning data. The vehicle positioning method can improve the positioning accuracy. The application also discloses a vehicle positioning device and a vehicle positioning system based on the driving behavior and the inertial navigation under the GNSS short-time failure.

Description

Vehicle positioning method, vehicle positioning device and system based on driving behavior and inertial navigation under GNSS short-time failure

Technical Field

The application relates to the technical field of vehicle positioning, in particular to a vehicle positioning method, a vehicle positioning device and a system under GNSS short-time failure based on driving behavior and inertial navigation.

Background

At present, with the influx of a great deal of funds, technology and talents into the automatic driving industry, intelligent automobiles can cope with most driving scenes. The mainstream high-precision positioning scheme for the automatic driving vehicle is as follows: the GPS (Global Navigation SATELLITE SYSTEM, GNSS) is combined with the inertial measurement unit (Inertial Measurement Unit, IMU) and adopts a Real-time differential positioning technology (Real-TIME KINEMATIC, RTK) of high-precision positioning to effectively realize centimeter-level positioning. The GNSS satellite navigation is responsible for defining absolute positions, the IMU inertial navigation is responsible for defining relative positions of the vehicle body, the position positioning supplement can be carried out on road sections shielded by satellite signals such as tunnels, and the self-position sensing of the vehicle body is realized through a fusion algorithm. And then combining with sensors such as a laser radar/a camera and the like, matching the environmental characteristics with a high-precision map through point cloud description, and further determining the position of the vehicle, thereby realizing accurate positioning with higher redundancy.

The positioning scheme fully utilizes the advantages of different positioning approaches, and plays a role in complementary advantages.

In the process of implementing the embodiment of the application, the related art is found to have at least the following problems:

With the rapid development of three-dimensional traffic, more and more tunnels, overpasses and the like are put into use, and due to the influence of non-line-of-sight and multipath effects, GNSS signals have the defects of long-term failure, local pollution and the like, especially, the defects of failure for a few minutes or even tens of minutes often exist in the tunnels, and the problem of local pollution of the GNSS signals caused by the overpasses, cities and the like is often difficult to predict and control. The IMU has a high frequency that can provide relative position information to assist positioning during periods of GNSS inactivity, but its positioning errors accumulate over time and distance, and the positioning errors of civilian low cost IMUs have a significant impact on automated driving high accuracy positioning.

It should be noted that the information disclosed in the above background section is only for enhancing understanding of the background of the application and thus may include information that does not form the prior art that is already known to those of ordinary skill in the art.

Disclosure of Invention

The following presents a simplified summary in order to provide a basic understanding of some aspects of the disclosed embodiments. This summary is not an extensive overview, and is intended to neither identify key/critical elements nor delineate the scope of such embodiments, but is intended as a prelude to the more detailed description that follows.

The embodiment of the application provides a vehicle positioning method under GNSS short-time failure based on driving behavior and inertial navigation, which is used for reducing the positioning error of an IMU and improving the positioning precision during the period of non-working of the GNSS.

In some embodiments, a method of vehicle localization in a GNSS short time failure based on driving behavior and inertial navigation, comprising:

Obtaining driving behavior information and determining driving state information according to the driving behavior information; the driving behavior information includes: steering wheel angle, vehicle lateral speed, and yaw rate; the driving state information includes: straight, left lane change and right lane change;

according to the driving behavior information and the driving state information, each preset motion constraint model occurrence probability; the preset motion constraint model comprises the following steps: a uniform linear motion model, a uniform acceleration linear motion model, a constant turning rate and speed motion model and a constant turning rate and acceleration model;

determining a mixed probability matrix according to the occurrence probabilities corresponding to all the preset motion constraint models;

Acquiring inertial positioning data, and calculating the overall estimation value and the overall covariance of the filtering result of each preset motion constraint model according to the inertial positioning data and the mixing probability; the inertial positioning data includes one or more of yaw rate information, course angle information, and acceleration information;

And correcting the inertial positioning data according to the overall estimated value and the overall covariance, and determining the position information of the vehicle according to the corrected inertial positioning data.

By adopting the vehicle positioning method under the GNSS short-time failure based on the driving behavior and the inertial navigation, the following technical effects can be achieved:

Driving behavior information such as steering wheel rotation angle, vehicle transverse speed and yaw rate belongs to short-term microscopic information, driving state information is determined according to the short-term microscopic data, driving state information such as straight driving, left lane changing and right lane changing belongs to long-term macroscopic information, and then in the process of predicting occurrence probability of each preset motion constraint model according to the driving state information and the driving state information, the driving state information can show long-term macroscopic constraint characteristics for vehicle positioning, namely, the occurrence probability of the preset motion constraint model can show long-term macroscopic constraint characteristics for vehicle positioning. The mixed probability of occurrence probabilities corresponding to all preset motion constraint models is enabled to show long-term macroscopic constraint characteristics, and then in the interactive multi-model filter (INTERACTING MULTIPLE MODEL FILTER, IMM), the mixed probability with the long-term macroscopic constraint characteristics is used as input, so that the filtering precision and stability of the IMM can be improved, and further the overall estimated value and the overall covariance with higher precision and stability are obtained.

In the process of correcting the inertial positioning data by using the overall estimated value and the overall covariance, the accumulated error of the IMU is reduced by using the data with certain long-term macroscopic constraint characteristics, the data with short-term microscopic constraint characteristics substituted in the correction process is reduced as much as possible, the high-frequency and short-term precise positioning characteristics provided by the IMU are reserved as much as possible, and the position information of the vehicle determined according to the corrected inertial positioning data has higher precision and robustness.

Optionally, modifying the inertial positioning data according to the overall estimate and the overall covariance, including:

Determining a first weight of the overall estimation value according to the overall covariance; the first weight is inversely related to the uncertainty represented by the overall covariance;

Determining a second weight of the inertial positioning data according to covariance of the inertial positioning data forming matrix; the second weight is inversely related to uncertainty represented by covariance of the inertial positioning data formation matrix;

Calculating a weighted average or a weighted sum according to the overall estimation value, the first weight, the inertial positioning data formation matrix and the second weight;

The second weighted average or weighted sum is used as the corrected inertial positioning data.

Optionally, determining the first weight of the overall estimation value according to the overall covariance includes:

Optionally, determining the second weight of the inertial positioning data from the covariance of the inertial positioning data constituent matrix comprises:

Where w _IMU is the second weight, P _IMU is the covariance of the inertial positioning data constituent matrix, trace (P _IMU) is the trace of P _IMU, M is the dimension of P _IMU, (P _IMU)_ii is the i-th diagonal element of P _IMU).

Optionally, calculating an overall estimation value and an overall covariance of a filtering result of each preset motion constraint model according to the inertial positioning data and the mixing probability, including:

The initial condition of each preset motion constraint model filtering algorithm is determined by the following method:

wherein, Predicting a state of the model j for the moment k-1; r is the number of models, r=4; Predicting a state of the model i for the moment k-1; p ^j,0 (k-1|k-1) is the covariance predictor of time k-1 for model j; a state covariance predicted value of the model i is a time k-1; the probability of mixing from model i to model j for time k-1;

p _ij is the transition probability from model i to model j, which is the transition probability matrix element of the prior Markov model; The occurrence probability of the model i is the moment k-1;

The overall estimate and the overall covariance are determined by:

p_j(k|k-1)＝φ_j(k-1)p^j,0(k-1|k-1)φ_j(k-1)^T+B_jQ_jB_j ^T

K_j(k)＝p_j(k|k-1)H^T[Hp_j(k|k-1)H^T+P]^-1

p_j(k|k)＝[I-K_j(k)H(k)]p_j(k|k-1)

wherein, Predicting a state of the model j for the moment k; phi _j (k-1) is a state transition matrix at the time k-1, which is used for describing the state change relation of the system from the time k-1 to the time k; The state predicted value of the model j at the moment k-1, namely the posterior state estimation at the previous moment; p _j (k|k-1) is the covariance prediction value of time k on model j; p ^j,0 (k-1|k-1) is the covariance predicted value of the model j at time k-1, i.e. the posterior covariance estimate at the previous time; b _j is a control input matrix for representing the influence of noise on the system state; q _j is a process covariance matrix for representing the magnitude of system process noise; b _j ^T is the transpose of the control input matrix B _j; k _j (K) is a Kalman gain matrix of the moment K to the model j, and determines the update weight of the observed value to the state estimation; h is an observation matrix for representing a relationship between the state vector and the observation vector; r is an observed noise covariance matrix; updating a matrix for the state of the model j at the moment k; z (k) is an observation vector matrix at the moment k and is composed of inertial positioning data; is the observation residual; p _j (k|k) is the covariance update matrix of time k to model j; i is an identity matrix.

Model probability updating is performed by the following likelihood functions:

wherein, lambada _j (k) is the likelihood function of the time k model j; is a transpose of the measurement method of model j, Is the residual covariance of model j; the model probability is updated as follows:

Wherein μ _j (k) is the posterior probability of the time k model j, μ _j (k-1) is the posterior probability of the time k-1 model j;

the overall estimate and the overall covariance are:

wherein, Is the overall estimate of time k; p (k|k) is the overall covariance of time k.

Optionally, determining driving state information according to the driving behavior information includes: obtaining driving behavior information of a plurality of continuous moments before the current moment to form an observation sequence of the driving behavior information; and determining driving state information according to the observation sequence of the driving behavior information.

Optionally, determining driving state information according to the observed sequence of driving behavior information includes:

And inputting the observation sequence of the driving behavior information into a hidden Markov model (Hidden Markov Model, HMM) to obtain the driving state information output by the HMM.

Optionally, predicting the occurrence probability of each preset motion constraint model according to the driving behavior information and the driving state information includes:

obtaining a plurality of sequences of the driving behavior information and the driving state information according to the sequence of the time stamps;

And inputting the multiple sequences of the driving behavior information and the driving state information into an artificial neural network model with sequence processing capability, and obtaining the occurrence probability of the multiple preset motion constraint models output by the artificial neural network model with sequence processing capability.

Optionally, determining a mixed probability matrix according to occurrence probabilities corresponding to all the preset motion constraint models includes:

Determining a transition probability matrix of a priori Markov model according to the occurrence probabilities corresponding to all the preset motion constraint models;

and determining the mixed probability matrix according to the transition probability.

In some embodiments, a vehicle positioning device under GNSS short time failure based on driving behavior and inertial navigation includes a driving state prediction module, a motion model probability prediction module, a first filtering module, a second filtering module, and a correction module.

The driving state prediction module is used for obtaining driving behavior information and determining driving state information according to the driving behavior information; the driving behavior information includes: steering wheel angle, vehicle lateral speed, and yaw rate; the driving state information includes: straight, left lane change and right lane change;

the motion model probability prediction module is used for predicting the occurrence probability of each preset motion constraint model according to the driving behavior information and the driving state information; the preset motion constraint model comprises the following steps: a uniform linear motion model, a uniform acceleration linear motion model, a constant turning rate and speed motion model and a constant turning rate and acceleration model;

the first filtering module is used for determining a mixed probability matrix according to the occurrence probabilities corresponding to all the preset motion constraint models;

The second filtering module is used for obtaining inertial positioning data and calculating the overall estimation value and the overall covariance of a Kalman filtering algorithm according to the inertial positioning data and the mixing probability; the inertial positioning data includes one or more of yaw rate information, course angle information, and acceleration information;

the correction module is used for correcting the inertial positioning data according to the overall estimated value and the overall covariance, and determining the position information of the vehicle according to the corrected inertial positioning data.

The vehicle positioning device under the GNSS short-time failure based on the driving behavior and the inertial navigation also has the technical effect of improving the positioning precision, and the specific analysis process refers to the analysis of the vehicle positioning method under the GNSS short-time failure based on the driving behavior and the inertial navigation, and is not repeated here.

In some embodiments, a vehicle positioning device under GNSS short time failure based on driving behavior and inertial navigation includes a processor and a memory storing program instructions, the processor being configured to execute the vehicle positioning method under GNSS short time failure based on driving behavior and inertial navigation provided in the foregoing embodiments when executing the program instructions.

The vehicle positioning device based on the driving behavior and the inertial navigation under the GNSS short-time failure also has the technical effect of improving the positioning precision.

In some embodiments, the vehicle positioning system includes the vehicle positioning device provided by the previous embodiments under GNSS short time failure based on driving behavior and inertial navigation.

Also, the vehicle positioning device has the technical effect of improving the positioning accuracy.

The foregoing general description and the following description are exemplary and explanatory only and are not restrictive of the application.

Drawings

One or more embodiments are illustrated by way of example and not limitation in the figures of the accompanying drawings, in which like references indicate similar elements, and in which:

fig. 1 is a schematic flow chart of a vehicle positioning method under GNSS short time failure based on driving behavior and inertial navigation according to an embodiment of the present application;

FIG. 2 is a schematic diagram of a vehicle positioning device under GNSS short time failure based on driving behavior and inertial navigation according to an embodiment of the present application;

FIG. 3 is a schematic diagram of a specific implementation flow of a method for positioning a vehicle under a GNSS short time failure based on driving behavior and inertial navigation according to an embodiment of the present application;

Fig. 4 is a schematic diagram of a vehicle positioning device under GNSS short time failure based on driving behavior and inertial navigation according to an embodiment of the present application.

Detailed Description

For a more complete understanding of the nature and the technical content of the embodiments of the present application, reference should be made to the following detailed description of embodiments of the application, taken in conjunction with the accompanying drawings, which are meant to be illustrative only and not limiting of the embodiments of the application. In the following description of the technology, for purposes of explanation, numerous details are set forth in order to provide a thorough understanding of the disclosed embodiments. However, one or more embodiments may still be practiced without these details. In other instances, well-known structures and devices may be shown simplified in order to simplify the drawing.

The terms first and second and the like in the description and in the claims of embodiments of the application and in the above-described figures are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate in order to describe embodiments of the application herein. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion.

The term "plurality" means two or more, unless otherwise indicated.

In the embodiment of the application, the character "/" indicates that the front object and the rear object are in an OR relationship. For example, A/B represents: a or B.

The term "and/or" is an associative relationship that describes an object, meaning that there may be three relationships. For example, a and/or B, represent: a or B, or, A and B.

Fig. 1 is a schematic flow chart of a vehicle positioning method under GNSS short time failure based on driving behavior and inertial navigation according to an embodiment of the present application. The vehicle positioning method under the GNSS short time failure based on the driving behavior and the inertial navigation can be executed by a vehicle positioning navigation device or can also be executed by mobile equipment.

Referring to fig. 1, a vehicle positioning method under GNSS short time failure based on driving behavior and inertial navigation includes:

s101, driving behavior information is obtained, and driving state information is determined according to the driving behavior information.

Wherein the driving behavior information includes: steering wheel angle, vehicle lateral speed, and yaw rate. The driving state information includes: straight, left lane change and right lane change.

In some implementations, the steering wheel angle may be obtained by a steering wheel angle sensor, the vehicle lateral speed may be obtained by a wheel speed sensor, and the yaw rate may be obtained by IMU detection.

The steering wheel angle sensor can be a built-in sensor of a vehicle or an additionally installed sensor; the wheel speed sensor can be a built-in sensor of a vehicle or an additionally installed sensor; the IMU may be a built-in IMU of the vehicle, or may be other IMUs fixed to the vehicle that are stationary relative to the vehicle.

The sensors of different classes have different sampling frequencies, and the sampling reference frequency can be determined according to the sampling frequencies of the sensors, and then the sampling data of each sensor is time-stamped and aligned according to the reference sampling frequency.

Wherein, in determining the sampling reference frequency according to the sampling frequencies of the plurality of sensors, the lowest sampling frequency of the plurality of sensors may be taken as the sampling reference frequency, or the average sampling frequency of the plurality of sensors may be taken as the sampling reference frequency.

In the process of performing time stamp alignment on the sampling data of each sensor according to the reference sampling frequency, a linear interpolation method can be adopted to perform time stamp alignment on the sampling data of each sensor; the sampled data of each sensor may also be time stamped using polynomial difference.

Of course, the above-listed two ways of determining the sampling reference frequency from the sampling frequencies of the plurality of sensors, and two ways of time-stamping the sampling data of each sensor according to the reference sampling frequency are merely exemplary. In a specific application scenario, a person skilled in the art can also determine the manner of determining the sampling reference frequency according to the actual application situation according to experience, and perform the manner of time stamp alignment.

The above-listed driving behavior information includes steering wheel rotation angle, vehicle lateral speed, and yaw rate, and is used to describe that the driving behavior information is microscopic vehicle characteristic information of the vehicle itself in a short period of time. In a specific application scenario, the person skilled in the art may also detect other specific types of driving behavior information of the vehicle itself, similar to the above-mentioned steering wheel information, vehicle lateral speed and yaw rate, in short-term microscopic level, which will not be described in detail here.

The above-listed driving state information includes straight, left lane change, and right lane change, and is used to illustrate that the driving state information is relatively long-term macroscopic vehicle-specific information. In a specific application scenario, the person skilled in the art may also obtain other types of specific driving status information similar to the long-term macroscopic driving status information of the straight driving, the left lane change and the right lane change, which are not described in detail herein.

The step of determining driving state information according to driving behavior information is used for converting short-term microcosmic constraint characteristics of vehicle positioning into long-term macroscopic constraint characteristics.

Optionally, determining driving state information according to driving behavior information includes: obtaining driving behavior information of a plurality of continuous moments before the current moment to form an observation sequence of the driving behavior information; and determining driving state information according to the observation sequence of the driving behavior information.

Specifically, determining driving state information according to an observation sequence of driving behavior information includes: the observation sequence of the driving behavior information is input into the HMM, and the driving state information output by the HMM is obtained.

For example, the observation sequence of the driving behavior information may be expressed as follows:

o=o ₁O₂O₃...O_T, where O is an observation sequence of driving behavior information, and T is a length of a time sequence; observation value at t time in observation sequence of driving behavior information Is the angle of rotation of the steering wheel,For the transverse velocity of the vehicle,Is yaw rate. Of course, in the specific application process, if the types of driving behavior information determined according to the actual situation are more, the number of elements in the O _t is adaptively increased, which is not described in detail herein.

The observation sequence O of the driving behavior information is input into the HMM, and the output of the HMM is the driving state information.

Of course, the HMM refers to a trained HMM.

In the training process of the HMM, an observation sequence of driving behavior information and a state sequence of driving state information are collected at the same time, for example, the observation sequence of driving behavior information and the state sequence of driving state information are recorded as { O, Q }, where O is the observation sequence of driving behavior information, Q is the state sequence of driving state information, q=q ₁q₂q₃...q_T, T is the length of time sequence, and driving state information can be obtained through manual recording or through a photographing device provided in a test road section; the state at the t-th time in the state sequence Q of the driving state information is a value set s= { S ₂,s₃,…,s_N } of Q _t,q_t, and N is the type number of the driving state information of the vehicle. In the embodiment of the present application, an exemplary explanation is given taking n=3 as an example (driving state information includes straight, left lane change, and right lane change).

According to the markov assumption, the state of the system at any time is only related to the state at the last time, and then the joint probability distribution of the observation sequence O of driving behavior information and the state sequence Q of driving state information is:

Wherein pi is the state probability of the vehicle driving state information at the initial moment, pi= (pi ₁,π₁,…,π_N), N is the type number of the vehicle driving state information, pi _i＝P(q₁＝s_i), i epsilon [1, N ];

A is the probability of the system transitioning between states, typically an N-dimensional matrix, with elements a _ij＝P(q_t+1＝s_j|q_t＝s_i), i, j e [1, N ], t e [1, T-1] in matrix A;

Phi is the probability that the system outputs observations in each state.

The system outputs the probability b _i(O_t)＝P(O_t|q_t＝s_i, phi of the value of the observed variable O in the state of t moment i, i epsilon [1, N ], t epsilon [1, T ].

The HMM for identifying the vehicle state information may be obtained by training the HMM with the obtained offline vehicle driving information and the vehicle state information.

Of course, the above observation sequence for processing driving behavior information by using HMM is only illustrative, and other algorithm models with processing sequence capability similar to Long Short-Term Memory (LSTM) and its deformation algorithm may be used to obtain driving state information.

S102, predicting the occurrence probability of each preset motion constraint model according to the driving behavior information and the driving state information.

The preset motion constraint model comprises the following steps: a uniform linear motion model, a uniform acceleration linear motion model, a constant turning rate and speed motion model and a constant turning rate and acceleration model.

obtaining a plurality of sequences of driving behavior information and driving state information according to the sequence of the time stamps;

And inputting a plurality of sequences of the driving behavior information and the driving state information into the artificial neural network model with the sequence processing capability, and obtaining the occurrence probability of each preset motion constraint model output by the artificial neural network model with the sequence processing capability.

In the case where the preset motion constraint model includes the above four models, the output of the models generally includes the occurrence probabilities of the four models of the uniform velocity linear motion model, the uniform acceleration linear motion model, the constant turning rate and velocity motion model, and the constant turning rate and acceleration model at the same time.

LSTM may be used as the artificial neural network model with sequence processing capabilities.

In LSTM, the reservation condition of the last moment is determined first, this function is completed by a forgetting gate, and the formula of this function is:

F_t＝σ(W_f·[h_t-1,X_t]+b_f)

Wherein F _t is a forgetting gate at time t, sigma is a sigmoid activation function, W _f is weight of the forgetting gate, h _t-1 is a hidden state at time t-1, X _t is vehicle driving information and vehicle state information input at time t, and b _f is bias of the forgetting gate.

The forgetting gate maps the input and the state at the last moment to a value between 0 and 1 through a Sigmoid function to determine the reservation condition of the state at the last moment, wherein 1 represents complete reservation and 0 represents rejection.

The output information at the current moment is then determined by the input gate.

The input gate firstly constructs candidate vectors through a tanh function, and the calculation formula is as follows:

wherein, As candidate vectors, W _c is a weight when constructing candidate vectors, h _t-1 is a hidden state at time t-1, X _t is vehicle driving information and vehicle state information input at time t, and b _c is a bias when constructing candidate vectors.

The input gate selects the forgetting proportion again, and the calculation formula is as follows:

i_t＝σ(W_i·[h_t-1,X_t]+b_i)

wherein i _t is an input gate at time t, σ is a Sigmoid function, W _i is a weight of the input gate, h _t-1 is a hidden state at time t-1, X _t is vehicle driving information and vehicle state information input at time t, and b _i is a bias of the input gate.

Finally, the output value is determined by the output gate, and the calculation formula is as follows:

O_t＝σ(W_o[h_t-1,X_t]+b_o)

h_t＝O_t·tanh(C_t)

Wherein C _t is the memory cell at time t, F _t is the forgetting gate at time t, C _t-1 is the memory cell at time t-1, i _t is the forgetting gate at time t, For the candidate vector of time t, O _t is the output gate of time t, σ is the Sigmoid function, W _o is the weight of the output gate, h _t-1 is the hidden state of time t, X _t is the vehicle driving information and the vehicle state information input at time t, b _o is the weight of the output gate, and h _t is the hidden state of time t.

The LSTM can be trained through the obtained observation sequence of the driving behavior information, the obtained observation sequence of the driving state information and the preset motion constraint model, and the occurrence probability of each preset motion constraint model is predicted.

The above list of LSTM is for illustrative purposes only, and other artificial neural network models with sequence processing capabilities may be selected by those skilled in the art in specific application scenarios. For example, the most basic recurrent neural network (Recurrent Neural Network, RNN) can be employed, namely: and inputting a plurality of sequences of driving behavior information and driving state information into the RNN to obtain the occurrence probability of each preset motion constraint model output by the RNN.

S103, determining a mixed probability matrix according to the occurrence probabilities corresponding to all the preset motion constraint models.

The transition probability matrix of the prior Markov model can be determined according to the occurrence probabilities corresponding to all the preset motion constraint models; a mixing probability matrix is determined from the transition probability matrix.

Under the condition that the preset motion constraint model comprises a uniform linear motion model, a uniform acceleration linear motion model, a constant turning rate and speed motion model and a constant turning rate and acceleration model, the transition probability matrix refers to transition probabilities of the four models, namely the uniform linear motion model, the uniform acceleration linear motion model, the constant turning rate and speed motion model and the constant turning rate and acceleration model. Of course, in the specific application process, if the preset motion constraint model further comprises other models, a transition probability matrix between all the preset motion constraint models is adaptively calculated.

The transition probability matrix is further described below:

wherein pi is a transition probability matrix, and the four models are ordered according to the following sequence: transition probabilities of the four models, namely a uniform linear motion model, a uniform acceleration linear motion model, a constant turning rate and speed motion model and a constant turning rate and acceleration model; each element in the transition probability matrix pi is explained as follows:

p ₁₁ =p (uniform linear motion→uniform linear motion)

P ₁₂ =p (uniform linear motion→uniform acceleration linear motion)

P ₁₃ =p (uniform linear motion→constant turning rate and velocity motion)

P ₁₄ =p (uniform linear motion→constant turning rate and acceleration motion)

P ₂₁ =p (uniform acceleration linear motion → uniform linear motion)

P ₂₂ =p (uniform acceleration linear motion → uniform acceleration linear motion)

P ₂₃ =p (uniform acceleration linear motion→constant turning rate and velocity motion)

P ₂₄ =p (uniform acceleration linear motion→constant turning rate and acceleration motion)

P ₃₁ =p (constant turning rate and speed motion→uniform linear motion)

P ₃₂ =p (constant turning rate and speed motion→uniform acceleration linear motion)

P ₃₃ = P (constant turning rate and speed movement → constant turning rate and speed movement)

P ₃₄ = P (constant turning rate and speed motion → constant turning rate and acceleration motion)

P ₄₁ =p (constant turning rate and acceleration motion→uniform linear motion)

P ₄₂ =p (constant turning rate and acceleration motion→uniform acceleration linear motion)

P ₄₃ = P (constant turning rate and acceleration motion → constant turning rate and velocity motion)

P ₄₄ = P (constant turning rate and acceleration motion → constant turning rate and acceleration motion)

In the process of determining the transition probability matrix of the prior Markov model, the transition probability matrix can be obtained by analyzing historical data, counting the transition frequencies among the four models, namely a uniform linear motion model, a uniform acceleration linear motion model, a constant turning rate and speed motion model and a constant turning rate and acceleration model, and then normalizing. Or one skilled in the art may also empirically set a suitable transition probability matrix.

S104, acquiring inertial positioning data, and calculating the overall estimation value and the overall covariance of the filtering result of each preset motion constraint model according to the inertial positioning data and the mixing probability.

The overall estimation value is an estimation value of the inertial positioning data, and the overall covariance is used for representing uncertainty of the estimation value of the inertial positioning data or statistical characteristics of an estimation value error of the inertial positioning data.

The overall estimation value and the overall covariance are filtering results of a filtering algorithm with a state estimation function.

The filtering algorithm may be a Kalman filter (KALMAN FILTER) and its extension algorithm, such as an Extended Kalman filter (Extended KALMAN FILTER, EKF), unscented Kalman filter (Unscented KALMAN FILTER, UKF), etc.

The filtering algorithm may also be other filtering algorithms than a particle Filter (PARTICLE FILTER), an H-INFINITY FILTER, a Bayesian Filter (Bayesian Filter), etc. KALMAN FILTER.

The inertial positioning data includes one or more of yaw rate information, heading angle information, and acceleration information. The inertial positioning data may be obtained by one or more of an IMU, a magnetic sensor, an accelerometer.

Alternatively, the inertial positioning data may include yaw rate information and acceleration information, or the inertial positioning data may include heading angle information and acceleration information, or the inertial positioning data may include yaw rate information, heading angle information, and acceleration information.

The inertial positioning data is a quantitative measure of the filtering algorithm used to correct and update the system state estimate. That is, the inertial positioning data is input into the filtering algorithm, so that the overall estimated value and the overall covariance output by the filtering algorithm can be obtained.

S105, correcting the inertial positioning data according to the overall estimated value and the overall covariance, and determining the position information of the vehicle according to the corrected inertial positioning data.

The position information of the vehicle can be calculated from the corrected inertial positioning data by:

Calculating integration of the corrected inertial positioning data at the GNSS failure moment, wherein the integration result is the position information of the vehicle; or calculating the position information of the vehicle from the corrected inertial positioning data by means of an inertial navigation algorithm (Inertial Navigation System, INS). Or the person skilled in the art can select an algorithm conforming to the actual situation according to own experience and the actual situation, and calculate the position information of the vehicle according to the corrected inertial positioning data.

Optionally, correcting the inertial positioning data according to the overall estimate and the overall covariance correction, including: determining an adjustment coefficient of the overall estimation value according to the overall covariance; obtaining the product of the overall estimation value and the adjustment coefficient; correcting the inertial positioning data according to the product to obtain corrected inertial positioning data; wherein the adjustment coefficient is inversely related to the uncertainty represented by the overall covariance.

The higher the correlation of the adjustment coefficient with the uncertainty represented by the overall covariance, e.g., the greater the absolute value of the correlation coefficient, the lower the degree of correction of the inertial positioning data by the overall estimate value. That is, only under the condition that the accuracy of the overall covariance is higher, the inertial positioning data is corrected to a greater extent by the overall covariance, otherwise, the inertial positioning data is corrected to a lesser extent by the overall covariance.

In a specific application process, in the case that the accuracy of the filtering result is confirmed to be higher by a test mode, the correlation degree between the adjustment coefficient and the uncertainty represented by the overall covariance can be set to be lower, for example, the absolute value of the correlation coefficient between the adjustment coefficient and the uncertainty is smaller; in the case where the accuracy of the filtering result is confirmed to be relatively low by a trial method, the degree of correlation between the adjustment coefficient and the uncertainty represented by the overall covariance may be set to be high, for example, the absolute value of the correlation coefficient between the two may be large. The correlation coefficient between the adjustment coefficient according with the actual situation and the uncertainty represented by the overall covariance can be set by a person skilled in the art, so as to obtain the adjustment coefficient according with the actual situation, further obtain relatively accurate corrected inertial positioning data, and finally obtain the position information of the vehicle.

Further, the inertial positioning data is corrected according to the product, and corrected inertial positioning data is obtained, including: and taking the sum of the product and the inertial positioning data as corrected inertial positioning data, wherein the sum is vector sum.

Optionally, correcting the inertial positioning data according to the overall estimate and the overall covariance, including: determining a first weight of the overall estimation value according to the overall covariance; the first weight is inversely related to the uncertainty represented by the overall covariance; determining a second weight of the inertial positioning data according to covariance of the inertial positioning data forming matrix; the second weight is inversely related to uncertainty represented by covariance of the inertial positioning data formation matrix; according to the overall estimation value, the first weight, the inertia positioning data forming matrix and the second weight, calculating a weighted average value or a weighted sum; the second weighted average or weighted sum is used as the corrected inertial positioning data.

The higher the correlation of the first weight with the uncertainty represented by the overall covariance, e.g., the greater the absolute value of the correlation coefficient, the lower the degree of correction of the inertial positioning data by the overall estimate value. That is, only under the condition that the accuracy of the overall covariance is higher, the inertial positioning data is corrected to a greater extent by the overall covariance, otherwise, the inertial positioning data is corrected to a lesser extent by the overall covariance.

In a specific application process, in the case that the accuracy of the filtering result is confirmed to be relatively high in a test manner, the correlation degree between the first weight and the uncertainty represented by the overall covariance can be set to be relatively low, for example, the absolute value of the correlation coefficient between the first weight and the uncertainty is relatively low; in case the accuracy of the experimentally confirmed filtering result is relatively low, the correlation between the first weight and the uncertainty represented by the overall covariance may be set to be high, for example, the absolute value of the correlation coefficient between the two is large. The correlation coefficient between the first weight according with the actual situation and the uncertainty represented by the overall covariance can be set by a person skilled in the art, so that the first weight according with the actual situation is obtained, further, relatively accurate corrected inertial positioning data is obtained, and finally, the position information of the vehicle is obtained.

In the above technical solution, the inertial positioning data needs to be corrected according to the overall estimation value and the overall covariance to obtain more accurate corrected inertial positioning data, and in the correction process, the first weight of the overall estimation value and the second weight of the inertial positioning data form a matrix are used.

The manner in which the first weight and the second weight are obtained is exemplarily described below.

Optionally, determining the first weight of the overall estimate according to the overall covariance includes:

Through the technical scheme, the first weight of the overall estimated value and the second weight of the inertial positioning data can be obtained.

Further, determining the first weight of the overall estimate from the overall covariance may be further optimized to:

Correspondingly, determining the second weight of the inertial positioning data according to the covariance of the inertial positioning data forming matrix comprises:

Wherein w _IMU is a second weight, a _IMU is a second adjustment coefficient, which can be obtained through experiments, P _IMU is covariance of the inertial positioning data forming matrix, trace (P _IMU) is trace of P _IMU, M is the dimension of P _IMU, (P _IMU)_ii is the i-th diagonal element of P _IMU).

The a _IMM is used for adjusting the correlation between the first weight w _IMM and the reciprocal of trace (P (k|k)), and the a _IMU is used for adjusting the correlation between the second weight w _IMU and the reciprocal of trace (P _IMU). In the application process, a _IMM and a _IMU may be set to 1 first, then a _IMM and a _IMU are adjusted separately, after each adjustment, the calculated vehicle position information is compared with the actually detected vehicle position information, and when the error between the calculated vehicle position information and the actually detected vehicle position information is smaller or minimum, a _IMU and/or a _IMU at that time will be reserved. Wherein, the "error is small" means that the prediction error meets the accuracy requirement; the "minimum error" may be the extreme or the minimum of the error.

By adopting the technical scheme, the first weight and the second weight can be more adaptive to the actual demand.

In the above embodiment, the step of calculating the overall estimation value and the overall covariance of the filtering result of each preset motion constraint model from the inertial positioning data and the mixing probability is briefly described, and the step of calculating the overall estimation value and the overall covariance of the filtering result of each preset motion constraint model from the inertial positioning data and the mixing probability is exemplarily described below with reference to the filtering algorithm KALMAN FILTER.

Optionally, calculating an overall estimation value and an overall covariance of the filtering result of each preset motion constraint model according to the inertial positioning data and the mixing probability, including:

the initial conditions of each preset motion constraint model filtering algorithm are determined by the following modes:

p _ij is the transition probability from model i to model j, which is the transition probability matrix element of the prior Markov model; the probability of occurrence for the time k-1 model i.

The overall estimate and the overall covariance are determined by:

p_j(k|k-1)＝φ_j(k-1)p^j,0(k-1|k-1)φ_j(k-1)^T+B_jQ_jB_j ^T

K_j(k)＝p_j(k|k-1)H^T[Hp_j(k|k-1)H^T+R]^-1

p_j(k|k)＝[I-K_j(k)H(k)]p_j(k|k-1)

It should be understood that this step is illustrated herein by way of example only with reference to FIG. KALMAN FILTER. In a specific application, the skilled artisan can adapt the above formulas when using other algorithms such as EKF, UKF, PARTICLE FILTER, H-INFINITY FILTER, or Bayesian Filter.

Model probability updating is performed by the following likelihood functions:

the overall estimate and the overall covariance are:

By adopting the technical scheme, the overall estimation value and the overall covariance can be obtained. The overall covariance may be used to calculate a first weight to facilitate a weighted average or a weighted sum based on the overall estimate, the first weight, a matrix of inertial positioning data, and a second weight.

In the foregoing embodiment, the second weight is obtained from the covariance of the inertial positioning data constituent matrix, and the manner of obtaining the covariance of the inertial positioning data constituent matrix is exemplified below.

The covariance of the inertial positioning data constituent matrix may be calculated by updating the matrix by:

Wherein P _IMU(t_k) is the covariance of the matrix formed by the updated inertial positioning data, F _IMU is the state transition matrix, P _IMU(t_k-1) is the covariance of the matrix formed by the inertial positioning data known at the time k-1, obtained by the update formula of the last time, Transpose of the state transition matrix F _IMU, Q _IMU is the process noise covariance.

Wherein, I is an identity matrix, and Δt is an update period, which may also be referred to as a discrete duration;

Where σ _IMU ² is the variance of the process noise, used to represent the strength of the acceleration noise.

The state transition matrix F _IMU is used to describe the change of the system between two states, which can be made up of yaw rate information and acceleration information, for example, in case the inertial positioning data includes yaw rate information and acceleration information:

In the case where the inertial positioning data includes course angle information and acceleration information, the system may be composed of course angle information and acceleration information, for example:

In the case where the inertial positioning data includes yaw rate information, course angle information, and acceleration information, the system may be composed of yaw rate information, course angle information, and acceleration information, for example:

the determination of the position information of the vehicle based on the corrected inertial positioning data is further exemplarily described below.

In the case where the inertial positioning data includes yaw rate information and acceleration information, course angle information is first calculated by integrating the yaw rate information:

wherein, As the heading angle information of the time instant k,For yaw rate information detected at time k-1, ω _IMU(t_k) is yaw rate information detected at time k, Δt is an update period, which may also be referred to as a discrete duration.

In the case that the inertial positioning data includes course angle information and acceleration information, the course angle information at the time k can be directly detected and obtained

In the case where the inertial positioning data includes yaw rate information, course angle information, and acceleration information, the first course angle information may be calculated by integrating the yaw rate information, and then final course angle information may be determined based on the first course angle information and the detected second course angle information. For example, an average value of the first course angle information and the second course angle information may be regarded as final course angle information, or a weighted average value of the first course angle information and the second course angle information may be regarded as final course angle information.

And then converting the acceleration information detected by the IMU into the acceleration in the world coordinate system:

Wherein a _w(t_k) is acceleration information of the vehicle at time k in the world coordinate system, For heading angle information at time k, a _IMU(t_k) is acceleration information detected at time k,Course angle information for time kA rotation matrix is formed:

Updating the speed information:

v(t_k+1)＝v(t_k)+a_w(t_k)Δt

Wherein v (t _k+1) is updated speed information, v (t _k) is speed information known at the current time, the speed information is obtained by an update formula of the last time, a _w(t_k) is acceleration information of the vehicle at the current time k in the world coordinate system, and Δt is an update period, which may also be called as discrete duration.

Updating the position information:

p(t_k+1)＝p(t_k)+v(t_k+1)Δt

Where p (t _k+1) is updated position information, p (t _k) is position information known at the current time, obtained by an update formula at the previous time, position information p (0) =0 at the time of GNSS start failure, v (t _k+1) is updated speed information, and Δt is an update period, which may also be referred to as a discrete duration.

After the GNSS fails, the latest updated position information is the position information of the vehicle.

The inertial positioning data related to the above embodiment of calculating the vehicle position information are all corrected inertial positioning data, that is, the yaw rate information in the above embodiment is corrected yaw rate information, the heading angle information in the above embodiment is corrected heading angle information, and the acceleration information in the above embodiment is corrected acceleration information.

The above specific process of calculating the vehicle position information is only illustrative, and in a specific application scenario, a person skilled in the art may select a specific algorithm adapted to the actual situation according to his own experience, which is not described in detail herein.

Fig. 2 is a schematic diagram of a vehicle positioning device under GNSS short time failure based on driving behavior and inertial navigation according to an embodiment of the present application. The vehicle positioning device under the GNSS short-time failure based on driving behavior and inertial navigation can be realized in the form of software, hardware or a combination of the two.

Referring to fig. 2, the vehicle positioning device under GNSS short time failure based on driving behavior and inertial navigation includes a driving state prediction module 21, a motion model probability prediction module 22, a first filtering module 23, a second filtering module 24, and a correction module 25.

The driving state prediction module 21 is configured to obtain driving behavior information, and determine driving state information according to the driving behavior information; the driving behavior information includes: steering wheel angle, vehicle lateral speed, and yaw rate; the driving state information includes: straight, left lane change and right lane change;

The motion model probability prediction module 22 is configured to predict occurrence probability of each preset motion constraint model according to driving behavior information and driving state information; the preset motion constraint model comprises the following steps: a uniform linear motion model, a uniform acceleration linear motion model, a constant turning rate and speed motion model and a constant turning rate and acceleration model;

the first filtering module 23 is configured to determine a mixed probability matrix according to occurrence probabilities corresponding to all preset motion constraint models;

The second filtering module 24 is configured to obtain inertial positioning data, and calculate an overall estimation value and an overall covariance of the kalman filtering algorithm according to the inertial positioning data and the mixing probability; the inertial positioning data includes one or more of yaw rate information, heading angle information, and acceleration information;

The correction module 25 is configured to correct the inertial positioning data according to the overall estimation value and the overall covariance, and determine the position information of the vehicle according to the corrected inertial positioning data.

Optionally, the correction module 25 includes:

A first determining unit configured to determine a first weight of an overall estimation value according to an overall covariance; the first weight is inversely related to the uncertainty represented by the overall covariance;

A second determining unit for obtaining a second weight of the inertial positioning data determined according to covariance of the inertial positioning data forming matrix; the second weight is inversely related to uncertainty represented by covariance of the inertial positioning data formation matrix;

The first calculation unit is used for calculating a weighted average value or a weighted sum according to the overall estimation value, the first weight, the inertia positioning data forming matrix and the second weight;

and a third determining unit, configured to use the second weighted average or weighted sum as the corrected inertial positioning data.

Optionally, the first determining unit is specifically configured to execute the following formula:

Optionally, the second determining unit is specifically configured to execute the following formula:

The second filtering module 54 is specifically configured to:

the overall estimate and the overall covariance are determined by:

p_j(k|k-1)＝φ_j(k-1)p^j,0(k-1|k-1)φ_j(k-1)^T+B_jQ_jB_j ^T

K_j(k)＝p_j(k|k-1)H^T[Hp_j(k|k-1)H^T+R]^-1

p_j(k|k)＝[I-K_j(k)H(k)]p_j(k|k-1)

wherein, Predicting a state of the model j for the moment k; phi _j (k-1) is a state transition matrix at the time k-1, which is used for describing the state change relation of the system from the time k-1 to the time k; p _j (k|k-1) is the covariance prediction value of time k on model j; b _j is a control input matrix for representing the influence of noise on the system state; q _j is a process covariance matrix for representing the magnitude of system process noise; k _j (K) is a Kalman gain matrix of the model j for the moment K; h is an observation matrix; r is an observed noise covariance matrix; updating a matrix for the state of the model j at the moment k; z (k) is an observation vector matrix at the moment k and is composed of inertial positioning data; Is the observation residual; p _j (k|k) is the covariance update matrix of time k to model j; i is an identity matrix;

model probability updating is performed by the following likelihood functions:

wherein, lambada _j (k) is the likelihood function of the time k model j; is a transpose of the measurement method of model j, Is the residual covariance of model j;

The model probability is updated as follows:

the overall estimate and the overall covariance are:

Optionally, the driving state prediction module 21 includes:

A first obtaining unit for obtaining driving behavior information of a plurality of continuous moments before a current moment to form an observation sequence of the driving behavior information;

and a fourth determining unit for determining driving state information according to the observation sequence of driving behavior information.

Optionally, the motion model probability prediction module 22 includes:

A second obtaining unit, configured to obtain a plurality of sequences of driving behavior information and driving state information according to a sequence of the time stamps;

And a fifth determining unit, configured to input a plurality of sequences of driving behavior information and driving state information into the artificial neural network model with sequence processing capability, and obtain occurrence probabilities of a plurality of preset motion constraint models output by the artificial neural network model with sequence processing capability.

Optionally, the first filtering module 23 includes:

the sixth determining unit is used for determining a transition probability matrix of the prior Markov model according to the occurrence probabilities corresponding to all the preset motion constraint models;

and a seventh determining unit, configured to determine a mixing probability matrix according to the transition probability.

Fig. 3 is a schematic flow chart of a specific implementation of a vehicle positioning method under GNSS short time failure based on driving behavior and inertial navigation according to an embodiment of the present application.

Referring to fig. 3, the method for positioning the vehicle under the GNSS short time failure based on driving behavior and inertial navigation includes the following 5 steps sequentially performed: the method comprises a driving behavior information acquisition step, a driving behavior state identification step, a motion constraint model prediction step, a key motion parameter estimation step and an auxiliary positioning step.

In the driving behavior information obtaining step, a steering wheel angle, a vehicle lateral speed and a vehicle yaw rate are obtained through an on-vehicle sensor, and the steering wheel angle, the vehicle lateral speed and the yaw rate are formed into a time sequence { O }, wherein O is an observation sequence of driving behavior information, and the driving behavior information comprises the steering wheel angle, the vehicle lateral speed and the vehicle yaw rate.

In the driving state identification step, firstly training an HMM by using training data acquired by a real vehicle test, then inputting a time sequence { O } obtained in the driving behavior information acquisition step into the trained HMM, and identifying the driving state information on line by the HMM, wherein the output result of the HMM is that the vehicle moves straight, the vehicle changes lanes left or the vehicle changes lanes right.

In the motion constraint model prediction, LSTM is firstly trained offline by using data acquired by a real vehicle test, then a time sequence { O } in a driving behavior information acquisition step and an output result of the HMM in a driving state identification step are input into the LSTM, and the LSTM outputs a uniform linear motion model probability, a uniform acceleration linear motion model probability, a constant turning rate and a speed motion model probability, and a constant turning rate and an acceleration model probability.

In the key motion parameter estimation step, four of the LSTM outputs: the constant velocity linear motion model probability, the uniform acceleration linear motion model probability, the constant turning rate and the velocity motion model probability, and the constant turning rate and the acceleration model probability form a constraint model of the vehicle navigation system, and an overall estimated value is obtained through improved interactive multimode by combining yaw angular velocity information obtained by the IMU, course angular information obtained by the magnetic sensor and acceleration information obtained by the accelerometer, wherein the overall estimated value comprises: yaw rate optimal estimation value, course angle optimal estimation value and acceleration optimal estimation value.

In the auxiliary positioning step, the overall estimated value obtained in the key motion parameter estimation step is used for carrying out auxiliary correction on the vehicle positioning under the condition that the vehicle GNSS fails in a short time by taking the pose of the current moment obtained by the IMU as an origin.

Fig. 4 is a schematic diagram of a vehicle positioning device under GNSS short time failure based on driving behavior and inertial navigation according to an embodiment of the present application. Referring to fig. 4, a vehicle positioning device 40 under GNSS short time failure based on driving behavior and inertial navigation includes:

A processor (processor) 41 and a memory (memory) 42, and may also include a communication interface (Communication Interface) 43 and a bus 44. The processor 41, the communication interface 43 and the memory 42 may communicate with each other via a bus 44. The communication interface 43 may be used for information transmission. The processor 41 may invoke logic instructions in the memory 42 to perform the vehicle localization method under GNSS short time failure based on driving behavior and inertial navigation provided by the previous embodiments.

Further, the logic instructions in the memory 42 described above may be implemented in the form of software functional units and stored in a computer readable storage medium when sold or used as a stand alone product.

The memory 42 is a computer readable storage medium that can be used to store a software program, a computer executable program, such as program instructions/modules corresponding to the methods in the embodiments of the present application. The processor 41 executes functional applications and data processing by running software programs, instructions and modules stored in the memory 42, i.e. implements the methods of the method embodiments described above.

Memory 42 may include a storage program area that may store an operating system, at least one application program required for functionality, and a storage data area; the storage data area may store data created according to the use of the terminal device, etc. In addition, memory 42 may include high-speed random access memory, and may also include non-volatile memory.

The embodiment of the application provides a computer readable storage medium, which stores computer executable instructions, wherein the computer executable instructions are configured to:

Predicting the occurrence probability of each preset motion constraint model according to the driving behavior information and the driving state information; the preset motion constraint model comprises the following steps: a uniform linear motion model, a uniform acceleration linear motion model, a constant turning rate and speed motion model and a constant turning rate and acceleration model;

Determining a mixed probability matrix according to occurrence probabilities corresponding to all preset motion constraint models;

acquiring inertial positioning data, and calculating the overall estimation value and the overall covariance of the filtering result of each preset motion constraint model according to the inertial positioning data and the mixing probability; the inertial positioning data includes one or more of yaw rate information, heading angle information, and acceleration information;

The computer readable storage medium described above may be a transitory computer readable storage medium.

The technical solution of the embodiment of the present application may be embodied in the form of a software product, where the software product is stored in a storage medium, and includes one or more instructions to cause a computer device (which may be a personal computer, a server, or a network device) to perform all or part of the steps of the method in the embodiment of the present application. And the aforementioned storage medium may be a non-transitory storage medium including: a plurality of media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, or a transitory storage medium.

The above description and the drawings illustrate embodiments of the application sufficiently to enable those skilled in the art to practice them. Other embodiments may involve structural, logical, electrical, process, and other changes. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the sequence of operations may vary. Portions and features of some embodiments may be included in, or substituted for, those of others. Moreover, the terminology used in the present application is for the purpose of describing embodiments only and is not intended to limit the claims. As used in the description of the embodiments and the claims, the singular forms "a," "an," and "the" (the) are intended to include the plural forms as well, unless the context clearly indicates otherwise. Furthermore, when used in the present disclosure, the terms "comprises," "comprising," and/or variations thereof, mean that the recited features, integers, steps, operations, elements, and/or components are present, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method or apparatus that includes such elements. In this context, each embodiment may be described with emphasis on the differences from the other embodiments, and the same similar parts between the various embodiments may be referred to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method sections disclosed in the embodiments, the description of the method sections may be referred to for relevance.

Those of skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. The skilled person may use different methods for each particular application to achieve the described functionality, but such implementation is not to be considered as beyond the scope of the embodiments of the application. It will be clearly understood by those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not described herein again.

In the embodiments disclosed herein, the disclosed methods, articles of manufacture (including but not limited to devices, apparatuses, etc.) may be practiced in other ways. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of elements may be merely a logical functional division, and there may be additional divisions when actually implemented, e.g., multiple elements or components may be combined or integrated into another system, or some features may be omitted or not performed. In addition, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interface, device or unit indirect coupling or communication connection, which may be in electrical, mechanical or other form. The elements illustrated as separate elements may or may not be physically separate, and elements shown as elements may or may not be physically located, or may be distributed over a plurality of network elements. Some or all of the units may be selected according to actual needs to implement the present embodiment. In addition, each functional unit in the embodiment of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit.

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. Each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

Claims

1. A vehicle positioning method under GNSS short time failure based on driving behavior and inertial navigation, comprising:

2. The vehicle positioning method according to claim 1, characterized in that correcting the inertial positioning data according to the overall estimation value and the overall covariance includes:

3. The vehicle positioning method according to claim 2, characterized in that,

Determining a first weight of the overall estimate from the overall covariance, comprising:

Determining a second weight of the inertial positioning data from the covariance of the inertial positioning data formation matrix, comprising:

Where w _IMU is the second weight, PI _MU is the covariance of the inertial positioning data constituent matrix, trace (P _IMU) is the trace of P _IMU, M is the dimension of P _IMU, (P _IMU)_ii is the i-th diagonal element of P _IMU).

4. The vehicle positioning method according to claim 1, characterized in that calculating an overall estimation value and an overall covariance of a filtering result of each of the preset motion constraint models from the inertial positioning data and the mixing probability, comprising:

The overall estimate and the overall covariance are determined by:

wherein, Predicting a state of the model j for the moment k; phi _j (k-1) is a state transition matrix at the time k-1, which is used for describing the state change relation of the system from the time k-1 to the time k; Predicting a state of the model j for the moment k-1; p _j (k|k-1) is the covariance prediction value of time k on model j; p ^j,0 (k-1|k-1) is the covariance predictor of time k-1 for model j; b _j is a control input matrix for representing the influence of noise on the system state; b _j ^T is the transpose of the control input matrix B _j; q _j is a process covariance matrix for representing the magnitude of system process noise; k _j (K) is a Kalman gain matrix of the model j for the moment K; h is an observation matrix; r is an observed noise covariance matrix; Updating a matrix for the state of the model j at the moment k; z (k) is an observation vector matrix of time k, and is composed of the inertial positioning data; is the observation residual; pj (k|k) is the covariance update matrix of time k to model j; i is an identity matrix;

model probability updating is performed by the following likelihood functions:

Wherein Λ _j (k) is a likelihood function of the time k model j; is a transpose of the measurement method of model j, Is the residual covariance of model j;

The model probability is updated as follows:

the overall estimate and the overall covariance are:

5. The vehicle positioning method according to any one of claims 1 to 4, characterized in that determining driving state information from the driving behavior information includes:

obtaining driving behavior information of a plurality of continuous moments before the current moment to form an observation sequence of the driving behavior information;

and determining driving state information according to the observation sequence of the driving behavior information.

6. The vehicle positioning method according to any one of claims 1 to 4, characterized in that predicting occurrence probability of each preset motion constraint model based on the driving behavior information and the driving state information includes:

7. The vehicle positioning method according to any one of claims 1 to 4, characterized in that determining a hybrid probability matrix according to occurrence probabilities corresponding to all of the preset motion constraint models includes:

8. A vehicle positioning device under GNSS short time failure based on driving behavior and inertial navigation, comprising:

The second filtering module is used for obtaining inertial positioning data and calculating the overall estimation value and the overall covariance of the filtering result of each preset motion constraint model according to the inertial positioning data and the mixing probability; the inertial positioning data includes one or more of yaw rate information, course angle information, and acceleration information;

And the correction module is used for correcting the inertial positioning data according to the overall estimated value and the overall covariance, and determining the position information of the vehicle according to the corrected inertial positioning data.

9. A vehicle positioning device under GNSS short time failure based on driving behaviour and inertial navigation, comprising a processor and a memory storing program instructions, characterized in that the processor is configured to execute the vehicle positioning method under GNSS short time failure based on driving behaviour and inertial navigation as claimed in any of the claims 1 to 7 when executing the program instructions.

10. A vehicle positioning system, comprising:

The vehicle positioning device under GNSS short time failure based on driving behavior and inertial navigation according to claim 8 or 9.