CN113376675B - Urban canyon positioning method based on GNSS/vision/Lidar fusion - Google Patents

Urban canyon positioning method based on GNSS/vision/Lidar fusion Download PDF

Info

Publication number
CN113376675B
CN113376675B CN202110571260.4A CN202110571260A CN113376675B CN 113376675 B CN113376675 B CN 113376675B CN 202110571260 A CN202110571260 A CN 202110571260A CN 113376675 B CN113376675 B CN 113376675B
Authority
CN
China
Prior art keywords
satellite
candidate
positioning
carrier
street
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110571260.4A
Other languages
Chinese (zh)
Other versions
CN113376675A (en
Inventor
孙蕊
刘正午
戴晔莹
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Aeronautics and Astronautics
Original Assignee
Nanjing University of Aeronautics and Astronautics
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University of Aeronautics and Astronautics filed Critical Nanjing University of Aeronautics and Astronautics
Priority to CN202110571260.4A priority Critical patent/CN113376675B/en
Publication of CN113376675A publication Critical patent/CN113376675A/en
Application granted granted Critical
Publication of CN113376675B publication Critical patent/CN113376675B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
    • G01S19/38—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system
    • G01S19/39—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system the satellite radio beacon positioning system transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
    • G01S19/42—Determining position
    • G01S19/45—Determining position by combining measurements of signals from the satellite radio beacon positioning system with a supplementary measurement

Abstract

The invention discloses an urban canyon positioning method based on GNSS/vision/Lidar fusion, which comprises the following steps of 1, judging satellite visibility by using a fisheye camera; step 2, further judging the satellite signal type based on a supervised learning mode; step 3, determining the weight of the satellite and carrying out multi-constellation positioning calculation based on the weight of the satellite to obtain an initial positioning solution; step 4, correcting the initial positioning solution and generating candidate positions; step 5, screening candidate positions and calculating candidate position weights according to satellite visibility of the candidate positions and the coincidence degree of the identified visible satellites in the step 1; and 6, calculating based on the candidate position weights to obtain a final positioning result. The method effectively improves the accuracy of the street crossing direction and the street following direction in the GNSS positioning of the urban canyon, and has better real-time performance and convenience.

Description

Urban canyon positioning method based on GNSS/vision/Lidar fusion
Technical Field
The invention belongs to the technical field of satellite positioning navigation, and particularly relates to an urban canyon positioning method based on GNSS/vision/Lidar fusion.
Background
The global navigation satellite system GNSS (Global Navigation Satellite System) can provide space-time information for the carrier, and has become an important means for supporting urban intelligent transportation and other applications. Although GNSS can obtain positioning information with high accuracy in an open area, in an urban canyon environment, due to the satellite being blocked by a high building, multipath (Multipath) and non-line of sight (NLOS, non Line Of Sight) reception are generated, and the NLOS and Multipath signals are used for GNSS positioning calculation to reduce positioning accuracy, so many scholars propose a method of excluding reflected signals from GNSS positioning calculation, thereby reducing the influence of the reflected signals on GNSS positioning, and improving the urban canyon GNSS positioning accuracy, such as: peyraud uses a 3D city model to identify NLOS signals,NLOS signals are excluded from GNSS positioning solutions. But this approach makes it difficult to identify all NLOS satellites. Meguro uses an omnidirectional infrared camera to identify invisible satellites and uses only visible satellites for GNSS positioning resolution. However, this approach treats the visible satellites as LOS signals, ignoring that the visible satellites may be reflected and diffracted to become Multipath signals, and the use of such signals for GNSS positioning solutions may reduce the accuracy of the positioning. Sanroma Sanchez proposed an NLOS detection method that utilized information obtained with a fisheye camera and C/N 0 To distinguish whether the satellite is an LOS or NLOS signal. However, NLOS signals are sometimes subject to strong reflections resulting in C/N 0 Greater than the threshold, therefore, only add C/N 0 The use of a signal class for distinguishing between signal classes is not yet reliable enough.
On the other hand, although in urban environments, excluding reflected satellite signals improves the accuracy of urban GNSS positioning, in urban environments, because buildings on both sides of a street block, the probability that satellite signals in the street direction are blocked by the buildings is much higher than the probability that satellite signals in the street direction are blocked by the buildings, and therefore, after excluding reflected satellite signals, satellites for GNSS positioning calculation are mostly in the street direction, resulting in positioning accuracy in the street direction being much better than positioning accuracy in the street direction. Groves, therefore, proposes a shadow matching method that predicts which satellites can be seen from different locations using a 3D city model and compares it to the measured satellite visibility to determine location, thereby improving accuracy in the cross-street direction. Although many scholars have improved on the method of shadow matching. Such as: shadow matching method utilizes C/N 0 The threshold is set to classify NLOS/LOS, but in urban environments, signals are reflected and refracted by buildings, and C/N occurs after some NLOS signals are strongly reflected 0 Exceeding the set threshold, only C/N is used 0 Distinguishing NLOS/LOS signals is unreliable, so Xu classifies signals using SVM (Support Vector Machine ), improving accuracy of NLOS/LOS classification, thereby improving location of shadow matchingPrecision. However, shadow matching requires that the 3D city model have a high accuracy and be updated in time, which is often difficult to achieve. Therefore, how to improve the positioning accuracy of GNSS in urban canyon environments is an urgent problem to be solved.
Disclosure of Invention
The invention aims to: aiming at the defects of the prior art, the invention provides an urban canyon positioning method based on GNSS/vision/Lidar fusion.
In order to solve the technical problems, the invention discloses an urban canyon positioning method based on GNSS/vision/Lidar fusion, which comprises the following steps:
step 1, judging satellite visibility by using a fisheye camera;
step 2, further judging the signal type of the visible satellite based on a supervised learning mode;
step 3, determining the weight of the satellite and carrying out multi-constellation positioning calculation based on the weight of the satellite to obtain an initial positioning solution;
step 4, correcting the initial positioning solution and generating candidate positions;
step 5, screening candidate positions and calculating candidate position weights according to the degree of coincidence between the satellite visibility of the candidate positions and the satellite visibility in the step 1;
and 6, calculating based on the candidate position weights to obtain a final positioning result.
In one implementation, the step 1 includes:
step 1-1, shooting a sky image by using a fisheye camera mounted on a carrier to point to the sky;
step 1-2, converting a sky image shot by a fisheye camera into a binary image, and distinguishing an image area of the sky from an image area shielded by a building;
step 1-3, projecting satellites into the sky image plane, wherein the projection formula is as follows:
wherein E is i The elevation angle of the ith satellite, i is a positive integer which is less than or equal to the sum of the total number of satellites of four constellations, alpha i Azimuth for the ith satellite, f c Focal length d of fish eye camera i For the distance between the position of the ith satellite projected to the image plane and the center of the image plane, beta is the included angle between the positive direction of the X axis of the local navigation coordinate system and the north direction,and->Is the coordinates of the center of the image plane,/>And->Coordinates projected into the sky image for the ith satellite; the four constellations comprise a global positioning system GPS, a Beidou satellite navigation system BDS, a Glonass global satellite navigation system GLONASS and a Galileo satellite navigation system GALILEO.
And 1-4, judging the visibility of the satellite, if the coordinates of the satellite projected into the sky image are in the sky area, judging the satellite as a visible satellite, otherwise, judging the satellite as an invisible satellite.
In urban canyon GNSS positioning, satellite signals are blocked and reflected by buildings, the type of the satellite signals is accurately identified, the urban canyon GNSS positioning accuracy is improved, the visibility of satellites is judged by shooting sky images by using a fisheye camera, invisible satellite signals are regarded as NLOS signals, the classification accuracy of the NLOS signals is improved, and meanwhile, the judgment range is narrowed for further judging that the visible satellite signals are LOS signals or Multipath signals.
In one implementation, the supervised learning method in step 2 uses a linear support vector machine SVM classifier.
The fish eye camera can only judge whether the satellite is a visible satellite or an invisible satellite, the invisible satellite can judge as an NLOS signal, but the satellite can be reflected and diffracted by a building, so that the visible satellite cannot judge as an LOS signal or a Multipath signal, the invention classifies the LOS signal and the Multipath signal of the visible satellite by using a supervised learning mode, and the SVM has good effect on satellite signal classification, so that the invention selects the SVM to classify the LOS signal and the Multipath signal of the visible satellite.
In one implementation, the step 2 includes:
step 2-1, in an off-line stage, LOS signals and Multipath signals are collected and marked in an urban environment;
step 2-2, training a linear SVM classifier by using the marked LOS signals and the Multipath signals, wherein a classification scoring formula of the linear SVM classifier is as follows:
u i ＝(v i /s) T α+γ
wherein u is i SVM score for ith satellite, v i The feature vector of the SVM of the ith satellite comprises the carrier-to-noise ratio C/N of the satellite signal 0 The pseudo-range residual error and the pseudo-range change rate, s is a kernel scale, alpha is a vector for fitting a linear coefficient, and gamma is the deviation of a linear SVM classifier;
step 2-3, in the online stage, inputting the feature vector of the satellite signal judged to be the visible satellite in the step 1 into an SVM scoring formula to obtain a predicted satellite signal type, when u i Predicting the ith satellite signal as LOS signal when the signal is not less than 0, and when u i <And predicting the ith satellite signal as a Multipath signal at 0.
In one implementation, the step 3 includes:
step 3-1, the satellite signal type obtained according to step 2-3 isCarrier-to-noise ratio C/N of satellite signal of LOS signal 0 And determining satellite weights by satellite elevation angles, wherein the total number of satellites with the satellite signal types of LOS signals obtained in the step 2-3 is J, and the satellite weights sw of the J-th satellite are recorded j The formula is:
wherein J is more than or equal to 1 and less than or equal to J, sigma j Is the standard deviation of error, S j In order for the scaling factor to be a factor,C/N 0j representing the carrier-to-noise ratio of the j-th satellite, E j Representing satellite elevation angle, a, of the j-th satellite 0 、a 1 And E is 0 Respectively constant terms, a 0 ＝3mm，a 1 ＝26mm，E 0 ＝20°；
Step 3-2, obtaining an initial position based on multi-constellation positioning calculation of a weighted least square method, firstly unifying a time system and a coordinate system of a global positioning system GPS (Global Positioning System), a Beidou satellite navigation system BDS (BeiDou Navigation Satellite System), a Gelnas global satellite navigation system GLONASS (GLObal NAvigation Satellite System) and a Galileo satellite navigation system GALILEO (Galileo satellite navigation system), unifying four systems under a space-time system of a GPS, wherein the four systems are provided with n BDS satellites, m GPS satellites, k GLONASS satellites and l GALILEO satellites, and the initial position of a carrier is set as n+m+k+l=JThe position of the navigation satellite is (x j ,y j ,z j ) Considering the satellite weights assigned in step 3-1, the observation equation may be listed as follows:
WAX＝WB
where W is the weight matrix of the satellite, w=diag (sw 1 ,sw 2 ,...,sw J ) A is the satellite direction residual rotation matrix,
the geometric distance from the j satellite to the initial position is the geometric distance from the j satellite to the initial position; x= [ Δx, Δy, Δz, cΔt ] BDS ,cΔt GPS ,cΔt GLS ,cΔt GAO ] T ，cΔt BDS ,cΔt GPS ,cΔt GLs ,cΔt GAo Equivalent distance errors from the carrier position to the BDS satellite, GPS satellite, GALILEO satellite and GLONASS satellite positions respectively caused by receiver clock deviation and the like, and B is a pseudo-range observation value residual vector +.>
Wherein ρ is j A pseudorange representing the j-th satellite of the actual measurement;
calculating initial value compensation quantity of the corresponding position by using a weighted least square method, and sequentially iterating until the compensation quantity is smaller than an allowable error range threshold value, thereby realizing carrier positioning calculation and obtaining:
X＝(A T W T WA) -1 A T W T WB
the initial positions of the carrier were thus obtained:
in one implementation, the step 4 includes:
step 4-1, measuring the distance from the carrier to the street building by using a pulse laser radar ranging technology, wherein the formula is as follows:
where c= 2.99792458 ×10 8 m/s is the propagation speed of light in vacuum, T is the transmission time of laser in the atmosphere, and d is the distance from the carrier to the building;
step 4-2, correcting the error of the initial position in the street crossing direction based on the 3D city model: if the distance d from the carrier to the left street is obtained, an error formula for correcting the street crossing direction of the initial position is as follows:
X repair tool ＝(x Repair tool ,y Repair tool ,z Repair tool ) T ＝(x Initially, the method comprises +(l-d)sinα,y Initially, the method comprises +(l-d)cosα,z Initially, the method comprises ) T
Wherein l is X Initially, the method comprises Distance to street, α is the angle between the street direction and the east direction.
Similarly, if the distance d from the carrier to the right street is obtained, an error formula for correcting the street crossing direction of the initial position is as follows:
X repair tool ＝(x Repair tool ,y Repair tool ,z Repair tool ) T ＝(x Initially, the method comprises -(l-d)sinα,y Initially, the method comprises -(l-d)cosα,z Initially, the method comprises ) T
If the distance from the carrier to the streets on both sides is obtained at the same time, one of them can be optionally corrected.
Step 4-3, generating according to the L meter interval delta L meter before and after the corrected position along the street directionCandidate position, the coordinates of the e-th generated candidate position are X e ，/>We call the corrected position the original candidate position.
In urban environments, due to the blockage of buildings on both sides of the street, the accuracy in the cross-street direction is lower than that along the street after the reflected satellites are eliminated. Therefore, groves proposes a shadow matching method to improve accuracy in the cross street direction, and many scholars have improved the shadow matching method. However, shadow matching requires that the 3D city model have a high accuracy and be updated in time, which is often difficult to achieve. According to the method, the distance from the carrier to the street is measured by using Lidar (Light detection and ranging, laser radar), the initial positioning result is corrected by using the distance, the positioning precision in the street crossing direction is improved, the requirement on the precision of the 3D city model is low, and the method has better instantaneity and convenience.
In one implementation, the step 5 includes:
and 5-1, predicting satellite visibility of each candidate position by using a 3D city model, namely calculating the lowest elevation angle of the satellite which is not blocked by a building under different azimuth angles by using the 3D city model, when the elevation angle of the satellite under the azimuth angles is larger than the lowest elevation angle, considering the satellite as visible, otherwise, considering the satellite as invisible satellite, comparing the satellite visibility predicted by using the 3D city model with the satellite visibility judged by using a fisheye camera at an e candidate point, scoring the e candidate point, and if the visibility of the i satellite predicted by using the 3D city model at the e candidate point is consistent with the i satellite judged at the e candidate point by using the fisheye camera, adding a score to the e candidate point, otherwise, scoring the e candidate point unchanged. Marking the e candidate position as Mce;
step 5-2, excluding the e candidate position if the score Mce of the e candidate position is smaller than the score of the original candidate position; if the score Mce of the e candidate position is greater than or equal to the score of the original candidate position, calculating the weight of the candidate position according to the ranking from large to small of the scores of the candidate positionsThe method comprises the steps of carrying out a first treatment on the surface of the The scores Mce of the N candidate positions in total are recorded to be greater than or equal to the scores of the original candidate positions,
eliminating candidate locations with scores lower than the original candidate location is beneficial to reducing interference of candidate locations with significant errors to final positioning accuracy.
Step 5, firstly, scoring the candidate position by utilizing the coincidence degree of satellite visibility of the 3D city model predicted candidate position and satellite visibility of the real position judged by the fish-eye camera, then determining the weight of the candidate position according to the score of the candidate position, wherein the more the satellite visibility of the candidate position is coincident with the satellite visibility of the real position, the more the candidate position is considered to be close to the real position, so that the candidate position is given a larger weight, and then weighting calculation is carried out to obtain a final positioning result, thereby improving the positioning precision along the street direction.
In one implementation, the candidate location weight formula calculated in step 5-2 is as follows:
wherein pw is q Represents the q candidate position weight, wherein q is more than or equal to 1 and less than or equal to N, N q Ranking from big to small showing scores according to the q-th candidate position, 1.ltoreq.n q ≤N。
The formula weights the ranking of the candidate locations according to the score of the candidate locations, and considers that the higher the score, the closer the candidate locations are to the real locations, and the greater the weight should be assigned. The weighted positioning calculation is beneficial to improving the positioning accuracy along the street direction according to the weight of the candidate position.
In one implementation, the formula for obtaining the final positioning result X based on the candidate position weight calculation in step 6 is as follows:
wherein X is q Representing the coordinates of the q-th candidate position.
The beneficial effects are that:
1. according to the method, the fisheye camera is judged to be a satellite of the visible satellite, LOS/Multipath classification is carried out by utilizing the SVM, so that the classification precision of LOS signals is improved, GNSS positioning calculation is carried out only by using the LOS signals, and the positioning precision is improved;
2. according to the method, the distance from the carrier to the two sides of the street is generated by using the laser radar, and the 3D city model is used for correcting the positioning error of the street, so that the precision of the street crossing direction is effectively improved, the requirement on the precision of the 3D city model is low, and the method has better instantaneity and convenience;
3. according to the method, the device and the system, the multiple candidate positions are further selected, satellite visibility of the multiple candidate positions is compared and analyzed with satellite visibility judged by the fisheye camera, the coincidence degree is judged, corresponding weights are given according to the coincidence degree, weighted positioning calculation is carried out according to the weights, and the accuracy of the urban canyon GNSS positioning in the street direction is effectively improved.
Drawings
The foregoing and/or other advantages of the invention will become more apparent from the following detailed description of the invention when taken in conjunction with the accompanying drawings and detailed description.
Fig. 1 is a flowchart of an urban canyon positioning method based on GNSS/vision/Lidar fusion according to an embodiment of the present application;
fig. 2 is a schematic view of projection of a satellite to a sky image plane according to an embodiment of the present disclosure;
fig. 3 is a schematic diagram of correcting an error of an initial position in a street crossing direction based on a 3D city model according to an embodiment of the present application.
Detailed Description
Embodiments of the present invention will be described below with reference to the accompanying drawings.
Fig. 1 is a flowchart of an urban canyon positioning method based on GNSS/vision/Lidar fusion according to an embodiment of the present application, including the following steps:
step 1, judging satellite visibility by using a fisheye camera;
step 2, further judging the signal type of the visible satellite based on a supervised learning mode;
step 3, determining the weight of the satellite and carrying out multi-constellation positioning calculation based on the weight of the satellite to obtain an initial positioning solution;
step 4, correcting the initial positioning solution and generating candidate positions;
step 5, screening candidate positions and calculating candidate position weights according to the degree of coincidence between the satellite visibility of the candidate positions and the satellite visibility in the step 1;
and 6, calculating based on the candidate position weights to obtain a final positioning result.
In this embodiment, the step 1 includes:
step 1-1, shooting a sky image by using a fisheye camera mounted on a carrier to point to the sky;
step 1-2, converting a sky image shot by a fisheye camera into a binary image, and distinguishing an image area of the sky from an image area shielded by a building;
step 1-3, projecting a satellite into the sky image plane, wherein a projection schematic diagram is shown in fig. 2, and a projection formula is as follows:
wherein E is i The elevation angle of the ith satellite, i is a positive integer which is less than or equal to the sum of the total number of satellites of four constellations, alpha i Azimuth for the ith satellite, f c Focal length d of fish eye camera i For the distance between the position of the ith satellite projected to the image plane and the center of the image plane, beta is the included angle between the positive direction of the X axis of the local navigation coordinate system and the north direction,and->Is the coordinates of the center of the image plane,/>And->Coordinates projected into the sky image for the ith satellite; the four constellations comprise a global positioning system GPS, a Beidou satellite navigation system BDS, a Glonass global satellite navigation system GLONASS and a Galileo satellite navigation system GALILEO.
And 1-4, judging the visibility of the satellite, if the coordinates of the satellite projected into the sky image are in the sky area, judging the satellite as a visible satellite, otherwise, judging the satellite as an invisible satellite.
In this embodiment, the supervised learning method in step 2 uses a linear support vector machine SVM classifier. The step 2 comprises the following steps: step 2-1, in an off-line stage, LOS signals and Multipath signals are collected and marked in an urban environment;
step 2-2, training a linear SVM classifier by using the marked LOS signals and the Multipath signals, wherein a classification scoring formula of the linear SVM classifier is as follows:
u i ＝(v i /s) T α+γ
wherein u is i Scoring of the ith satellite, v i The feature vector of the SVM of the ith satellite comprises the carrier-to-noise ratio C/N of the satellite signal 0 The pseudo-range residual error and the pseudo-range change rate, s is a kernel scale, alpha is a vector for fitting a linear coefficient, and gamma is the deviation of a linear SVM classifier;
step 2-3, online stage, willIn the step 1, the feature vector of the satellite signal of the visible satellite is judged to be input into an SVM scoring formula, the predicted satellite signal type is obtained, and when u i Predicting the ith satellite signal as LOS signal when the signal is not less than 0, and when u i <And predicting the ith satellite signal as a Multipath signal at 0.
In this embodiment, the step 3 includes:
step 3-1, the carrier-to-noise ratio C/N of the satellite signal with the type of LOS signal obtained in step 2-3 0 And determining satellite weights by satellite elevation angles, wherein the total number of satellites with the satellite signal types of LOS signals obtained in the step 2-3 is J, and the satellite weights sw of the J-th satellite are recorded j The formula is:
wherein J is more than or equal to 1 and less than or equal to J, sigma j Is the standard deviation of error, S j In order for the scaling factor to be a factor,C/N0j represents the carrier-to-noise ratio of the jth satellite, E j Representing satellite elevation angle, a, of the j-th satellite 0 、a 1 And E is 0 Respectively constant terms, a 0 ＝3mm，a 1 ＝26mm，E 0 ＝20°；
Step 3-2, obtaining an initial position based on multi-constellation positioning calculation of a weighted least square method, firstly unifying a time system and a coordinate system of a global positioning system GPS, a Beidou satellite navigation system BDS, a Geronnass global satellite navigation system GLONASS and a Galileo satellite navigation system GALILEO, unifying four systems under a space-time system of the GPS, wherein n BDS satellites, m GPS satellites, k GLONASS satellites and l GALILEO satellites are arranged, n+m+k+l=J, and setting the initial position of a carrier as the initial position of the carrierThe position of the navigation satellite is (x j ,y j ,z j ) Considering the satellite weights assigned in step 3-1, the observation equation may be listed as follows:
WAX＝WB
where W is the weight matrix of the satellite, w=diag (sw 1 ,sw 2 ,...,sw J ) A is the satellite direction residual rotation matrix,
the geometric distance from the j satellite to the initial position is the geometric distance from the j satellite to the initial position; x= [ Δx, Δy, Δz, cΔt ] BDS ,cΔt GPS ,cΔt GLS ,cΔt GAO ] T ，cΔt BDS ,cΔt GPS ,cΔt GLs ,cΔt GAo Equivalent distance errors from the carrier position to the BDS satellite, GPS satellite, GALILEO satellite and GLONASS satellite positions respectively caused by receiver clock deviation and the like, and B is a pseudo-range observation value residual vector +.>
Wherein ρ is j A pseudorange representing the j-th satellite of the actual measurement;
calculating initial value compensation quantity of the corresponding position by using a weighted least square method, and sequentially iterating until the compensation quantity is smaller than an allowable error range threshold value, thereby realizing carrier positioning calculation and obtaining:
X＝(A T W T WA) -1 A T W T WB
the initial positions of the carrier were thus obtained: in this embodiment, the error range threshold is 10 -5 。
In this embodiment, the step 4 includes:
step 4-1, measuring the distance from the carrier to the street building by using a pulse laser radar ranging technology, wherein the formula is as follows:
where c= 2.99792458 ×10 8 m/s is the propagation speed of light in vacuum, T is the transmission time of laser in the atmosphere, and d is the distance from the carrier to the building;
step 4-2, correcting the error of the initial position in the street crossing direction based on the 3D city model: as shown in fig. 3, if the distance d from the carrier to the left street is obtained, the error formula for correcting the initial position across street direction is:
X repair tool ＝(x Repair tool ,y Repair tool ,z Repair tool ) T ＝(x Initially, the method comprises +(l-d)sinα,y Initially, the method comprises +(l-d)cosα,z Initially, the method comprises ) T
Wherein l is X Initially, the method comprises Distance to street, α is the angle between the street direction and the east direction.
Similarly, if the distance d from the carrier to the right street is obtained, an error formula for correcting the street crossing direction of the initial position is as follows:
X repair tool ＝(x Repair tool ,y Repair tool ,z Repair tool ) T ＝(x Initially, the method comprises -(l-d)sinα,y Initially, the method comprises -(l-d)cosα,z Initially, the method comprises ) T
If the distance from the carrier to the streets on both sides is obtained at the same time, one of them can be optionally corrected.
Step 4-3, generating according to the L meter interval delta L meter before and after the corrected position along the street directionCandidate position, the coordinates of the e-th generated candidate position are X e ，/>We call the corrected position the original candidate position. In this embodiment, L is 5, and Δl is 1.
In this embodiment, the step 5 includes:
and 5-1, predicting satellite visibility of each candidate position by using a 3D city model, namely calculating the lowest elevation angle of the satellite which is not blocked by a building under different azimuth angles by using the 3D city model, when the elevation angle of the satellite under the azimuth angles is larger than the lowest elevation angle, considering the satellite as visible, otherwise, considering the satellite as invisible satellite, comparing the satellite visibility predicted by using the 3D city model with the satellite visibility judged by using a fisheye camera at an e candidate point, scoring the e candidate point, and if the visibility of the i satellite predicted by using the 3D city model at the e candidate point is consistent with the i satellite judged at the e candidate point by using the fisheye camera, adding a score to the e candidate point, otherwise, scoring the e candidate point unchanged. Marking the e candidate position as Mce;
step 5-2, excluding the e candidate position if the score Mce of the e candidate position is smaller than the score of the original candidate position; if the score Mce of the e candidate position is greater than or equal to the score of the original candidate position, calculating the weight of the candidate position according to the ranking from big to small of the score of the candidate position; the scores Mce of the N candidate positions in total are recorded to be greater than or equal to the scores of the original candidate positions,
in this embodiment, the candidate location weight formula calculated in step 5-2 is as follows:
wherein pw is q Represents the q candidate position weight, wherein q is more than or equal to 1 and less than or equal to N, N q Ranking from big to small showing scores according to the q-th candidate position, 1.ltoreq.n q ≤N。
In this embodiment, the formula for obtaining the final positioning result X based on the candidate position weight calculation in step 6 is as follows:
wherein X is q Representing the coordinates of the q-th candidate position.
The present invention provides an urban canyon positioning method based on GNSS/vision/Lidar fusion, and the method and the way for realizing the technical scheme are numerous, the above description is only a specific embodiment of the present invention, and it should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and the improvements and modifications should also be regarded as the protection scope of the present invention. The components not explicitly described in this embodiment can be implemented by using the prior art.

Claims (8)

1. An urban canyon positioning method based on GNSS/vision/Lidar fusion is characterized by comprising the following steps:
step 1, judging satellite visibility by using a fisheye camera;
step 2, further judging the signal type of the visible satellite based on a supervised learning mode;
step 3, determining the weight of the satellite and carrying out multi-constellation positioning calculation based on the weight of the satellite to obtain an initial positioning solution;
step 4, correcting the initial positioning solution and generating candidate positions;
step 5, screening candidate positions and calculating candidate position weights according to the degree of coincidence between the satellite visibility of the candidate positions and the satellite visibility in the step 1;
step 6, obtaining a final positioning result based on candidate position weight calculation;
the step 4 comprises the following steps:
step 4-1, measuring the distance from the carrier to the street building by using a pulse laser radar ranging technology, wherein the formula is as follows:
where c= 2.99792458 ×10 8 m/s is the propagation speed of light in vacuum, T is the transmission time of laser in the atmosphere, and d is the distance from the carrier to the building;
step 4-2, correcting the error of the initial position in the street crossing direction based on the 3D city model: if the distance d from the carrier to the left street is obtained, an error formula for correcting the street crossing direction of the initial position is as follows:
X repair tool ＝(x Repair tool ,y Repair tool ,z Repair tool ) T ＝(x Initially, the method comprises +(l-d)sinα,y Initially, the method comprises +(l-d)cosα,z Initially, the method comprises ) T
Wherein l is X Initially, the method comprises The distance from the street, alpha is the included angle between the street direction and the east direction;
similarly, if the distance d from the carrier to the right street is obtained, an error formula for correcting the street crossing direction of the initial position is as follows:
X repair tool ＝(x Repair tool ,y Repair tool ,Z Repair tool ) T ＝(x Initially, the method comprises -(l-d)sinα,y Initially, the method comprises -(l-d)cosα,z Initially, the method comprises ) T
If the distance from the carrier to the streets on two sides is obtained at the same time, one of the distances can be optionally corrected;
step 4-3, generating according to the L meter interval delta L meter before and after the corrected position along the street directionCandidate location, e-th generated candidateThe coordinates of the position being X e ，/>We call the corrected position the original candidate position.
2. The urban canyon positioning method based on GNSS/visual/Lidar fusion of claim 1, wherein step 1 comprises:
step 1-1, shooting a sky image by using a fisheye camera mounted on a carrier to point to the sky;
step 1-2, converting a sky image shot by a fisheye camera into a binary image, and distinguishing an image area of the sky from an image area shielded by a building;
step 1-3, projecting satellites into the sky image plane, wherein the projection formula is as follows:
wherein E is i The elevation angle of the ith satellite, i is a positive integer which is less than or equal to the sum of the total number of satellites of four constellations, alpha i Azimuth for the ith satellite, f c Focal length d of fish eye camera i For the distance between the position of the ith satellite projected to the image plane and the center of the image plane, beta is the included angle between the positive direction of the X axis of the local navigation coordinate system and the north direction,and->Is the coordinates of the center of the image plane,/>And->Coordinates projected into the sky image for the ith satellite; the four constellations comprise a global positioning system GPS, a Beidou satellite navigation system BDS, a GLONASS and a Galileo satellite navigation system GALILEO;
and 1-4, judging the visibility of the satellite, if the coordinates of the satellite projected into the sky image are in the sky area, judging the satellite as a visible satellite, otherwise, judging the satellite as an invisible satellite.
3. The urban canyon positioning method based on GNSS/vision/Lidar fusion according to claim 2, wherein the supervised learning method in step 2 adopts a linear Support Vector Machine (SVM) classifier.
4. The urban canyon positioning method according to claim 3, wherein said step 2 comprises:
step 2-1, in an off-line stage, LOS signals and Multipath signals are collected and marked in an urban environment;
step 2-2, training a linear SVM classifier by using the marked LOS signals and the Multipath signals, wherein a classification scoring formula of the linear SVM classifier is as follows:
u i ＝(v i /s) T α+γ
wherein u is i Scoring of the ith satellite, v i The feature vector of the SVM of the ith satellite comprises the carrier-to-noise ratio C/N of the satellite signal 0 The pseudo-range residual error and the pseudo-range change rate, s is a kernel scale, alpha is a vector for fitting a linear coefficient, and gamma is the deviation of a linear SVM classifier;
step 2-3, in an online stage, judging the satellite signal of the visible satellite in the step 1Is input into an SVM scoring formula to obtain a predicted satellite signal type, when u i Predicting the ith satellite signal as LOS signal when the signal is not less than 0, and when u i <And predicting the ith satellite signal as a Multipath signal at 0.
5. The urban canyon positioning method based on GNSS/visual/Lidar fusion of claim 4, wherein the step 3 comprises:
step 3-1, the carrier-to-noise ratio C/N of the satellite signal with the type of LOS signal obtained in step 2-3 0 And determining satellite weights by satellite elevation angles, wherein the total number of satellites with the satellite signal types of LOS signals obtained in the step 2-3 is J, and the satellite weights sw of the J-th satellite are recorded j The formula is:
wherein J is more than or equal to 1 and less than or equal to J, sigma j Is the standard deviation of error, S j In order for the scaling factor to be a factor,C/N 0j representing the carrier-to-noise ratio of the j-th satellite, E j Representing satellite elevation angle, a, of the j-th satellite 0 、a 1 And E is 0 Respectively constant terms, a 0 ＝3mm，a 1 ＝26mm，E 0 ＝20°；
Step 3-2, obtaining an initial position based on multi-constellation positioning calculation of a weighted least square method, firstly unifying a time system and a coordinate system of a global positioning system GPS, a Beidou satellite navigation system BDS, a Geronnass global satellite navigation system GLONASS and a Galileo satellite navigation system GALILEO, unifying four systems under a space-time system of the GPS, wherein the four systems are provided with n BDS satellites, m GPS satellites, k GLONASS satellites and l GALILEO satellitesStar, n+m+k+l=j, assuming the initial position of the carrier asThe position of the navigation satellite is (x j ,y j ,z j ) Considering the satellite weights assigned in step 3-1, the observation equation may be listed as follows:
WAX＝WB
where W is the weight matrix of the satellite, w=diag (sw 1 ,sw 2 ,...,sw J ) A is the satellite direction residual rotation matrix,
the geometric distance from the j satellite to the initial position is the geometric distance from the j satellite to the initial position; x= [ Δx, Δy, Δz, cΔt ] BDS ，cΔt GPS ，cΔt GLS ，cΔt GAO ] T ，cΔt BDS ，cΔt GPS ，cΔt GLS ，cΔt GAO Equivalent distance errors of carrier positions to BDS satellite, GPS satellite, GALILEO satellite and GLONASS satellite positions respectively caused by receiver clock bias, B is a pseudo-range observation value residual vector,'>
Wherein ρ is j A pseudorange representing the j-th satellite of the actual measurement;
calculating initial value compensation quantity of the corresponding position by using a weighted least square method, and sequentially iterating until the compensation quantity is smaller than an allowable error range threshold value, thereby realizing carrier positioning calculation and obtaining:
X＝(A T W T WA) -1 A T W T WB
the initial positions of the carrier were thus obtained:
6. the urban canyon positioning method based on GNSS/visual/Lidar fusion of claim 5, wherein step 5 comprises:
5-1, predicting satellite visibility of each candidate position by using a 3D city model, namely calculating the lowest elevation angle of the satellite, which is not blocked by a building, under different azimuth angles by using the 3D city model, when the elevation angle of the satellite under the azimuth angles is larger than the lowest elevation angle, considering the satellite as visible, otherwise, considering the satellite as invisible satellite, comparing the satellite visibility predicted by using the 3D city model with the satellite visibility judged by using a fisheye camera at an e candidate point, scoring the e candidate point, and if the visibility of the i satellite predicted by using the 3D city model at the e candidate point is consistent with the i satellite judged at the e candidate point by using the fisheye camera, adding a score to the e candidate point, otherwise, scoring the e candidate point; marking the e candidate position as Mce;
step 5-2, excluding the e candidate position if the score Mce of the e candidate position is smaller than the score of the original candidate position; if the score Mce of the e candidate position is greater than or equal to the score of the original candidate position, calculating the weight of the candidate position according to the ranking from big to small of the score of the candidate position; the scores Mce of the N candidate positions in total are recorded to be greater than or equal to the scores of the original candidate positions,
7. the urban canyon positioning method based on GNSS/visual/Lidar fusion of claim 6, wherein the candidate location weights calculated in step 5-2 are as follows:
wherein pw is q Represents the q candidate position weight, wherein q is more than or equal to 1 and less than or equal to N, N q Ranking from big to small showing scores according to the q-th candidate position, 1.ltoreq.n q ≤N。
8. The urban canyon positioning method based on GNSS/visual/Lidar fusion of claim 7, wherein the step 6 obtains the final positioning result X based on candidate position weight calculation by the formula:
wherein X is q Representing the coordinates of the q-th candidate position.
CN202110571260.4A 2021-05-25 2021-05-25 Urban canyon positioning method based on GNSS/vision/Lidar fusion Active CN113376675B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110571260.4A CN113376675B (en) 2021-05-25 2021-05-25 Urban canyon positioning method based on GNSS/vision/Lidar fusion

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110571260.4A CN113376675B (en) 2021-05-25 2021-05-25 Urban canyon positioning method based on GNSS/vision/Lidar fusion

Publications (2)

Publication Number Publication Date
CN113376675A CN113376675A (en) 2021-09-10
CN113376675B true CN113376675B (en) 2024-01-30

Family

ID=77571873

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110571260.4A Active CN113376675B (en) 2021-05-25 2021-05-25 Urban canyon positioning method based on GNSS/vision/Lidar fusion

Country Status (1)

Country Link
CN (1) CN113376675B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20230382565A1 (en) * 2022-05-31 2023-11-30 Qualcomm Incorporated Space vehicle geometry based machine learning for measurement error detection and classification
CN115657101B (en) * 2022-10-26 2023-09-15 北京航空航天大学 Method for assisting GNSS-INS (Global navigation satellite System-inertial navigation System) high-precision navigation positioning by using fisheye camera
CN115856976B (en) * 2023-02-27 2023-06-02 智道网联科技(北京)有限公司 Fusion positioning method and device for automatic driving vehicle and electronic equipment

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107064974A (en) * 2017-02-28 2017-08-18 广东工业大学 A kind of localization method and system for suppressing urban canyons multipath satellite-signal
CN110579780A (en) * 2019-08-28 2019-12-17 桂林电子科技大学 Shadow matching improvement algorithm based on Beidou GEO satellite
CN111051924A (en) * 2017-09-28 2020-04-21 苹果公司 Three-dimensional urban model and shadow mapping for improved height localization in urban environments
CN112666588A (en) * 2020-11-06 2021-04-16 南京航空航天大学 Positioning method based on scene matching and machine learning in urban canyon environment

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10495762B2 (en) * 2017-05-19 2019-12-03 Qualcomm Incorporated Non-line-of-sight (NLoS) satellite detection at a vehicle using a camera

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107064974A (en) * 2017-02-28 2017-08-18 广东工业大学 A kind of localization method and system for suppressing urban canyons multipath satellite-signal
CN111051924A (en) * 2017-09-28 2020-04-21 苹果公司 Three-dimensional urban model and shadow mapping for improved height localization in urban environments
CN110579780A (en) * 2019-08-28 2019-12-17 桂林电子科技大学 Shadow matching improvement algorithm based on Beidou GEO satellite
CN112666588A (en) * 2020-11-06 2021-04-16 南京航空航天大学 Positioning method based on scene matching and machine learning in urban canyon environment

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Efficient Satellite Selection Method for Instantaneous RTK-GNSS in Challenging Environments;Hiroko TOKURA等;《TRANSACTIONS OF THE JAPAN SOCIETY FOR AERONAUTICAL AND SPACE SCIENCES》;第221-229页 *
Rui Sun等.Improving GPS Code Phase Positioning Accuracy in Urban Environments Using Machine Learning.《IEEE INTERNET OF THINGS JOURNAL,》.2021,第7065-7078页. *
城市峡谷中基于GPS双极化天线的定位方法;王冠宇等;《导航定位学报》;第13-19页 *
鱼眼图像辅助下的低成本GPS城市区域定位研究;吴妍雯;《中国优秀硕士学位论文全文数据库基础科学辑》;第25-63页 *

Also Published As

Publication number Publication date
CN113376675A (en) 2021-09-10

Similar Documents

Publication Publication Date Title
CN113376675B (en) Urban canyon positioning method based on GNSS/vision/Lidar fusion
Ng et al. A computation effective range-based 3D mapping aided GNSS with NLOS correction method
CN107966724B (en) It is a kind of based on 3D city model auxiliary urban canyons in satellite positioning method
CN107621645A (en) Deception jamming signal detection method based on single receiver
Bai et al. Using Sky‐pointing fish‐eye camera and LiDAR to aid GNSS single‐point positioning in urban canyons
Groves et al. Shadow matching: Improved GNSS accuracy in urban canyons
Wen et al. Exclusion of GNSS NLOS receptions caused by dynamic objects in heavy traffic urban scenarios using real-time 3D point cloud: An approach without 3D maps
CN113031031B (en) GNSS signal accurate classification-based weighted positioning method in urban canyon
Bai et al. Real-time GNSS NLOS detection and correction aided by sky-pointing camera and 3D LiDAR
Sun et al. Using dual‐polarization GPS antenna with optimized adaptive neuro‐fuzzy inference system to improve single point positioning accuracy in urban canyons
Sun et al. An integrated urban positioning algorithm using matching, particle swam optimized adaptive neuro fuzzy inference system and a spatial city model
Suzuki Mobile robot localization with GNSS multipath detection using pseudorange residuals
Miura et al. GPS error correction by multipath adaptation
Sun et al. Stacking ensemble learning for non-line-of-sight detection of global navigation satellite system
Zhang et al. GNSS shadow matching based on intelligent LOS/NLOS classifier
CN115685260A (en) Identification deception jamming identification method, device, platform and medium for GLS reference station
CN113504553B (en) GNSS positioning method based on accurate 3D city model in urban canyon
Bradbury Prediction of urban GNSS availability and signal degradation using virtual reality city models
Wang Investigation of shadow matching for GNSS positioning in urban canyons
Tang et al. GNSS localization propagation error estimation considering environmental conditions
Ng et al. GNSS NLOS pseudorange correction based on skymask for smartphone applications
Ng et al. Range-based 3D mapping aided GNSS with NLOS correction based on skyplot with building boundaries
CN113933876B (en) Multi-star communication time difference positioning data fusion processing method
Hsu et al. Assessment of QZSS L1-SAIF for 3D map-based pedestrian positioning method in an urban environment
CN109613571A (en) Satellite positioning method in a kind of urban canyons based on dual polarized antenna

Legal Events

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