Map matching method based on hidden Markov model

Abstract

The invention discloses a map matching method based on a hidden Markov model, and belongs to the technical field of map matching. According to the method, a high-precision map is introduced, topological information such as lane-level front-back connection and lane change in the high-precision map and higher-precision position information of each lane are fully utilized, and lane matching is performed on continuous vehicle track points through a hidden Markov model, so that more accurate lane-level positioning correction is realized; and the time complexity and the space complexity of the operation are relatively low, so that the method is suitable for being deployed in a vehicle end calculation processing unit.

Description

Technical Field

[0001] The present invention belongs to the technical field of map matching, in particular to a map matching method based on a hidden Markov model. Background Art

[0002] The traditional vehicle positioning correction method based on the hidden Markov model only uses the standard map, so it can only perform road-level positioning of the vehicle, or whether the vehicle is traveling on the bridge deck of an overpass, and the positioning correction accuracy is not high.

[0003] After searching, the Chinese invention patent: A dynamic map matching method based on a high-order hidden Markov model (publication number CN112632202A, publication date 2021.04.09), this application discloses a dynamic map matching method based on a high-order hidden Markov model. The method obtains candidate sections of GPS points through GPS point data and road network information, and then obtains candidate points of GPS points in each candidate section, and then calculates the high-order hidden Markov model transition probability and the state transition probability of the high-order hidden Markov model of each candidate point of the GPS point, and uses the Viterbi algorithm to solve the high-order hidden Markov model to achieve matching of GPS points and sections in the road network, which can accurately achieve dynamic map matching. However, the shortcoming of this application is that it does not use high-precision maps, and the accuracy of positioning correction is not high. Summary of the Invention

[0004] 1. Technical problem to be solved by the invention

[0005] The purpose of the present invention is to solve the problem that the existing vehicle positioning correction method does not use high-precision maps and the positioning correction accuracy is not high.

[0006] 2. Technical solution

[0007] In order to achieve the above object, the technical solution provided by the present invention is:

[0008] A map matching method based on a hidden Markov model of the present invention comprises the following steps:

[0009] S1. Obtain the GNSS positioning observation sequence Z0, Z2...Z from the beginning time point 0 to time point T with a length of T through the GNSS device. T , where Z i =(lon, lat), which is the longitude and latitude coordinate information, where lon represents the longitude value and lat represents the latitude value;

[0010] S2, coordinate transformation;

[0011] S3, read high-precision map information;

[0012] S4. Build a time series hidden Markov framework model;

[0013] S5. Extract the lane connection relationship from the high-precision map and calculate the state transition matrix based on the connection relationship;

[0014] S6. Set the value of the state quantity X in the model to be the lane number;

[0015] S7, using the trajectory information and using a 10-meter radius as a threshold, extracting the relevant lanes as possible states;

[0016] S8. Calculate the observation probability of each state using the normal distribution through the two-dimensional position observation of each point in the trajectory;

[0017] S9. Calculate the maximum possible final state through the Viterbi algorithm;

[0018] S10. Obtain the time sequence state sequence corresponding to the maximum possible state through reverse calculation.

[0019] Preferably, the step S2 is specifically to convert the observed quantity into Cartesian coordinates, that is, Z i =(x, y), where x represents the abscissa value and y represents the ordinate value.

[0020] Preferably, the step S3 specifically comprises forming a square with a side length of L with each observation point in the observation sequence Z as the center, taking the union to form the largest including square, and obtaining map information of the corresponding area from the map module.

[0021] Preferably, the time series hidden Markov framework model is specifically composed of five parts: state space S, observation space O, state transition probability matrix A, observation probability matrix B and initial state probability distribution π; wherein the state space S represents all lanes in which the vehicle may travel, the observation space O is the location information obtained by the GNSS device, the state transition probability matrix A represents the probability of the vehicle transferring from one lane to another, the observation probability matrix B represents the probability of observing specific location information in a certain lane, and the initial state probability distribution π represents the probability distribution of the vehicle's initial position in each lane.

[0022] Preferably, step S5 is specifically to extract lane lines within the initial range and obtain lane line relationships based on the positioning information obtained by the GNSS device and combine the initial range of 20 meters of the positioning position with the high-precision map, and calculate the state transition matrix.

[0023] Preferably, the Viterbi algorithm in step S9 is specifically to record δ for time point t and state j. t(j) is the probability value of the sequence with the maximum probability of being in state j at time t among all possible partial state sequences up to time t; record ψ t (j) is the state of the maximum probability sequence at time t-1; when initialized, δ1(j) = π j b j (o1), ψ1(j)=0; in recursive calculation, for t=2,3,...,T, j=1,2,...,N, calculate and Calculate the final state probability P*=max[δ t (i)], the final state i*=argmax[δ t (i)], where π j is the initial state probability, b j (o t ) is the observed o in state j t The probability of a ij is the probability of transitioning from state i to state j.

[0024] Preferably, the reverse calculation in step S10 is specifically to start from the final state i and recursively t=T-1,T-2,...,1, backtrack and calculate the optimal state sequence q1*,q2*,...,q t *, where q t represents the optimal state at time t, Indicates that time t+1 is in state q t+1 , the index of the best previous state at time t.

[0025] 3. Beneficial effects

[0026] Compared with the prior art, the technical solution provided by the present invention has the following beneficial effects:

[0027] The present invention provides a map matching method based on a hidden Markov model. The method introduces a high-precision map and makes full use of topological information such as lane-level continuity and lane changes in the high-precision map, as well as the more accurate position information of each lane. Lane matching is performed on continuous vehicle trajectory points through the hidden Markov model, thereby achieving more accurate lane-level positioning correction. The time and space complexity of the operation are relatively small, and the method is suitable for deployment in the vehicle-side computing processing unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a logic flow chart of the method process of the present invention;

[0029] Figure 2Schematic diagram of the hidden Markov probability calculation framework of the present invention;

[0030] Figure 3 This is a flow chart of the state transition probability calculation of the present invention;

[0031] Figure 4 This is a flow chart of the probability initialization method of the present invention;

[0032] Figure 5 Flowchart for calculating the observation probability of the present invention;

[0033] Figure 6 This is a flow chart of the maximum probability inference method of the present invention. DETAILED DESCRIPTION

[0034] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0035] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0036] In this application, terms such as "upper," "lower," "left," "right," "front," "back," "top," "bottom," "inner," "outer," "center," "vertical," "horizontal," "transverse," and "longitudinal" indicate positions or locations based on the positions or locations shown in the accompanying drawings. These terms are primarily intended to better describe this application and its embodiments and are not intended to limit the devices, elements, or components indicated to having a specific orientation, or to being constructed or operated in a specific orientation.

[0037] Furthermore, some of the above terms may be used to express other meanings besides indicating a position or location. For example, the term "on" may also be used to indicate a dependency or connection in certain circumstances. Those skilled in the art will understand the specific meanings of these terms in this application based on the specific circumstances.

[0038] Furthermore, the terms "installed," "disposed," "provided with," "connected," "connected," and "socketed" should be interpreted broadly. For example, they can refer to fixed connections, removable connections, or integral structures; mechanical connections or electrical connections; direct connections, indirect connections through an intermediary, or internal communication between two devices, elements, or components. Those skilled in the art will understand the specific meanings of these terms in this application based on the specific circumstances.

[0039] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0040] Example 1

[0041] Refer to the attached Figure 1-6 A map matching method based on a hidden Markov model in this embodiment includes the following steps:

[0042] S1. Obtain the GNSS positioning observation sequence Z0, Z2...Z from the beginning time point 0 to time point T with a length of T through the GNSS device. T , where Z i =(lon, lat), which is the longitude and latitude coordinate information, where lon represents the longitude value and lat represents the latitude value;

[0043] S2, coordinate transformation;

[0044] S3, read high-precision map information;

[0045] S4. Build a time series hidden Markov framework model;

[0046] S5. Extract the lane connection relationship from the high-precision map and calculate the state transition matrix based on the connection relationship;

[0047] S6. Set the value of the state quantity X in the model to be the lane number;

[0048] S7, using the trajectory information, using a 10-meter radius as a threshold, extracting the relevant lanes as possible states;

[0049] S8. Calculate the observation probability of each state using the normal distribution through the two-dimensional position observation of each point in the trajectory;

[0050] S9. Calculate the maximum possible final state through the Viterbi algorithm;

[0051] S10. Obtain the time sequence state sequence corresponding to the maximum possible state through reverse calculation.

[0052] The step S2 is specifically to convert the observed quantity into Cartesian coordinates, namely Z i =(x, y), where x represents the abscissa value and y represents the ordinate value.

[0053] The step S3 specifically comprises forming a square with a side length of L with each observation point in the observation sequence Z as the center, taking the union to form the largest including square, and obtaining the map information of the corresponding area from the map module.

[0054] The temporal hidden Markov framework model specifically consists of five parts: state space S, observation space O, state transition probability matrix A, observation probability matrix B, and initial state probability distribution π; where the state space S represents all lanes that the vehicle may travel, the observation space O is the location information obtained by the GNSS device, the state transition probability matrix A represents the probability of the vehicle transferring from one lane to another, the observation probability matrix B represents the probability of observing specific location information in a certain lane, and the initial state probability distribution π represents the probability distribution of the vehicle's initial position in each lane.

[0055] Step S5 specifically involves extracting lane lines within the initial range and obtaining lane line relationships based on the positioning information obtained by the GNSS device and combining the initial 20-meter radius of the positioning position with the high-precision map, and calculating a state transition matrix.

[0056] The Viterbi algorithm in step S9 is specifically to record δ for time point t and state j. t (j) is the probability value of the sequence with the maximum probability of being in state j at time t among all possible partial state sequences up to time t; record ψ t (j) is the state of the maximum probability sequence at time t-1; when initialized, δ1(j) = π j b j (o1), ψ1(j)=0; in recursive calculation, for t=2,3,...,T, j=1,2,...,N, calculate and Calculate the final state probability P*=max[δ t (i)], the final state i*=argmax[δ t (i)], where π j is the initial state probability, b j (o t ) is the observed o in state jt The probability of a ij is the probability of transitioning from state i to state j.

[0057] The reverse calculation in step S10 is specifically to start from the final state i and use the recursive formula q t =ψ t+1 (q t+1 *), t=T-1,T-2,...,1, and backtrack to calculate the optimal state sequence q1*,q2*,...,q t *, where q t represents the optimal state at time t, Indicates that time t+1 is in state q t+1 , the index of the best previous state at time t.

[0058] The method of this embodiment introduces high-precision maps, making full use of topological information such as lane-level continuity and lane changes in the high-precision maps, as well as more accurate lane position information, and uses a hidden Markov model to perform lane matching on continuous vehicle trajectory points, thereby achieving more accurate lane-level positioning correction. The time and space complexities of the calculation are relatively small, making it suitable for deployment in the vehicle-side computing processing unit.

[0059] The above-mentioned embodiments only express a certain implementation method of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the patent scope of the present invention. It should be pointed out that for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent of the present invention shall be based on the attached claims.

Claims

1. A map matching method based on a hidden Markov model, characterized by: Includes the following steps S1. Obtain the GNSS positioning observation sequence Z0, Z2...Z from the beginning time point 0 to time point T with a length of T through the GNSS device. T , where Z i =(lon,lat), which is the longitude and latitude coordinate information, where lon represents the longitude value and lat represents the latitude value; S2, coordinate transformation; S3, read high-precision map information; S4. Build a time series hidden Markov framework model; S5. Extract the lane connection relationship from the high-precision map and calculate the state transition matrix based on the connection relationship; S6. Set the value of the state quantity X in the model to be the lane number; S7, using the trajectory information and using a 10-meter radius as a threshold, extracting the relevant lanes as possible states; S8. Calculate the observation probability of each state using the normal distribution through the two-dimensional position observation of each point in the trajectory; S9. Calculate the maximum possible final state through the Viterbi algorithm; S10. Obtain the time sequence state sequence corresponding to the maximum possible state through reverse calculation.

2. A map matching method based on a hidden Markov model according to claim 1, characterized in that: The step S2 is specifically to convert the observed quantity into Cartesian coordinates, namely Z i =(x, y), where x represents the abscissa value and y represents the ordinate value.

3. The map matching method based on a hidden Markov model according to claim 2, characterized in that: The step S3 specifically comprises forming a square with a side length of L with each observation point in the observation sequence Z as the center, taking the union to form the largest including square, and obtaining the map information of the corresponding area from the map module.

4. A map matching method based on a hidden Markov model according to claim 3, characterized in that: The temporal hidden Markov framework model specifically consists of five parts: state space S, observation space O, state transition probability matrix A, observation probability matrix B, and initial state probability distribution π; where the state space S represents all lanes that the vehicle may travel, the observation space O is the location information obtained by the GNSS device, the state transition probability matrix A represents the probability of the vehicle transferring from one lane to another, the observation probability matrix B represents the probability of observing specific location information in a certain lane, and the initial state probability distribution π represents the probability distribution of the vehicle's initial position in each lane.

5. The map matching method based on a hidden Markov model according to claim 4, characterized in that: Step S5 specifically involves extracting lane lines within the initial range and obtaining lane line relationships based on the positioning information obtained by the GNSS device and combining the initial 20-meter radius of the positioning position with the high-precision map, and calculating a state transition matrix.

6. The map matching method based on a hidden Markov model according to claim 5, characterized in that: The Viterbi algorithm in step S9 is specifically to record δ for time point t and state j. t (j) is the probability value of the sequence with the maximum probability of being in state j at time t among all possible partial state sequences up to time t; record ψ t (j) is the state of the maximum probability sequence at time t-1; when initialized, δ1(j) = π j b j (o1), ψ1(j) = 0; in recursive calculation, for t = 2, 3, ..., T, j = 1, 2, ..., N, calculate δ t (j) = max[δ t-1 (i) a ij ]·b j (o t ) and ψ t (j) = argmax[δ t-1 (i) a ij ]; calculate the final state probability P*=max[δ t (i)], the final state i*=argmax[δ t (i)], where π j is the initial state probability, b j (o t ) is the observed o in state j t The probability of a ij is the probability of transitioning from state i to state j.

7. The map matching method based on a hidden Markov model according to claim 5, characterized in that: The reverse calculation in step S10 is specifically to start from the final state i and use the recursive formula q t =ψ t+1 (q t+1 *), t=T-1,T-2,...,1, and backtrack to calculate the optimal state sequence q1*,q2*,...,q t *, where q t represents the optimal state at time t, ψ t+1 Indicates that time t+1 is in state q t+1 , the index of the best previous state at time t.

Images