Estimation device, estimation method, and storage medium

Abstract

The estimation device (1) comprises: a dead reckoning block (100) for calculating the dead reckoning position based on a parameter (p t ) of the dynamic model (DM), and the internal boundary information (I i ) of dead reckoning, estimate the state quantity (Z t ); a map matching block (120) based on map information (I m ), and external information (I) obtained from the outside of the vehicle (4) o ) map matching, observation state quantity (Z m ); and a parameter correction block (140), based on the state quantity (Z t ) relative to the state quantity (Z) observed by map matching m ) offset (δz), and the correction feedback to the dead reckoning parameter (p t ).

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application is based on patent application No. 2019-210764 filed in Japan on November 21, 2019, and the contents of the basic application are cited in their entirety by reference. Technical Field

[0003] The present disclosure relates to an estimation technique for estimating a state quantity of a vehicle. Background Art

[0004] Conventionally, the state quantity including the position of the vehicle is estimated by dead reckoning based on the inner boundary information obtained from the inner boundary of the vehicle. As one of such estimation techniques, the disclosed technique of Patent Document 1 uses a Kalman filter based on a state equation that is a dynamic model in order to correct the navigation error included in the estimation result in dead reckoning.

[0005] Patent Document 1: Japanese Patent Application Publication No. 2008-249639

[0006] In the disclosed technology of Patent Document 1, if the parameters included in the dynamics model as the parameters that change the vehicle behavior themselves have errors, the accuracy of dead reckoning is reduced. Among the parameters that change the vehicle behavior, the variation corresponding to important factors other than the vehicle behavior, such as the road condition, becomes an error. However, such parameter errors cannot be correctly corrected or compensated by the Kalman filter, and the estimation accuracy of the state quantity is reduced. Summary of the invention

[0007] The subject of the present disclosure is to provide an estimation device that improves the estimation accuracy of a state quantity. Another subject of the present disclosure is to provide an estimation method that improves the estimation accuracy of a state quantity. Still another subject of the present disclosure is to provide a storage medium that improves the estimation accuracy of a state quantity.

[0008] Hereinafter, technical means of the present disclosure for solving the problems will be described.

[0009] A first aspect of the present disclosure is an estimation device for estimating a state quantity including a position of a vehicle, comprising:

[0010] a dead reckoning unit that estimates a state quantity by dead reckoning based on a dynamics model including a parameter that changes the behavior of the vehicle and internal boundary information acquired from an internal boundary of the vehicle;

[0011] a map matching unit that observes the state quantity by map matching based on map information representing a driving environment of the vehicle and external information acquired from outside the vehicle; and

[0012] The parameter correction unit corrects a parameter fed back to the dead reckoning based on an offset of a state quantity estimated by the dead reckoning relative to a state quantity observed by the map matching.

[0013] A second aspect of the present disclosure is a method for estimating a state quantity including a position of a vehicle, comprising:

[0014] a dead reckoning step of estimating a state quantity by dead reckoning based on a dynamics model including parameters for changing the behavior of the vehicle and internal boundary information acquired from an internal boundary of the vehicle;

[0015] a map matching step of observing the state quantity by map matching based on map information representing the driving environment of the vehicle and external information acquired from the outside of the vehicle; and

[0016] The parameter correction step corrects the parameters fed back to the dead reckoning based on the offset of the state quantity estimated by the dead reckoning relative to the state quantity observed by the map matching.

[0017] A third aspect of the present disclosure is a storage medium storing an estimation program including a command for causing a processor to execute in order to estimate a state quantity including a position of a vehicle.

[0018] The commands include:

[0019] a dead reckoning step of estimating a state quantity by dead reckoning based on a dynamics model including parameters for changing the behavior of the vehicle and internal boundary information acquired from an internal boundary of the vehicle;

[0020] a map matching step of observing the state quantity by map matching based on map information representing the driving environment of the vehicle and external information acquired from the outside of the vehicle; and

[0021] The parameter correction step corrects the parameters fed back to the dead reckoning based on the offset of the state quantity estimated by the dead reckoning relative to the state quantity observed by the map matching.

[0022] According to the first to third aspects, the correction of the parameter that changes the vehicle behavior is based on the offset of the estimated state quantity of dead reckoning relative to the observed state quantity of map matching. According to this correction, the parameter can be accurately corrected and fed back to the dead reckoning. Therefore, by using a dynamic model including the fed-back parameter, the estimation accuracy of the state quantity of dead reckoning can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a block diagram showing the overall configuration of an estimation device according to one embodiment.

[0024] Figure 2 This is a block diagram showing a detailed configuration of an estimation device according to one embodiment.

[0025] Figure 3 Schematic diagram for explaining a dead reckoning block according to an embodiment.

[0026] Figure 4 Schematic diagram for explaining a dead reckoning block according to an embodiment.

[0027] Figure 5 This is a block diagram showing a detailed configuration of an estimation device according to a modified example of one embodiment.

[0028] Figure 6 It is a schematic diagram for explaining a map matching block according to one embodiment.

[0029] Figure 7 It is a schematic diagram for explaining a parameter correction block according to one embodiment.

[0030] Figure 8 This is a block diagram showing a detailed configuration of an estimation device according to a modified example of one embodiment.

[0031] Fig. 9 This is a block diagram showing a detailed configuration of an estimation device according to a modified example of one embodiment.

[0032] Fig.10 Detailed description of the invention is given in the following figure.

[0033] Fig.11 This is a flowchart showing an estimation method according to a modification of one embodiment.

[0034] Fig.12 This is a flowchart showing an estimation method according to a modification of one embodiment. DETAILED DESCRIPTION

[0035] Hereinafter, one embodiment of the present disclosure will be described based on the drawings.

[0036] like Figure 1 as well as Figure 2 As shown, the estimation device 1 according to one embodiment is mounted on a vehicle 4 together with a sensor unit 2 and a map unit 3. The sensor unit 2 includes an outside world sensor 22 and an inside world sensor 24.

[0037] The external sensor 22 acquires information that can be used for estimating the motion of the vehicle 4 from the outside world that is the surrounding environment of the vehicle 4 as external information I. o The external sensor 22 can also obtain external information by detecting objects outside the vehicle 4. oThe external sensor 22 of the detection type is, for example, one or more of LiDAR (Light Detection and Ranging / Laser Imaging Detection and Ranging), camera, radar, and sonar. The external sensor 22 can also obtain external information by receiving signals from a wireless communication system outside the vehicle 4. o The receiving type of external sensor 22 is, for example, one or more of a GNSS (Global Navigation Satellite System) receiver and an ITS (Intelligent Transport Systems) receiver. Hereinafter, the present embodiment will be described by taking as an example a case where a LiDAR is mounted on the vehicle 4 as at least the external sensor 22, which detects reflected light from a reflection point by irradiating light to the outside world and generates a point group image.

[0038] The inner space sensor 24 acquires information that can be used for motion estimation of the vehicle 4 from the inner space that is the internal environment of the vehicle 4 as the inner space information I. i The inner boundary sensor 24 can also obtain the inner boundary information by detecting a specific motion physical quantity in the inner boundary of the vehicle 4. i The inner limit sensor 24 of the detection type is, for example, one or more of an inertial sensor, a vehicle speed sensor, and a steering angle sensor. Hereinafter, the present embodiment will be described by taking as an example a case where a gyro sensor as an inertial sensor for detecting the angular velocity of the vehicle body, a vehicle speed sensor for detecting the speed of the vehicle body, and a steering angle sensor for detecting the steering angle of the wheel relative to the vehicle body are mounted on the vehicle 4 as at least the inner limit sensor 24.

[0039] The map unit 3 is configured to include a non-temporary storage of map information I m The map unit 3 may be a database of a locator that can be used for advanced driving assistance or automatic driving control of the vehicle 4. The map unit 3 may also be a database of a navigation device that guides the driving of the vehicle 4. The map unit 3 may also be composed of a combination of multiple types of these databases.

[0040] Map Information I m The information representing the driving environment of the vehicle 4 is converted into two-dimensional or three-dimensional data. mAt least one or more road information including, for example, the position, shape, and road surface condition of the road itself. m It may also include one or more identification information such as signs attached to the road and the position and shape of lane dividing lines. m For example, structural information indicating one or more of the positions and shapes of buildings and traffic lights facing the road may be included.

[0041] like Figure 1 As shown, the estimation device 1 is connected to the sensor unit 2 and the map unit 3 via one or more of a LAN (Local Area Network), a wiring harness, and an internal bus, for example. The estimation device 1 is configured to include at least one dedicated computer. The dedicated computer constituting the estimation device 1 may also be an ECU (Electronic Control Unit) dedicated to driving control that performs advanced driving assistance or automatic driving control of the vehicle 4. The dedicated computer constituting the estimation device 1 may also be an ECU of a locator utilized for advanced driving assistance or automatic driving control of the vehicle 4. The dedicated computer constituting the estimation device 1 may also be an ECU of a navigation device that navigates the driving of the vehicle 4. The dedicated computer constituting the estimation device 1 may also be an ECU of a communication control device that controls communication between the vehicle 4 and the outside world.

[0042] The estimation device 1 is constructed to include such a dedicated computer, and at least includes a memory 10 and a processor 12. The memory 10 is a non-transitory tangible storage medium such as a semiconductor memory, a magnetic medium, an optical medium, etc., which non-temporarily stores programs and data that can be read by the computer. The processor 12 includes, for example, one or more of a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), and a RISC (Reduced Instruction Set Computer)-CPU as a core. The processor 12 executes a plurality of commands included in the estimation program stored in the memory 10. Thus, the estimation device 1 is as follows. Figure 2 As shown, a plurality of functional blocks are constructed for estimating the state quantity including the position of the vehicle 4. As described above, in the estimation device 1, in order to estimate the state quantity of the vehicle 4, the estimation program stored in the memory 10 causes the processor 12 to execute a plurality of instructions, thereby constructing a plurality of functional blocks.

[0043] The plurality of functional blocks constructed by the estimation device 1 include a dead reckoning block 100, a map matching block 120, a parameter correction block 140, and an error compensation block 160. The dead reckoning block 100 is based on a dynamics model DM (Dynamics Model) and internal boundary information I i The dead reckoning of the vehicle 4 is used to estimate the state quantity of the vehicle 4. For this purpose, the dead reckoning block 100 includes a behavior estimation sub-block 102 and a state quantity estimation sub-block 104. In the following description, dead reckoning is expressed as DR (Dead Reckoning).

[0044] The behavior estimation sub-block 102 is implemented by Figure 3 The dynamic model DM shown estimates the behavior of the driving state of the vehicle 4 at the latest time t (hereinafter simply referred to as the vehicle behavior). Specifically, the dynamic model DM is a two-wheel model, etc., after the vehicle behavior is modeled based on dynamics. The dynamic model DM of this embodiment is modeled so as to be able to estimate the slip angle β as a kinematic physical quantity of a specific vehicle behavior. Figure 3 In , γ and ψ are the angular velocity and steering angle of the vehicle 4, respectively. Figure 3 In , F, G and L are the lateral force, lateral acceleration and wheelbase of the vehicle 4 respectively. Figure 3 In , r added to β, F, and L is a suffix indicating a component of the rear model of the vehicle 4. Figure 3 , f added to β, F, and L is a suffix indicating a component of the front model of the vehicle 4 .

[0045] Towards Figure 2 The behavior estimation sub-block 102 shown in FIG. 1 is input with the internal boundary information I obtained by the internal boundary sensor 24 at the latest time t. i The angular velocity γ t and the rudder angle ψ t The behavior estimation sub-block 102 receives as input a parameter p related to the dynamics at the latest time t (hereinafter referred to as a dynamic parameter) which is corrected and fed back by a parameter correction block 140 described in detail later. t . Dynamic parameter p t The dynamic parameter p in this embodiment includes one or more physical quantities that change according to important factors other than vehicle behavior. t At least the dynamic friction coefficient of the road surface on which the vehicle 4 is traveling. In addition to the dynamic friction coefficient, the dynamic parameter p t The weight of the vehicle 4 including the weight of the passengers may also be included.

[0046] For all but the angular velocity γ t and the rudder angle ψ t The dynamic parameter p is input to the behavior estimation sub-block 102. t For example, according to the parameter pt The change of makes the slip angle β estimated as the vehicle behavior at the latest time t by the dynamic model DM of this embodiment t Therefore, the behavior estimation sub-block 102 takes the dynamic parameter p as t With angular velocity γ t and the rudder angle ψ t The model function M based on the dynamic model DM is used as a variable d The following formula 1, for the slip angle β at the latest time t t Perform an estimation operation.

[0047] [Formula 1]

[0048] β t =M d (γ t , ψ t , p t )··FORMULA 1

[0049] Figure 2 The state quantity estimation sub-block 104 shown in FIG. 1 estimates the state quantity Z of the vehicle 4 at the latest time t based on the estimation result of the dynamic model DM of the behavior estimation sub-block 102. t Specifically, by using the two-dimensional coordinate position x of the vehicle 4 t ,y t and the yaw angle θ t The vector form of the following equation 2 defines the state quantity Z at the latest time t t .

[0050] [Formula 2]

[0051]

[0052] like Figure 4 As shown, the two-dimensional coordinate position x at the last estimated time t-1 is used. t-1 ,y t-1 The following equations 3 and 4 of the linear function represent the state quantity Z constituting the latest time t t The two-dimensional coordinate position x t ,y t In equations 3 and 4, β t is the slip angle at the latest time t estimated by the behavior estimation sub-block 102. In equations 3 and 4, V t is the speed of vehicle 4. In equations 3 and 4, θ t is the state quantity Z that constitutes the latest time t t The yaw angle is calculated by using the value θ at the last estimation time t-1. t-1 The following equation 5 represents the yaw angle θ tIn equations 3, 4, and 5, Δt is the difference between the latest time t and the last estimated time t-1, that is, the estimated time interval. In equation 5, γ t is the angular velocity of the vehicle 4 at the latest time t.

[0053] [Formula 3]

[0054] x t =x t-1 +Vt·Δt·cos(θ t +β t )··FORMULA 3

[0055] [Formula 4]

[0056] y t =y t-1 +V t ·Δt·sin(θ t +β t )··FORMULA 4

[0057] [Formula 5]

[0058] θ t =θ t-1 +γ t ·Δt··FORMULA 5

[0059] like Figure 2 As shown, the state quantity estimation sub-block 104 is input with the inner boundary information I obtained by the inner boundary sensor 24 at the latest time t. i The vehicle speed V t and the angular velocity γ t The state quantity estimation sub-block 104 receives as input the slip angle β at the latest time t estimated by the dynamics model DM of the behavior estimation sub-block 102. t The state quantity Z stored in the previous estimation in association with the previous time t-1 is also input to the state quantity estimation sub-block 104 by reading from the estimation storage area 180 in the memory 10. t-1 Based on these inputs, the state quantity estimation sub-block 104 estimates the state quantity Z at the latest time t, which constitutes the equation 2 in which equations 3, 4, and 5 are substituted. t The two-dimensional coordinate position x t ,y t and the yaw angle θ t The state quantity estimation sub-block 104 stores the result of the current estimation calculation in the estimation storage area 180 in association with the latest time t.

[0060] Here, the second terms on the right sides of equations 3 and 4 substituted into equation 1 and the second term on the right side of equation 5 are combined into the following equation 6, which is a displacement function ΔZ that represents the estimated displacement amount of the state amount at the estimated time interval Δt by a vector expression. This equation 6 can also be said to be obtained by using the slip angle β t The model function M of the kinetic model DM d , define the dynamic parameter p t The displacement function ΔZ is used as a variable. According to the above, by using the state quantity Z at the last estimation time t-1 t-1 The following equation 7 of the displacement function ΔZ at the latest time t represents the state quantity Z at the latest time t: t Therefore, if Figure 5 As shown in the modified example, the DR block 100 is based on the angular velocity γ t 、Rudder angle ψ t , vehicle speed V t , dynamic parameter p t And the last state quantity Z t-1 The input of the slip angle β t The state quantity Z of equation 7, which is substituted into equation 6, is estimated by the calculation of t The two-dimensional coordinate position x t ,y t and the yaw angle θ t .

[0061] [Formula 6]

[0062]

[0063] [Formula 7]

[0064] Z t =Z t-1 +ΔZ(p t )··FORMULA 7

[0065] Return to Figure 2 The map matching block 120 performs the mapping based on the map information I m and external information o The state quantity of the vehicle 4 is observed by performing map matching. In the following description, map matching is expressed as MM (Map Matching).

[0066] Specifically, the state quantity Z at the latest time t estimated by the DR block 100 is input to the MM block 120. t The state quantity Z corresponding to the input is input to the MM block 120 by reading from the map unit 3. t The two-dimensional coordinate position x in t ,y t Corresponding map information I mThe MM block 120 is input with a point group image acquired at the latest time t by the LiDAR in the external sensor 22 as the external information I o Based on these inputs, the MM block 120 obtains the map information I m Extract as Figure 6 As shown in the figure, the observation point group S obtained by observing the external object in the point group image o Multiple feature points S for matching m In addition, Figure 6 In the figure, the observation point group S is represented by white circles. o On the other hand, multiple feature points S are represented by black circles m .

[0067] The MM block 120 is based on the observation point group S o Multiple feature points S for matching m , observe the state quantity Z of vehicle 4 at the latest time t m At this time, the MM block 120 uses the two-dimensional coordinate position x of the vehicle 4 m ,y m and the yaw angle θ m The vector form of the following equation 8 is defined, and the state quantity Z m Perform an estimation operation.

[0068] [Formula 8]

[0069]

[0070] Figure 2 The parameter correction block 140 shown is based on the state quantity Z estimated by DR t like Figure 7 As shown in FIG. 1 , relative to the state quantity Z observed by the MM block 120 m The offset δz of the offset, the corrected dynamic parameter p t To this end, the parameter correction block 140 has a minimization sub-block 142 and an adjustment sub-block 144 .

[0071] The state quantity Z at the latest time t estimated by the DR block 100 is input to the minimization sub-block 142. t , and the state quantity Z at time t observed by MM block 120 m Based on these inputs, the minimization sub-block 142 obtains the estimated state quantity Z by performing a difference operation of the following equation 9: t Relative to the observed state quantity Z m The offset is the offset δz.

[0072] [Formula 9]

[0073] δz=Z m -Z t··FORMULA 9

[0074] Minimize the prediction of sub-block 142 as Figure 7 The estimated state quantity Z of the DR block 100 at the latest time t in the obtained offset δz is shown as follows: t According to the dynamic parameter p t The change δp of the deviation δz p Specifically, the deviation δz p As the dynamic parameters p t 、p t-1 The difference between the displacement function ΔZ at the latest time t and the previous time t-1 as a variable is defined by the following equation 10. Here, equation 10 uses the dynamic parameter p at the latest time t and the previous time t-1. t 、p t-1 The amount of change δp between the two is transformed into the following equation 11. The amount of deviation δz p It can also be used as a dynamic parameter p that combines the latest time t and the last time t-1 t 、p t-1 The value of the displacement function ΔZ with the change amount δp between the two as a variable is approximated by the following equation 12.

[0075] [Formula 10]

[0076] δz p =ΔZ(p t )-ΔZ(p t-1 )··FORMULA 10

[0077] [Formula 11]

[0078] δz p =ΔZ(p t-1 +δp)-Z(p t-1 )··FORMULA 11

[0079] [Formula 12]

[0080] δz p =Δz(δp)··FORMULA 12

[0081] Therefore, the dynamic parameter p at the previous time t-1 is input to the minimization sub-block 142 by reading from the estimation storage area 180 in the memory 10. t-1 Based on the input, the minimization sub-block 142 uses the minimization function M of the following equation 13: a The optimization operation is performed to obtain the deviation δz with the minimum absolute value of the difference (i.e., the absolute difference) from the offset δz p At this time, the minimization sub-block 142 is based on the dynamic parameter p t-1Substituting into equation 11 or equation 12, we can also obtain the deviation δz p The corresponding optimal value of the variation δp.

[0082] [Formula 13]

[0083] M a =argmin|δz-δz p |··FORMULA 13

[0084] By reading from the estimation storage area 180 Figure 2 The adjustment sub-block 144 shown has as input the dynamic parameter p at the last time t-1 t-1 The adjustment sub-block 144 is fed with the deviation δz p The optimal value of the variation δp is correspondingly predicted by the minimization sub-block 142. The adjustment sub-block 144 adjusts the Kalman gain K based on these inputs. p and the dynamic parameter p at the last time t-1 t-1 The weighted calculation of the following formula 14 is used together with the optimal value of the change amount δp to obtain the dynamic parameter p at the latest time t. t This can be said to be the Kalman filter, which corrects the dynamic parameter p according to the optimal value of the change δp. t The adjustment sub-block 144 converts the modified dynamic parameter p into t The DR is stored in the estimation storage area 180 in association with the latest time t, and is fed back to the DR block 100 as a dynamic parameter related to the next latest time.

[0085] [Formula 14]

[0086] p t =K p ·p t-1 +(1-K p )·(p t-1 +δp)··FORMULA 14

[0087] Figure 2 The error compensation block 160 shown in FIG. 1 is based on the state quantity Z observed by the MM block 120. m The state quantity Z estimated by the DR block 100 is filtered. t to compensate for the error.

[0088] Specifically, the state quantity Z at the latest time t estimated by the DR block 100 is input to the error compensation block 160. t , and the state quantity Z at the latest time t observed by the MM block 120 m The error compensation block 160 is based on these inputs by converting the Kalman gain K zThe observed state quantity Z of MM block 120 m The weighted operation of the following formula 15 used together compensates the estimated state quantity Z of the DR block 100 t This can be said to be the error of the observed state quantity Z through the Kalman filter m Fusion, thereby determining the estimated state quantity Z t In addition, the estimated state quantity Z determined by error compensation t The output from the error compensation block 160 is utilized, for example, for advanced driving assistance or automatic driving control of the vehicle 4 .

[0089] [Formula 15]

[0090] Z t ←(1-Kz)·Zt+K z ·Z m ··FORMULA 15

[0091] Here, equation 15 is transformed by using the offset δz and expressed by the following equation 16. Therefore, Figure 8 as well as Fig. 9 As shown in the modified example, the error compensation block 160 can also be changed by replacing the Kalman gain K z The weighted operation of the following equation 16 used together with the acquired amount δz of the minimization sub-block 142 compensates the estimated state amount Z of the DR block 100 t In addition, Figure 8 Shown relative to Figure 2 On the other hand, Fig. 9 Shown relative to Figure 5 Further variations of the variations of .

[0092] [Formula 16]

[0093] Z t ←(1-Kz)·Z t +K z ·(Z t +δ z )··FORMULA 16

[0094] Based on the above, in this embodiment, the DR block 100 corresponds to the “dead reckoning unit”, and the MM block 120 corresponds to the “map matching unit”. In addition, in this embodiment, the parameter correction block 140 corresponds to the “parameter correction unit”, and the error compensation block 160 corresponds to the “error compensation unit”.

[0095] Below, according to Fig.10The flow of the estimation method for estimating the state quantity of the vehicle 4 by the estimation device 1 through the cooperation of the functional blocks 100, 120, 140 and 160 is described. In addition, this flow is executed at each estimation timing that the vehicle 4 needs to be estimated. In addition, in this flow, each "S" refers to a plurality of steps executed according to a plurality of commands included in the estimation program.

[0096] In S101, the DR block 100 performs a vehicle behavior change based on the dynamic parameters p t The dynamic model DM and the inner boundary information I obtained from the inner boundary of the vehicle 4 i DR, estimate the state Z of vehicle 4 t In S102, the MM block 120 generates a map based on the map information I representing the driving environment of the vehicle 4. m , and external information I obtained from the outside of the vehicle 4 o MM, observe the state Z of vehicle 4 m .

[0097] In S103, the parameter correction block 140 calculates the state quantity Z estimated by DR in S101. t The state quantity Z observed by MM in S102 m The offset δz of the offset, the corrected dynamic parameter p t At this time, the dynamic parameter p changes according to important factors other than vehicle behavior. t The parameter correction block 140 takes the parameters including the coefficient of kinetic friction of the road surface on which the vehicle 4 is traveling as the correction object. Therefore, the parameter correction block 140 predicts the estimated state quantity Z of DR in the offset quantity δz. t According to the dynamic parameter p as the correction object t The change δp of the deviation δz p At this time, the parameter correction block 140 predicts the deviation δz so that the absolute value of the difference with the offset δz is minimized. p The parameter correction block 140 also calculates the deviation from the prediction by δz p The optimal value of the corresponding variation δp, the modified dynamic parameter p t The parameter correction block 140 further converts the corrected dynamic parameter p t The DR in the DR block 100 of S101 of the next process is fed back.

[0098] In S104, the error compensation block 160 calculates the state quantity Z observed by MM based on S102. m Filtering, compensation and determination of the state quantity Z estimated by DR in S101 t The above results in the current execution of this process being completed. In addition, S103 and S104 can be Fig.10 Follow the steps in the order shown, or Fig.11 The modification shown in the figure is performed in the reverse order, or Fig.12 The above steps are executed simultaneously and in parallel as in the other variations shown.

[0099] Based on the above, in this embodiment, S101 corresponds to the “dead reckoning process”, and S102 corresponds to the “map matching process”. In addition, in this embodiment, the set of S103 corresponds to the “parameter correction process”, and S104 corresponds to the “error compensation process”.

[0100] (Effect)

[0101] Hereinafter, the effects of the above-described embodiment will be described.

[0102] According to the present embodiment, the dynamic parameter p that changes the vehicle behavior t The correction of the estimated state quantity Z based on DR t Relative to the observed state quantity Z of MM m By this correction, the dynamic parameter p can be correctly corrected. t And fed back to DR. Therefore, by including the feedback dynamic parameters p t The dynamic model DM can improve the state quantity Z of DR t The estimation accuracy.

[0103] According to this embodiment, the estimated state quantity Z of DR in the offset δz t According to the dynamic parameter p t The change δp of the deviation δz p After the prediction, according to the deviation from the prediction δz p The corresponding change δp, the modified dynamic parameter p t According to this correction, the dynamic parameter p can accurately reflect the variation δp which is an important factor in the generation of the offset δz. t Therefore, by including the feedback dynamic parameters p t The dynamic model DM can improve the state quantity Z of DR t The estimation accuracy.

[0104] According to the present embodiment, the deviation amount δz having the smallest difference from the offset amount δz p The corresponding change δp is calculated based on the dynamic parameter p including the correction according to the change. t The DR of the dynamic model DM can make the estimated state quantity Z t Close to the observed state quantity Z mThat is, the state quantity Z of DR can be improved t The estimation accuracy.

[0105] According to the present embodiment, the dynamic parameters p including the dynamic friction coefficient of the road surface on which the vehicle is traveling are corrected based on the offset δz. t The dynamic parameter p that changes according to important factors other than vehicle behavior t According to this correction, the dynamic parameter p can be made into an error by using the variation corresponding to the important factor other than the vehicle behavior (ie, the variation δp). t Therefore, by including the dynamic parameter p fed back according to the correction t The dynamic model DM can improve the state quantity Z of DR t The estimation accuracy.

[0106] According to this embodiment, the observed state quantity Z based on MM m Filtering, compensating DR's estimated state quantity Z t According to the error compensation, the dynamic parameter p is fed back according to the correction based on the offset δz. t The dynamic model DM is combined with the state quantity Z t higher estimation accuracy.

[0107] (Other Embodiments)

[0108] Although one embodiment has been described above, the present disclosure is not limited to this embodiment and can be applied to various embodiments and combinations within the scope not departing from the gist of the present disclosure.

[0109] Specifically, the dedicated computer constituting the estimation device 1 of the modified example may also include at least one of a digital circuit and an analog circuit as a processor. Here, in particular, the digital circuit is at least one of an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), a SOC (System on a Chip), a PGA (Programmable Gate Array), and a CPLD (Complex Programmable Logic Device). In addition, such a digital circuit may also have a memory storing a program.

[0110] In the estimation device 1 of the modified example, the yaw angle θ may be replaced by t ,θm Or in addition to this, for example, vehicle-related physical quantities such as speed are used as the state quantity Z of the estimated and observed object. t , Z m In the estimation device 1 of the modified example, instead of using the weighted calculation using the Kalman filter, the dynamic parameter p at the previous time t-1 may be used. t-1 The addition of the change amount δp is used to correct the dynamic parameter p in the parameter correction block 140 and S103. t .

[0111] In the estimation device 1 of the modified example, the error compensation block 160 and S104 may also be implemented to compensate the estimated state quantity Z of the DR block 100 and S101 based on the state quantity estimated by a principle different from that of the MM block 120 and S102. t In the estimation device 1 of the modified example, the error compensation block 160 and S104 may be omitted, and the estimated state quantity Z of the DR block 100 and S101 may be directly output. t as a determined value.

Claims

1. An estimation device, which is an estimation device for estimating a state quantity including a position of a vehicle, wherein: have: a dead reckoning unit that estimates the state quantity by dead reckoning based on a dynamics model and internal boundary information, wherein the dynamics model includes parameters that change the behavior of the vehicle, and the internal boundary information is obtained from the internal boundary of the vehicle; a map matching unit that observes the state quantity by map matching based on map information and external information, wherein the map information represents a driving environment of the vehicle and the external information is obtained from outside the vehicle; and a parameter correction unit that corrects the parameter fed back to the dead reckoning based on an amount of deviation of the state amount estimated by the dead reckoning from the state amount observed by the map matching, The parameter correction unit predicts a deviation amount of the state amount estimated by the dead reckoning according to a variation amount of the parameter, among the offset amounts, and corrects the parameter by the variation amount corresponding to the predicted deviation amount. The parameter correction unit predicts the deviation amount generated according to the variation amount of the parameter as a variation amount function δz based on at least one of the following formulas: p : δz p =ΔZ(p t )-ΔZ(p t-1 )、 δz p =ΔZ(p t-1 +δp)-ΔZ(p t-1 ),as well as δz p =ΔZ(δp), in, p t represents the parameter at time t, p t-1 The parameter representing the time t-1 before the time t. δp represents the parameter p at time t-1 t-1 and the parameter p at time t t The amount of change between .

2. The estimation device according to claim 1, wherein: The parameter correction unit predicts the deviation amount that minimizes the difference from the offset amount.

3. The estimation device according to claim 1 or 2, wherein: The parameter correction unit corrects the parameter that changes based on an important factor other than the behavior.

4. The estimation device according to claim 3, wherein: The parameter correction unit corrects the parameters including a coefficient of dynamic friction of a road surface on which the vehicle is traveling.

5. The estimation device according to claim 1 or 2, wherein: The estimation device further includes an error compensation unit configured to compensate for an error in the state quantity estimated by the dead reckoning by filtering based on the state quantity observed by the map matching.

6. An estimation method, which is a method for estimating a state quantity including a position of a vehicle, wherein: include: A dead reckoning step of estimating the state quantity by dead reckoning based on a dynamics model and internal boundary information, wherein the dynamics model includes parameters that change the behavior of the vehicle, and the internal boundary information is obtained from the internal boundary of the vehicle; a map matching step of observing the state quantity by map matching based on map information and external information, wherein the map information represents the driving environment of the vehicle and the external information is obtained from outside the vehicle; and a parameter correction step of correcting the parameter fed back to the dead reckoning based on an amount of deviation of the state amount estimated by the dead reckoning relative to the state amount observed by the map matching, The parameter correction step predicts a deviation amount of the state amount estimated by the dead reckoning according to a variation amount of the parameter, among the offset amounts, and corrects the parameter by the variation amount corresponding to the predicted deviation amount. The parameter correction step predicts the deviation amount generated according to the variation amount of the parameter as a variation amount function δz based on at least one of the following formulas: p : δz p =ΔZ(p t )-ΔZ(p t-1 )、 δz p =ΔZ(p t-1 +δp)-ΔZ(p t-1 ),as well as δz p =ΔZ(δp), in, p t represents the parameter at time t, p t-1 The parameter representing the time t-1 before the time t. δp represents the parameter p at time t-1 t-1 and the parameter p at time t t The amount of change between .

7. The estimation method according to claim 6, wherein: The parameter correction step predicts the deviation amount that minimizes the difference from the offset amount.

8. The estimation method according to claim 6 or 7, wherein: The parameter correction step corrects the parameter that changes due to important factors other than the behavior.

9. The estimation method according to claim 8, wherein: The parameter correction step corrects the parameters including the coefficient of dynamic friction of the road surface on which the vehicle is traveling.

10. The estimation method according to claim 6 or 7, wherein: The estimation method further includes an error compensation step of compensating for an error of the state quantity estimated by the dead reckoning by filtering based on the state quantity observed by the map matching.

11. A storage medium storing an estimation program, the estimation program including instructions for a processor to execute in order to estimate a state quantity including a position of a vehicle, wherein: The above command includes: A dead reckoning step of estimating the state quantity by dead reckoning based on a dynamics model and internal boundary information, wherein the dynamics model includes parameters that change the behavior of the vehicle, and the internal boundary information is obtained from the internal boundary of the vehicle; a map matching step of observing the state quantity by map matching based on map information and external information, wherein the map information represents the driving environment of the vehicle and the external information is obtained from outside the vehicle; and a parameter correction step of correcting the parameter fed back to the dead reckoning based on an amount of deviation of the state amount estimated by the dead reckoning relative to the state amount observed by the map matching, The parameter correction step predicts a deviation amount of the state amount estimated by the dead reckoning according to a variation amount of the parameter, among the offset amounts, and corrects the parameter by the variation amount corresponding to the predicted deviation amount. The parameter correction step predicts the deviation amount generated according to the variation amount of the parameter as a variation amount function δz based on at least one of the following formulas: p : δz p =ΔZ(p t )-ΔZ(p t-1 )、 δz p =ΔZ(p t-1 +δp)-ΔZ(p t-1 ),as well as δz p =ΔZ(δp), in, p t represents the parameter at time t, p t-1 The parameter representing the time t-1 before the time t. δp represents the parameter p at time t-1 t-1 and the parameter p at time t t The amount of change between .

12. The storage medium according to claim 11, wherein: The parameter correction step predicts the deviation amount that minimizes the difference from the offset amount.

13. The storage medium according to claim 11 or 12, wherein: The parameter correction step corrects the parameter that changes due to important factors other than the behavior.

14. The storage medium according to claim 13, wherein: The parameter correction step corrects the parameters including the coefficient of dynamic friction of the road surface on which the vehicle is traveling.

15. The storage medium according to claim 11 or 12, wherein: The command further includes an error compensation step of compensating for an error of the state quantity estimated by the dead reckoning by filtering based on the state quantity observed by the map matching.

Images