(19)
(11)EP 3 726 428 A1

(12)EUROPEAN PATENT APPLICATION
published in accordance with Art. 153(4) EPC

(43)Date of publication:
21.10.2020 Bulletin 2020/43

(21)Application number: 18888791.3

(22)Date of filing:  12.12.2018
(51)International Patent Classification (IPC): 
G06K 9/62(2006.01)
(86)International application number:
PCT/CN2018/120627
(87)International publication number:
WO 2019/114757 (20.06.2019 Gazette  2019/25)
(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR
Designated Extension States:
BA ME
Designated Validation States:
KH MA MD TN

(30)Priority: 15.12.2017 CN 201711346317

(71)Applicant: Nio Nextev Limited
Hong Kong (CN)

(72)Inventor:
  • PENG, Siwei
    Hong Kong (CN)

(74)Representative: Vossius & Partner Patentanwälte Rechtsanwälte mbB 
Siebertstrasse 3
81675 München
81675 München (DE)

  


(54)OPTIMIZATION METHOD AND APPARATUS FOR MULTI-SENSOR TARGET INFORMATION FUSION, COMPUTER DEVICE, AND RECORDING MEDIUM


(57) An optimization method and apparatus for multi-sensor target information fusion, a computer device, and a recording medium. The optimization method for multi-sensor target information fusion comprises the steps of: obtaining, for each time, a fusion prediction result for all sensors concerning a target state at the current time; obtaining actual measurement results from each sensor concerning the target state at the current time; obtaining, for each set of actual measurement results, an optimal estimation result for a corresponding sensor concerning the target state at the current time based on the fusion prediction result and the set of actual measurement results; and fusing optimal estimation results for all the sensors to determine a corresponding weight of each of the optimal estimation results and thus to obtain an optimal fusion estimation result concerning the target state at the current time.




Description

Technical Field



[0001] The invention relates to the field of automobiles, and more particularly to an optimization method and apparatus for multi-sensor target information fusion, a computer device, and a recording medium.

Background Art



[0002] Currently, in the field of automobiles around the world, "intelligentization" is a clear development direction. The National Highway Traffic Safety Administration (NHTSA) has also given a standard for levels of automobile driving, from auxiliary driving to automated driving, but whether it is the auxiliary driving or the automated driving, it is necessary to fuse results from target sensing sensors, so as to reduce overlapped targets and compensate for deficiencies of the results from different sensors.

[0003] However, when it comes to the fusion of sensor results related to a target, this will involve the issue of weight assignment among the results from multiple sensors used for sensing the target. Such fusion requires processing of a large number of sensor results in real time and filtering of these results to provide a decision module with a stably and linearly varying target group, and thus usually involves the consumption of a large amount of processing power, and a conventional embedded processor chip has a relatively low processing performance, which result in the assignment of a fusion weight being artificially determined only according to characteristics of the sensors and the maximum error of the results when such a fusion process is carried out by means of the conventional embedded processor chip. The fusion weight obtained in this way is a weight defined in advance, and therefore, especially in the case of a sensor jump, there are problems that the weight assignment is irrational and is not universally applicable, and then an accurate target fusion estimation result cannot be obtained.

Summary of the Invention



[0004] In order to overcome one or more of the above-mentioned shortcomings or other shortcomings, the invention is implemented by using the technical solutions as follows.

[0005] According to one aspect of the invention, an optimization method for multi-sensor target information fusion is provided, which comprises: step S1: obtaining, for each time, a fusion prediction result for all sensors concerning a target state at the current time; step S2: obtaining actual measurement results from each sensor concerning the target state at the current time; step S3: obtaining, for each set of actual measurement results, an optimal estimation result for a corresponding sensor concerning the target state at the current time based on the fusion prediction result and the set of actual measurement results; and step S4: fusing optimal estimation results for all the sensors to determine a corresponding weight of each of the optimal estimation results and thus to obtain an optimal fusion estimation result concerning the target state at the current time.

[0006] Further, in one aspect of the invention, step S3 comprises: step S31: calculating, for each set of actual measurement results, a corresponding transformation matrix based on the fusion prediction result and the set of actual measurement results; step S32: calculating a corresponding covariance of each set of actual measurement results; step S33: calculating, for each set of actual measurement results, a corresponding Kalman gain based on the corresponding transformation matrix and the corresponding covariance; and step S34: calculating, for each set of actual measurement results, the corresponding optimal estimation result for the corresponding sensor concerning the target state at the current time based on the fusion prediction result, the corresponding Kalman gain, the set of actual measurement results, and the corresponding transformation matrix.

[0007] Further, in one aspect of the invention, step S4 comprises: step S41: determining the corresponding weight of each of the optimal estimation results according to the corresponding covariance of each set of actual measurement results; and step S42: calculating the optimal fusion estimation result for all the sensors concerning the target state at the current time based on each of the optimal estimation results and the corresponding weight.

[0008] Further, in one aspect of the invention, the method also comprises: step S5: correcting the covariance obtained in step S32 according to the transformation matrix obtained in step S31 and the Kalman gain obtained in step S33, so as to obtain a corrected covariance.

[0009] Further, in one aspect of the invention, in step S32, the corresponding covariance of each set of actual measurement results at the current time is obtained with a corrected covariance obtained at a previous time.

[0010] Further, in one aspect of the invention, in step S1, the fusion prediction result at the current time is obtained with an optimal fusion estimation result concerning the target state at the previous time.

[0011] According to another aspect of the invention, an optimization apparatus for multi-sensor target information fusion is provided, which is characterized by comprising: unit 1 for obtaining, for each time, a fusion prediction result for all sensors concerning a target state at the current time; unit 2 for obtaining actual measurement results from each sensor concerning the target state at the current time; unit 3 for obtaining, for each set of actual measurement results, an optimal estimation result for a corresponding sensor concerning the target state at the current time based on the fusion prediction result and the set of actual measurement results; and unit 4 for fusing optimal estimation results for all the sensors to determine a corresponding weight of each of the optimal estimation results and thus to obtain an optimal fusion estimation result concerning the target state at the current time.

[0012] Further, in another aspect of the invention, unit 3 comprises: unit 3A for calculating, for each set of actual measurement results, a corresponding transformation matrix based on the fusion prediction result and the set of actual measurement results; unit 3B for calculating a corresponding covariance of each set of actual measurement results; unit 3C for calculating, for each set of actual measurement results, a corresponding Kalman gain based on the corresponding transformation matrix and the corresponding covariance; and unit 3D for calculating, for each set of actual measurement results, the corresponding optimal estimation result for the corresponding sensor concerning the target state at the current time based on the fusion prediction result, the corresponding Kalman gain, the set of actual measurement results, and the corresponding transformation matrix.

[0013] Further, in another aspect of the invention, unit 4 comprises: unit 4A for determining the corresponding weight of each of the optimal estimation results according to the corresponding covariance of each set of actual measurement results; and unit 4B for calculating the optimal fusion estimation result for all the sensors concerning the target state at the current time based on each of the optimal estimation results and the corresponding weight.

[0014] Further, in another aspect of the invention, the optimization apparatus also comprises: unit 5 for correcting the covariance obtained in unit 3B according to the transformation matrix obtained in unit 3A and the Kalman gain obtained in unit 3C, so as to obtain a corrected covariance.

[0015] Further, in another aspect of the invention, in unit 3B, the corresponding covariance of each set of actual measurement results at the current time is obtained with a corrected covariance obtained at a previous time.

[0016] Further, in another aspect of the invention, in unit 1, the fusion prediction result at the current time is obtained with an optimal fusion estimation result concerning the target state at the previous time.

[0017] According to a further aspect of the invention, a computer device is provided, which comprises a memory, a processor, and a computer program stored on the memory and operable on the processor, wherein the program implements, when executed by the processor, the steps of the method according to one aspect of the invention.

[0018] According to still another aspect of the invention, a recording medium having a computer program stored thereon is provided, wherein the program is executed by a computer to implement the steps of the method according to one aspect of the invention.

[0019] According to yet another aspect of the invention, an auxiliary driving method is provided, which comprises an optimization method for multi-sensor target information fusion according to one aspect of the invention.

[0020] According to still yet another aspect of the invention, an auxiliary driving system is provided, which comprises an optimization apparatus for multi-sensor target information fusion according to another aspect of the invention.

[0021] Compared with the prior art, the invention can obtain one or more of the beneficial effects as follows:
  1. (1) according to the invention, a fusion weight can be calculated in real time to ensure that the result of weight assignment is appropriate in most scenarios; and
  2. (2) according to the invention, Kalman filtering is applied to each set of measurement results from each sensor to obtain a corresponding covariance and optimal estimation result for all the measurement results, and then an optimal fusion estimation result concerning a target state is calculated, so that each corresponding weight of the optimal estimation result can be determined in conjunction with the covariance of the measurement results from the sensor. Therefore, even if the sensor is affected by hardware performance and environment and a detection result does not conform to a physical motion principle, the weight can be adjusted in real time to realize the optimal fusion estimation of the target and to improve the performance of fusion estimation.

Brief Description of the Drawings



[0022] 

FIG. 1 is an example flowchart of an optimization method for multi-sensor target information fusion according to one implementation of the invention.

FIG. 2 is an example sub-flowchart of step S3 in FIG. 1 according to one embodiment of the invention.

FIG. 3 is an example sub-flowchart of step S4 in FIG. 1 according to one embodiment of the invention.

FIG. 4 is an example block diagram of an optimization apparatus for multi-sensor target information fusion according to one implementation of the invention.

FIG. 5 is an example block diagram of a computer device according to one implementation of the invention that is configured to perform an optimization method for multi-sensor target information fusion according to one implementation of the invention.


Detailed Description of Embodiments



[0023] An optimization method and apparatus for multi-sensor target information fusion, a computer device, and a recording medium according to the invention will be further described in detail below in conjunction with the accompanying drawings. It should be noted that the following detailed description of embodiments is exemplary rather than limiting, and is intended to provide a basic understanding of the invention, but not to confirm key or decisive elements of the invention or limit the scope of protection.

[0024] The invention is described below with reference to block diagram illustrations, block diagrams and/or flowcharts of the method and apparatus in the embodiments of the invention. It should be understood that each block of the flowchart illustrations and/or the block diagrams and combinations of the flowchart illustrations and/or the block diagrams may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing devices to form a machine, such that the instructions, which are executed by the processor of the computer or the other programmable data processing devices, create components for implementing functions/operations specified in the flowcharts and/or blocks and/or one or more flow block diagrams.

[0025] These computer program instructions may be stored in a computer-readable memory and may instruct a computer or other programmable processors to implement functions in a specific manner, so that these instructions stored in the computer-readable memory constitute an article of manufacture, which contains instruction components for implementing the functions/operations specified in one or more blocks of the flowcharts and/or block diagrams.

[0026] These computer program instructions may be loaded onto a computer or other programmable data processors to enable a series of operational steps to be executed on the computer or the other programmable processors, so as to constitute a computer-implemented process, so that these instructions executed on the computer or the other programmable data processors provide steps for implementing the functions or operations specified in one or more blocks of the flowcharts and/or block diagrams. It should also be noted that in some alternative implementations, the functions/operations shown in the blocks may occur in an order different from that shown in the flowchart. For example, two blocks shown in sequence may actually be executed substantially simultaneously or may sometimes be executed in a reverse order, which depends on the functions/operations involved.

[0027] The optimization method and apparatus for multi-sensor target information fusion according to the invention can be applied, for example, in a scenario where targets around a vehicle are sensed. In such a scenario, for example, a motion state of any one of targets around the vehicle can be characterized by a relative longitudinal position to the vehicle, a longitudinal speed, a longitudinal acceleration, a relative transverse position to the vehicle, a transverse speed, and a transverse acceleration, and each set of measurement results sensed by each of multiple sensors on the vehicle is related to the six items above or can be calculated to obtain values of the six items. Where there are differences in terms of the characteristic of each sensor and the measurement error for each item, the optimization method and apparatus for multi-sensor target information fusion according to the invention that are described in detail below can be used to obtain an optimal fusion result for any target and rationally determine a fusion weight.

[0028] FIG. 1 is an example flowchart of an optimization method for multi-sensor target information fusion according to one implementation of the invention. As shown in FIG. 1, method S100 comprises the step of: obtaining, for each time, a fusion prediction result for all sensors concerning a target state at the current time (step S1).

[0029] In one example, for a time t, the following equation (1) can be used to determine a fusion prediction result concerning the target state of a target:

where (t) is a fusion prediction result for all sensors concerning the target state at the time t, F is a system state transition matrix, X(t - 1) is an optimal fusion estimation result concerning the target state at a time t-1 (described later), and W(t) is a system noise.

[0030] In one embodiment, as shown in FIG. 1, method S100 may further comprise the step of: obtaining actual measurement results from each sensor concerning the target state at the current time (step S2).

[0031] In one embodiment, as shown in FIG. 1, method S100 may further comprise the step of: obtaining, for each set of actual measurement results, an optimal estimation result for a corresponding sensor concerning the target state at the current time based on the fusion prediction result and the set of actual measurement results (step S3).

[0032] The details of step S3 will be described in detail in conjunction with FIG. 2.

[0033] Specifically, in one embodiment, as shown in FIG. 2, step S3 comprises the step of: calculating, for each set of actual measurement results, a corresponding transformation matrix based on the fusion prediction result and the set of actual measurement results (step S31).

[0034] In one example, for each set of actual measurement results obtained by each sensor, the following equation (2) can be used to determine the corresponding transformation matrix:

where Zik(t) is a kth set of actual measurement results from an ith sensor concerning the target state at the time t, Hik is a corresponding transformation matrix of the kth set of actual measurement results from the ith sensor, X(t) is the fusion prediction result for all the sensors concerning the target state at the time t, and V(t) is a measurement noise.

[0035] In one embodiment, as shown in FIG. 2, step S3 may further comprise the step of: calculating a corresponding covariance of each set of actual measurement results (step S32).

[0036] In one example, for each set of actual measurement results obtained by each sensor, the following equation (3) can be used to determine the corresponding covariance:

where ik(t) is a corresponding covariance of the kth set of actual measurement results from the ith sensor concerning the target state at the time t, F is the system state transition matrix, FT is a transposed matrix of the system state transition matrix, Pik(t - 1) is a corresponding covariance of an optimal estimation result of the kth set of actual measurement results from the ith sensor at the time t-1 (described later), and Q is the covariance of a system process noise.

[0037] In one embodiment, as shown in FIG. 2, step S3 may further comprise the step of: calculating, for each set of actual measurement results, a corresponding Kalman gain based on the corresponding transformation matrix and the corresponding covariance (step S33).

[0038] In one example, for each set of actual measurement results obtained by each sensor, the following equation (4) can be used to determine the corresponding Kalman gain:

where kgik(t) is a corresponding Kalman gain of the kth set of actual measurement results from the ith sensor concerning the target state at the time t, ik(t) is the corresponding covariance of the kth set of actual measurement results from the ith sensor concerning the target state at the time t, Hik is the corresponding transformation matrix of the kth set of actual measurement results from the ith sensor, HikT is a transposed matrix of the corresponding transformation matrix of the kth set of actual measurement results from the ith sensor, and R is the covariance of a measurement process noise.

[0039] In one embodiment, as shown in FIG. 2, step S3 may further comprise the step of: calculating, for each set of actual measurement results, the corresponding optimal estimation result for the corresponding sensor concerning the target state at the current time based on the fusion prediction result, the corresponding Kalman gain, the set of actual measurement results, and the corresponding transformation matrix (step S34).

[0040] In one example, for each set of actual measurement results obtained by each sensor, the following equation (5) can be used to determine the corresponding optimal estimation result:

where Xik(t) is a corresponding optimal estimation result of the kth set of actual measurement results from the ith sensor at the time t, (t) is the fusion prediction result for all the sensors concerning the target state at the time t, kgik(t) is the corresponding Kalman gain of the kth set of actual measurement results from the ith sensor concerning the target state at the time t, Zik(t) is the kth set of actual measurement results from the ith sensor concerning the target state at the time t, and Hik is the corresponding transformation matrix of the kth set of actual measurement results from the ith sensor.

[0041] After the optimal estimation result for the corresponding sensor concerning the target state at the current time is calculated for each set of actual measurement results, the process returns to FIG. 1. In one embodiment, as shown in FIG. 1, method S 100 may further comprise the step of: fusing optimal estimation results for all the sensors to determine a corresponding weight of each of the optimal estimation results and thus to obtain an optimal fusion estimation result concerning the target state at the current time (step S4).

[0042] In one example, the following equation (6) can be used to obtain the optimal fusion estimation result concerning the target state at the current time:

where X(t) is an optimal fusion estimation result for all the sensors concerning the target state at the time t, f is a fusion function, Xik(t) is the corresponding optimal estimation result of the kth set of actual measurement results from the ith sensor at the time t, and ik(t) is the corresponding covariance of the kth set of actual measurement results from the ith sensor concerning the target state at the time t.

[0043] The details of step S4 will be described in detail in conjunction with FIG. 3.

[0044] Specifically, in one embodiment, as shown in FIG. 3, step S4 comprises the step of: determining the corresponding weight of each of the optimal estimation results according to the corresponding covariance of each set of actual measurement results (step S41).

[0045] In one example, after the equation (3) is used to calculate corresponding covariances (i.e., 11(t),12(t),...,ik(t)) of all sets of actual measurement results from all the sensors, a weight (i.e., w11(t),w12(t),...,wik(t)) is assigned to each corresponding optimal estimation result (i.e., X11(t),X12(t),...,Xik(t)) of each set of actual measurement results according to the magnitude of the covariances.

[0046] In one embodiment, as shown in FIG. 3, step S4 may further comprise the step of: calculating the optimal fusion estimation result for all the sensors concerning the target state at the current time based on each of the optimal estimation results and the corresponding weight (step S42).

[0047] In one example, the assigned weights (i.e., w11(t),w12(t),...,wik(t)) are used to perform a weighted operation on the corresponding optimal estimation results (i.e., X11(t),X12(t),...,Xik(t)), so as to obtain the optimal fusion estimation result for all the sensors concerning the target state at the time t. In addition, as shown in the equation (1), X(t) at the time t can also be used to calculate (t+1) at a time t + 1.

[0048] Optionally, in one embodiment, method S100 may further comprise the step of: correcting the covariance obtained in step S32 according to the transformation matrix obtained in step S31 and the Kalman gain obtained in step S33, so as to obtain a corrected covariance (step S5, not shown), wherein the corrected covariance can be used to calculate a corresponding covariance of corresponding actual measurement results at the next time of the current time (see equation (3) above).

[0049] In one example, the following equation (7) is used to obtain the corrected covariance at the current time:

where Pik(t) is a corrected covariance of the kth set of actual measurement results from the ith sensor at the time t, I is an identity matrix, kgik(t) is the corresponding Kalman gain of the kth set of actual measurement results from the ith sensor concerning the target state at the time t that is obtained in step S33, Hik is the corresponding transformation matrix of the kth set of actual measurement results from the ith sensor that is obtained in step S31, and ik(t) is the corresponding covariance of the kth set of actual measurement results from the ith sensor concerning the target state at the time t that is obtained in step S32, and in addition, Pik(t) at the time t can also be used to calculate ik(t+1) at the time t + 1.

[0050] Weights can be adjusted in real time by means of the steps above, so as to obtain an accurate target fusion estimation result.

[0051] In addition, it should be noted that although the sequence between steps is shown in FIGs. 1, 2 and 3, those skilled in the art should understand that FIGs. 1, 2 and 3 are merely examples, and the sequential relationship between the steps is not limited to the case shown in FIGs. 1, 2 and 3. For example, step S1 in FIG. 1 may be performed after step S2, or the two steps may be performed simultaneously. For another example, step S31 in FIG. 2 may be performed after step S32, or the two steps may be performed simultaneously, etc.

[0052] Next, the optimization apparatus for multi-sensor target information fusion for performing the method shown in FIG. 1 is described with reference to FIG. 4.

[0053] As shown in FIG. 4, the apparatus 100 comprises unit 1 101, which is configured to obtain, for each time, a fusion prediction result for all sensors concerning a target state at the current time.

[0054] In one example, for a time t, the following equation (8) can be used to determine a fusion prediction result concerning the target state of a target:

where (t) is a fusion prediction result for all sensors concerning the target state at the time t, F is a system state transition matrix, X(t - 1) is an optimal fusion estimation result concerning the target state at a time t-1 (described later), and W(t) is a system noise.

[0055] In one embodiment, as shown in FIG. 4, the apparatus 100 may further comprise unit 2, which is configured to obtain actual measurement results from each sensor concerning the target state at the current time.

[0056] In one embodiment, as shown in FIG. 4, the apparatus 100 may further comprise unit 3, which is configured to obtain, for each set of actual measurement results, an optimal estimation result for a corresponding sensor concerning the target state at the current time based on the fusion prediction result and the set of actual measurement results.

[0057] The internal structure of unit 3 103 will be described below in detail.

[0058] Specifically, in one embodiment, unit 3 103 comprises unit 3A (not shown), which is configured to calculate, for each set of actual measurement results, a corresponding transformation matrix based on the fusion prediction result and the set of actual measurement results.

[0059] In one example, for each set of actual measurement results obtained by each sensor, the following equation (9) can be used to determine the corresponding transformation matrix:

where Zik(t) is a kth set of actual measurement results from an ith sensor concerning the target state at the time t, Hik is a corresponding transformation matrix of the kth set of actual measurement results from the ith sensor, (t) is the fusion prediction result for all the sensors concerning the target state at the time t, and V(t) is a measurement noise.

[0060] In one embodiment, unit 3 103 may further comprise unit 3B (not shown), which is configured to calculate a corresponding covariance of each set of actual measurement results.

[0061] In one example, for each set of actual measurement results obtained by each sensor, the following equation (10) can be used to determine the corresponding covariance:

where ik(t) is a corresponding covariance of the kth set of actual measurement results from the ith sensor concerning the target state at the time t, F is the system state transition matrix, FT is a transposed matrix of the system state transition matrix, Pik(t - 1) is a corresponding covariance of an optimal estimation result of the kth set of actual measurement results from the ith sensor at the time t-1 (described later), and Q is the covariance of a system process noise.

[0062] In one embodiment, unit 3 103 may further comprise unit 3C (not shown), which is configured to calculate, for each set of actual measurement results, a corresponding Kalman gain based on the corresponding transformation matrix and the corresponding covariance.

[0063] In one example, for each set of actual measurement results obtained by each sensor, the following equation (11) can be used to determine the corresponding Kalman gain:

where kgik(t) is a corresponding Kalman gain of the kth set of actual measurement results from the ith sensor concerning the target state at the time t, ik(t) is the corresponding covariance of the kth set of actual measurement results from the ith sensor concerning the target state at the time t, Hik is the corresponding transformation matrix of the kth set of actual measurement results from the ith sensor, HikT is a transposed matrix of the corresponding transformation matrix of the kth set of actual measurement results from the ith sensor, and R is the covariance of a measurement process noise.

[0064] In one embodiment, unit 3 103 may further comprise unit 3D (not shown), which is configured to calculate, for each set of actual measurement results, the corresponding optimal estimation result for the corresponding sensor concerning the target state at the current time based on the fusion prediction result, the corresponding Kalman gain, the set of actual measurement results, and the corresponding transformation matrix.

[0065] In one example, for each set of actual measurement results obtained by each sensor, the following equation (12) can be used to determine the corresponding optimal estimation result:

where Xik(t) is a corresponding optimal estimation result of the kth set of actual measurement results from the ith sensor at the time t, (t) is the fusion prediction result for all the sensors concerning the target state at the time t, kgik(t) is the corresponding Kalman gain of the kth set of actual measurement results from the ith sensor concerning the target state at the time t, Zik(t) is the kth set of actual measurement results from the ith sensor concerning the target state at the time t, and Hik is the corresponding transformation matrix of the kth set of actual measurement results from the ith sensor.

[0066] After the optimal estimation result for the corresponding sensor concerning the target state at the current time is calculated for each set of actual measurement results, the process returns to FIG. 4. In one embodiment, as shown in FIG. 4, the apparatus 100 may further comprise unit 4 104, which is configured to fuse optimal estimation results for all the sensors to determine a corresponding weight of each of the optimal estimation results and thus to obtain an optimal fusion estimation result concerning the target state at the current time.

[0067] In one example, the following equation (13) can be used to obtain the optimal fusion estimation result concerning the target state at the current time:

where X(t) is an optimal fusion estimation result for all the sensors concerning the target state at the time t, f is a fusion function, Xik(t) is the corresponding optimal estimation result of the kth set of actual measurement results from the ith sensor at the time t, and ik(t) is the corresponding covariance of the kth set of actual measurement results from the ith sensor concerning the target state at the time t. The internal structure of unit 4 104 will be described below in detail.

[0068] Specifically, in one embodiment, unit 4 104 comprises unit 4A (not shown), which is configured to determine the corresponding weight of each of the optimal estimation results according to the corresponding covariance of each set of actual measurement results.

[0069] In one example, after the equation (10) is used to calculate corresponding covariances (i.e., 11(t),12(t),...,ik(t)) of all sets of actual measurement results from all the sensors, a weight (i.e., w11(t),w12(t),...,wik(t)) is assigned to each corresponding optimal estimation result (i.e., X11(t),X12(t),...,Xik(t)) of each set of actual measurement results according to the magnitude of the covariances.

[0070] In one embodiment, unit 4 104 may further comprise unit 4B (not shown), which is configured to calculate the optimal fusion estimation result for all the sensors concerning the target state at the current time based on each of the optimal estimation results and the corresponding weight.

[0071] In one example, the assigned weights (i.e., w11(t),w12(t),...,wik(t)) are used to perform a weighted operation on the corresponding optimal estimation results (i.e., X11(t),X12(t),...,Xik(t)), so as to obtain the optimal fusion estimation result for all the sensors concerning the target state at the time t. In addition, as shown in the equation (8), X(t) at the time t can also be used to calculate (t +1) at a time t + 1.

[0072] Optionally, in one embodiment, the apparatus 100 may further comprise unit 5 (not shown), which is configured to correct the covariance obtained in unit 3B according to the transformation matrix obtained in unit 3A and the Kalman gain obtained in unit 3C, so as to obtain a corrected covariance, wherein the corrected covariance can be used to calculate a corresponding covariance of corresponding actual measurement results at the next time of the current time (see equation (10) above).

[0073] In one example, the following equation (14) is used to obtain the corrected covariance at the current time:

where Pik(t) is a corrected covariance of the kth set of actual measurement results from the ith sensor at the time t, I is an identity matrix, kgik(t) is the corresponding Kalman gain of the kth set of actual measurement results from the ith sensor concerning the target state at the time t that is obtained in step S33, Hik is the corresponding transformation matrix of the kth set of actual measurement results from the ith sensor that is obtained in step S31, and ik(t) is the corresponding covariance of the kth set of actual measurement results from the ith sensor concerning the target state at the time t that is obtained in step S32, and in addition, Pik(t) at the time t can also be used to calculate ik(t + 1) at the time t + 1.

[0074] Weights can be adjusted in real time by means of the units above, so as to obtain an accurate target fusion estimation result.

[0075] When applied to auxiliary driving, the optimization method and apparatus for multi-sensor target information fusion according to one implementation of the invention can enable an auxiliary driving system to employ more optimized data, thereby facilitating its decision-making and control, for example, making, based on the optimized data, a better decision for auxiliary driving functions or scenarios, such as adaptive cruise and emergency braking, with such functions or scenarios further including vehicle stability control and the like.

[0076] Although implementations of the optimization method and apparatus for multi-sensor target information fusion have been mainly described in the invention, the invention is not limited to these implementations, and the invention may be implemented in a way of: an auxiliary driving method comprising the method described above or an auxiliary driving system comprising the apparatus described above, or a computer device for performing the method described above or a computer program for performing the method described above, or a computer program for implementing functions of the apparatus described above or a computer-readable recording medium having the computer program recorded thereon.

[0077] FIG. 5 shows a computer device according to one implementation of the invention that is configured to perform an optimization method for multi-sensor target information fusion according to one implementation of the invention. As shown in FIG. 5, a computer device 200 comprises a memory 201 and a processor 202. Although not shown, the computer device 200 further comprises a computer program that is stored on the memory 201 and operable on the processor 202. The program implements, when executed by the processor, the steps, for example as shown in FIGs. 1, 2 and 3, of an optimization method for multi-sensor target information fusion according to one implementation of the invention.

[0078] In addition, as described above, the invention may also be implemented as a recording medium, which stores a program for enabling a computer to perform an optimization method for multi-sensor target information fusion according to one implementation of the invention.

[0079] Here, various types of recording media, such as disks (e.g., a magnetic disk, an optical disk, etc.), cards (e.g., a memory card, an optical card, etc.), semiconductor memories (e.g., a ROM, a non-volatile memory, etc.), and tapes (e.g., a magnetic tape, a cassette tape, etc.), can be used as the recording medium.

[0080] By recording, in these recording media, a computer program that enables a computer to perform the optimization method for multi-sensor target information fusion in the implementations above or a computer program that enables a computer to implement functions of the optimization apparatus for multi-sensor target information fusion in the implementations above and circulating the computer program, costs are reduced, and portability and versatility are improved.

[0081] Furthermore, the recording medium is loaded on a computer, the computer reads the computer program recorded in the recording medium and stores same in a memory, and a processor (central processing unit (CPU) and a micro processing unit (MPU)) provided on the computer reads out and executes the computer program from the memory, whereby the optimization method for multi-sensor target information fusion in the implementations above can be performed, and functions of the optimization apparatus for multi-sensor target information fusion in the implementations above can be implemented.

[0082] Those of ordinary skill in the art should understand that the invention is not limited to the implementations above, and the invention can be implemented in many other forms without departing from the essence and scope thereof. Therefore, the presented examples and implementations are regarded to be schematic rather than restrictive, and without departing from the spirit and scope of the invention that are defined by the appended claims, the invention may cover various changes and replacements.


Claims

1. An optimization method for multi-sensor target information fusion, characterized by comprising:

step S1: obtaining, for each time, a fusion prediction result for all sensors concerning a target state at the current time;

step S2: obtaining actual measurement results from each sensor concerning the target state at the current time;

step S3: obtaining, for each set of actual measurement results, an optimal estimation result for a corresponding sensor concerning the target state at the current time based on the fusion prediction result and the set of actual measurement results; and

step S4: fusing optimal estimation results for all the sensors to determine a corresponding weight of each of the optimal estimation results and thus to obtain an optimal fusion estimation result concerning the target state at the current time.


 
2. The optimization method according to claim 1, characterized in that step S3 comprises:

step S31: calculating, for each set of actual measurement results, a corresponding transformation matrix based on the fusion prediction result and the set of actual measurement results;

step S32: calculating a corresponding covariance of each set of actual measurement results;

step S33: calculating, for each set of actual measurement results, a corresponding Kalman gain based on the corresponding transformation matrix and the corresponding covariance; and

step S34: calculating, for each set of actual measurement results, the corresponding optimal estimation result for the corresponding sensor concerning the target state at the current time based on the fusion prediction result, the corresponding Kalman gain, the set of actual measurement results, and the corresponding transformation matrix.


 
3. The optimization method according to claim 2, characterized in that step S4 comprises:

step S41: determining the corresponding weight of each of the optimal estimation results according to the corresponding covariance of each set of actual measurement results; and

step S42: calculating the optimal fusion estimation result for all the sensors concerning the target state at the current time based on each of the optimal estimation results and the corresponding weight.


 
4. The optimization method according to claim 2, characterized by further comprising:
step S5: correcting the covariance obtained in step S32 according to the transformation matrix obtained in step S31 and the Kalman gain obtained in step S33, so as to obtain a corrected covariance.
 
5. The optimization method according to claim 4, characterized in that in step S32, the corresponding covariance of each set of actual measurement results at the current time is obtained with a corrected covariance obtained at a previous time.
 
6. The method according to any of claims 1 to 5, characterized in that in step S1, the fusion prediction result at the current time is obtained with an optimal fusion estimation result concerning the target state at the previous time.
 
7. An optimization apparatus for multi-sensor target information fusion, characterized by comprising:

unit 1 for obtaining, for each time, a fusion prediction result for all sensors concerning a target state at the current time;

unit 2 for obtaining actual measurement results from each sensor concerning the target state at the current time;

unit 3 for obtaining, for each set of actual measurement results, an optimal estimation result for a corresponding sensor concerning the target state at the current time based on the fusion prediction result and the set of actual measurement results; and

unit 4 for fusing optimal estimation results for all the sensors to determine a corresponding weight of each of the optimal estimation results and thus to obtain an optimal fusion estimation result concerning the target state at the current time.


 
8. The optimization apparatus according to claim 7, characterized in that unit 3 comprises:

unit 3A for calculating, for each set of actual measurement results, a corresponding transformation matrix based on the fusion prediction result and the set of actual measurement results;

unit 3B for calculating a corresponding covariance of each set of actual measurement results;

unit 3C for calculating, for each set of actual measurement results, a corresponding Kalman gain based on the corresponding transformation matrix and the corresponding covariance; and

unit 3D for calculating, for each set of actual measurement results, the corresponding optimal estimation result for the corresponding sensor concerning the target state at the current time based on the fusion prediction result, the corresponding Kalman gain, the set of actual measurement results, and the corresponding transformation matrix.


 
9. The optimization apparatus according to claim 8, characterized in that unit 4 comprises:

unit 4A for determining the corresponding weight of each of the optimal estimation results according to the corresponding covariance of each set of actual measurement results; and

unit 4B for calculating the optimal fusion estimation result for all the sensors concerning the target state at the current time based on each of the optimal estimation results and the corresponding weight.


 
10. The optimization apparatus according to claim 8, characterized by further comprising:
unit 5 for correcting the covariance obtained in unit 3B according to the transformation matrix obtained in unit 3A and the Kalman gain obtained in unit 3C, so as to obtain a corrected covariance.
 
11. The optimization apparatus according to claim 10, characterized in that in unit 3B, the corresponding covariance of each set of actual measurement results at the current time is obtained with a corrected covariance obtained at a previous time.
 
12. The optimization apparatus according to any of claims 7 to 11, characterized in that in unit 1, the fusion prediction result at the current time is obtained with an optimal fusion estimation result concerning the target state at the previous time.
 
13. A computer device, comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, characterized in that the program implements, when executed by the processor, the steps of the method according to any of claims 1 to 6.
 
14. A recording medium having a computer program stored thereon, characterized in that the program is executed by a computer to implement the steps of the method according to any of claims 1 to 6.
 
15. An auxiliary driving method, characterized by comprising an optimization method for multi-sensor target information fusion according to any of claims 1 to 6.
 
16. An auxiliary driving system, characterized by comprising an optimization apparatus for multi-sensor target information fusion according to any of claims 7 to 12.
 




Drawing













Search report