Multi-sensor fusion target positioning method, electronic equipment and storage medium

Abstract

The invention provides a multi-sensor fusion target positioning method, electronic equipment and a storage medium, and the method comprises the steps: obtaining the measurement information of a current moment and the track information of a previous moment, determining the track prediction information of the current moment according to the track information, carrying out the matching of the measurement information and the track prediction information, obtaining a track measurement matching pair, and carrying out the positioning of a target. For each track measurement matching pair, a pseudo measurement value is constructed based on the track prediction value and the measurement value in the track measurement matching pair, and the track value of the corresponding target at the current moment is determined according to the pseudo measurement value and the track prediction value, so that target positioning is realized, the final positioning result can be closer to the real position of the target, and the positioning accuracy is improved. The problems of track splitting, jittering, pulling and the like caused by large measurement difference between sensors during fusion are solved, so that the stability of the driving function and the driving safety are ensured.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a multi-sensor fusion target positioning method, electronic device, and storage medium. Background Art

[0002] In autonomous driving systems, multi-sensor data fusion is a key technology for improving target detection and tracking accuracy. Radar and cameras, as common sensors, each have distinct advantages and limitations. Radar performs well at long distances and in adverse weather conditions, generally achieving high accuracy in longitudinal position and velocity measurements, but exhibiting lower accuracy in lateral distance measurement and velocity measurement. Cameras, on the other hand, excel at close-range recognition, with better lateral accuracy than longitudinal at medium and long distances. However, they are susceptible to factors such as vehicle pitch, non-coplanarity, and target irregularities, leading to unstable ranging.

[0003] However, when using multi-source sensor fusion, since each sensor is independent and the measurement results are affected by the environment, there is uncertainty. The ranging gap between the camera and radar is large, including deviations and measurement errors, which may cause the final fused track to split, jitter, and pull, ultimately affecting the stability of the driving function. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, the present application aims to provide a multi-sensor fusion target positioning method, electronic device and storage medium to solve the problems of track splitting, jittering, pulling, etc. caused by large measurement gaps between sensors, and ensure the stability of driving functions.

[0005] The present invention provides a multi-sensor fusion target positioning method, which includes:

[0006] Acquire measurement information at a current moment and track information at a previous moment, wherein the measurement information includes a measurement value of each target, and the track information includes a track value of each target;

[0007] Determining track prediction information at a current moment based on the track information, wherein the track prediction information includes track prediction values of each target;

[0008] Matching the measurement information with the track prediction information to obtain a track measurement matching pair;

[0009] For each track measurement matching pair, a pseudo measurement value of the corresponding target is constructed based on the track prediction value and the measurement value in the track measurement matching pair, and the track value of the corresponding target at the current moment is determined according to the pseudo measurement value and the track prediction value.

[0010] Optionally, matching the measurement information with the track prediction information to obtain a track measurement matching pair includes:

[0011] Combining the measurement values in the measurement information with the track prediction values in the track prediction information in pairs to obtain track measurement pairs;

[0012] For each track measurement pair, construct a temporary pseudo measurement value based on the track prediction value and the measurement value in the track measurement pair, and determine the associated cost of the track measurement pair according to the temporary pseudo measurement value and the track prediction value;

[0013] A track measurement matching pair is determined according to the associated cost of each track measurement pair.

[0014] Optionally, constructing a temporary pseudo-measurement value based on the track prediction value and the measured value in the track measurement pair includes:

[0015] Determining an estimated track gain parameter of the target at the current moment based on the track gain parameter of the target corresponding to the track prediction value at the previous moment;

[0016] Determining a measurement gain parameter based on a track prediction value and a measurement value in the track measurement pair;

[0017] determining a temporary gain parameter of the track-measurement pair based on the estimated track gain parameter and the measurement gain parameter;

[0018] A temporary pseudo measurement value is determined according to the temporary gain parameter and the measurement value of the track measurement pair.

[0019] Optionally, determining a temporary gain parameter of the track-measurement pair based on the estimated track gain parameter and the measurement gain parameter includes:

[0020] If the sensor corresponding to the track measurement centering measurement value has not participated in track updating, determining the temporary gain parameter based on the measurement gain parameter;

[0021] If the sensor corresponding to the measurement value of the track measurement has participated in track update, and the target corresponding to the measurement value is different from the target updated by the sensor, then the temporary gain parameter is set to a preset value;

[0022] If the sensor corresponding to the measurement value in the track measurement has participated in track update, and the target corresponding to the measurement value is the same as the target updated by the sensor, the temporary gain parameter is determined based on the estimated track gain parameter and the measurement gain parameter.

[0023] Optionally, determining an association cost of the track measurement pair according to the temporary pseudo-measurement value and the track prediction value includes:

[0024] determining a speed-associated cost based on the target speed in the temporary pseudo-measurement value and the target speed in the track prediction value;

[0025] determining a position association cost based on the target position in the temporary pseudo-measurement value and the target position in the track prediction value;

[0026] An association cost of the track measurement pair is determined according to the speed association cost and the position association cost.

[0027] Optionally, constructing a pseudo measurement value of the corresponding target based on the track prediction value and the measurement value in the track measurement matching pair includes:

[0028] determining a measurement gain parameter based on a track prediction value and a measurement value in the track measurement matching pair;

[0029] Determining a track gain parameter of the target at the current moment based on the track gain parameter of the target corresponding to the track prediction value at the previous moment and the measurement gain parameter;

[0030] A pseudo measurement value of the target is determined based on the measurement values in the track measurement matching pair and a track gain parameter at a current moment.

[0031] Optionally, determining the track gain parameter of the target at the current moment based on the track gain parameter of the target corresponding to the track prediction value at the previous moment and the measurement gain parameter includes:

[0032] If the target corresponding to the track prediction value in the track measurement matching pair is updated for the first time by the sensor corresponding to the measurement value, or the target corresponding to the track prediction value has not been updated by the sensor within a preset number of consecutive times, the track gain parameter of the target at the current moment is determined based on the measurement gain parameter; otherwise, the track gain parameter of the target at the current moment is determined based on the track gain parameter of the target at the previous moment and the measurement gain parameter.

[0033] Optionally, determining the pseudo measurement value of the target based on the measurement value in the track measurement matching pair and the track gain parameter at the current moment includes:

[0034] determining a gain parameter range based on a type of target corresponding to a track prediction value in the track measurement matching pair;

[0035] It is determined whether the track gain parameter at the current moment is within the gain parameter range; if not, the track gain parameter at the current moment is updated based on the gain parameter range.

[0036] An embodiment of the present application further provides an electronic device, comprising:

[0037] processor and memory;

[0038] The processor is configured to execute the steps of the multi-sensor fusion target positioning method provided in any embodiment of the present application by calling the program or instruction stored in the memory.

[0039] An embodiment of the present application further provides a computer-readable storage medium, which stores a program or instruction, and the program or instruction enables a computer to execute the steps of the multi-sensor fusion target positioning method provided by any embodiment of the present application.

[0040] In summary, the present application proposes a target positioning method based on multi-sensor fusion. The method obtains the measurement information at the current moment and the track information at the previous moment, determines the track prediction information at the current moment based on the track information, and then matches the measurement information with the track prediction information to obtain a track measurement matching pair. For each track measurement matching pair, a pseudo measurement value of the corresponding target is constructed based on the track prediction value and the measurement value in the track measurement matching pair, and the track value of the corresponding target at the current moment is determined based on the pseudo measurement value and the track prediction value to achieve target positioning. Compared with directly fusing the measurement value and the track prediction value, this method calculates the pseudo measurement value through the track prediction value and the measurement value in each track measurement matching pair, and can obtain a result close to the actual position of the target. Therefore, by combining the pseudo measurement value and the track prediction value, the track value at the current moment is determined, which can make the final positioning result closer to the actual position of the target, solve the problems of track splitting, jittering, pulling, etc. caused by the large measurement gap between sensors during fusion, so as to ensure the stability of the driving function and driving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the specific implementation methods or the description of the prior art. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0042] Figure 1 This is a flow chart of a multi-sensor fusion target positioning method provided in an embodiment of the present application;

[0043] Figure 2 This is a schematic diagram of a simulation result provided by an embodiment of the present application;

[0044] Figure 3 This is a schematic structural diagram of a multi-sensor fusion target positioning device provided in an embodiment of the present application;

[0045] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] The present application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the relevant invention and are not intended to limit the invention. It should also be noted that, for ease of description, only portions relevant to the invention are shown in the accompanying drawings.

[0047] 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.

[0048] As mentioned in the background technology, to address the problems in the prior art, this application proposes a target positioning method based on multi-sensor fusion. Figure 1 This is a flow chart of a multi-sensor fusion target positioning method provided by an embodiment of the present application. Figure 1 , the multi-sensor fusion target positioning method specifically includes:

[0049] S110 , obtaining measurement information at the current moment and track information at the previous moment, wherein the measurement information includes the measurement value of each target, and the track information includes the track value of each target.

[0050] In the embodiments of the present application, the current measurement information may be information detected by a sensor in real time with respect to a target. Considering that the sampling frequencies (or sampling intervals) of various sensors generally vary, obtaining the current measurement information can be understood as receiving measurement information transmitted in real time by a sensor (radar or camera).

[0051] The measurement information can consist of the measurement values of each target. Specifically, the measurement values may include target position, target velocity, target acceleration, target size, target category, measurement noise, target ID, and the ID of the lane in which the target is located. The measurement noise can be obtained by querying a preset table, which is obtained by piecewise fitting the error between the true value system and the actual measurement value. Target velocity and target acceleration are ground-relative parameters.

[0052] In addition to the measurement values of each target, the measurement information can also include the vehicle body posture at the corresponding moment, timestamp, and sensor ID. The vehicle body posture includes the local coordinate system posture (or global coordinate system posture), vehicle speed information, yaw rate information, and gear position information.

[0053] In order to ensure the accuracy of the measurement information, the measurement information may be pre-processed after being obtained, such as removing abnormal values, filling blank values, etc.

[0054] In the embodiments of this application, the track information at the last moment can be understood as the positioning result obtained by this method at the last moment, including the track value of each target. Specifically, the track value may include the target position, target speed, target acceleration, target size, target category, target ID, the ID of the lane where the target is located, and the target timestamp.

[0055] In addition to the track values of each target, track information can also include the vehicle body posture, the track gain parameters corresponding to each target, and the detection target ID corresponding to each sensor. The track gain parameters can be understood as the weight of the fusion between the track and measurement of the corresponding target, reflecting the deviation between the track and measurement, that is, the measurement gap between sensors (caused by deviation or measurement error).

[0056] S120: Determine the track prediction information at the current moment based on the track information.

[0057] The track prediction information includes the track prediction value of each target. Specifically, after obtaining the track information at the previous moment, time and space alignment can be performed to extrapolate the track information at the previous moment to the current moment to predict the track information at the current moment and obtain the track prediction information at the current moment.

[0058] For example, based on the track value of each target in the track information at the previous moment, combined with the motion model, the position of each target at the current moment can be inferred to obtain the track prediction information at the current moment, so as to align the track information at the previous moment to the sensor moment.

[0059] S130: Match the measurement information with the track prediction information to obtain a track measurement matching pair.

[0060] The measurement values in the measurement information and the track prediction values in the track prediction information may be matched, so that the measurement values and track prediction values belonging to the same target are determined as a track measurement matching pair.

[0061] In a specific embodiment, matching the measurement information with the track prediction information to obtain a track measurement matching pair includes the following steps:

[0062] Step 11: Combine the measurement values in the measurement information with the track prediction values in the track prediction information in pairs to obtain track measurement pairs;

[0063] Step 12: For each track measurement pair, construct a temporary pseudo-measurement value based on the track prediction value and the measurement value in the track measurement pair, and determine the associated cost of the track measurement pair based on the temporary pseudo-measurement value and the track prediction value;

[0064] Step 13: Determine track measurement matching pairs based on the associated costs of each track measurement pair.

[0065] In step 11, each measurement value in the measurement information and each track prediction value in the track prediction information may be combined in pairs to form a plurality of track measurement pairs, so as to obtain all possible track value-measurement value combinations.

[0066] Furthermore, in step 12, for each track measurement pair, the association cost between the track prediction value and the measurement value can be calculated by constructing a temporary pseudo-measurement value, so as to facilitate subsequent determination of whether they match based on the association cost.

[0067] Regarding step 12 above, in one example, constructing a temporary pseudo-measurement value based on the track prediction value and the measured value in the track measurement pair includes the following steps:

[0068] Step 121: Determine an estimated track gain parameter of the target at the current moment based on the track gain parameter of the target corresponding to the track prediction value at the previous moment;

[0069] Step 122: Determine a measurement gain parameter based on the track prediction value and the measured value in the track measurement pair;

[0070] Step 123: Determine a temporary gain parameter of the track-measurement pair based on the estimated track gain parameter and the measured gain parameter;

[0071] Step 124: Determine a temporary pseudo measurement value based on the temporary gain parameter and the measurement value of the track measurement.

[0072] The track gain parameter may be composed of a track gain value and a track gain variance. In step 121, the track gain parameter of the target corresponding to the track prediction value may be obtained from the track information at the previous moment, and then the track gain parameter at the previous moment may be optimally estimated at the current moment to obtain the estimated track gain parameter of the target at the current moment.

[0073] For example, the estimated track gain parameter at the current moment can be determined by the following formula:

[0074] GainVar t|t-1 =GainVar t-1 +Q Gain *d t *d t ;

[0075] Gain t|t-1 =Gain t-1 ;

[0076] Where, GainVar t|t-1 Gain is the estimated track gain variance at the current moment, t|t-1 GainVar is the estimated track gain value at the current moment; t-1 Gain is the track gain variance at the previous moment, t-1 is the track gain value at the previous moment; Q Gain is the process noise; d t It is the time difference between the current moment and the previous moment, in seconds.

[0077] Furthermore, in step 122, a measurement gain parameter may be calculated based on the track prediction value and the measured value in the track measurement pair. The measurement gain parameter may reflect the difference between the measurement and the track. The measurement gain parameter may be composed of the measurement gain value and the measurement gain noise.

[0078] As shown in the following formula:

[0079]

[0080] In the formula, (x t ,y t ) represents the longitudinal and transverse positions in the track prediction value, (x m ,y m ) represents the vertical and horizontal position of the measured value, Indicates the vertical and horizontal measurement noise of the measurement preset. Indicates the longitudinal and transverse covariance of the track, which can be calculated after the filtering is completed at the last moment. Gain m Indicates the measurement gain value, R Gain Represents the measured gain noise.

[0081] Furthermore, in step 123, a temporary gain parameter of the track measurement pair may be determined based on the estimated track gain parameter and the measured gain parameter. Specifically, the temporary gain parameter may be determined based on whether the sensor corresponding to the measured value in the track measurement pair has participated in track updating.

[0082] Regarding step 123 above, in an optional embodiment, determining a temporary gain parameter of the track-measurement pair based on the estimated track gain parameter and the measurement gain parameter includes:

[0083] If the sensor corresponding to the track measurement value has not participated in the track update, a temporary gain parameter is determined based on the measurement gain parameter;

[0084] If the sensor corresponding to the measurement value in the track measurement has participated in the track update, and the target corresponding to the measurement value is different from the target updated by the sensor, the temporary gain parameter is set to the preset value;

[0085] If the sensor corresponding to the measurement value in the track measurement has participated in track update, and the target corresponding to the measurement value is the same as the target updated by the sensor, a temporary gain parameter is determined based on the estimated track gain parameter and the measurement gain parameter.

[0086] The temporary gain parameters include a temporary gain value and a temporary gain variance.

[0087] Specifically, in the first case, if the sensor corresponding to the measurement value in the track measurement (i.e., the sensor that collects the measurement information) has not participated in the track update, that is, the measurement value detected by the sensor is not integrated when the track value is determined at each historical moment, in order to prevent the sensor from being missed by mistake, the measurement gain value can be directly determined as the temporary gain value, and the measurement gain noise can be determined as the temporary gain variance.

[0088] In the second case, if the sensor corresponding to the measurement value in the track measurement has participated in the track update, and the target updated by the sensor is different from the target corresponding to the measurement value, it means that the measurement value of the sensor at the current moment may be erroneous, and the sensor has detected the wrong target. Therefore, the temporary gain value and the temporary gain variance can be set to 0.

[0089] In the third case, if the sensor corresponding to the track measurement value has participated in the track update, and the target updated by the sensor is the same as the target corresponding to the measurement value, it means that the sensor's measurement value at the current moment is accurate and the sensor has detected the correct target. Therefore, the estimated track gain parameter and the measurement gain parameter can be fused to obtain the temporary gain parameter. As shown in the following formula:

[0090]

[0091] Where Gain and GainVar are the temporary gain value and temporary gain variance, respectively.

[0092] Through the above optional implementation method, the targets updated by the sensor corresponding to the measurement value during the historical track calculation process can be combined to divide three situations and determine the temporary gain parameters respectively, which can ensure the accuracy of the temporary gain parameters and thus ensure the correctness of the temporary pseudo-measurement values constructed based on the temporary gain parameters.

[0093] After obtaining the temporary gain parameter, in step 124, a temporary pseudo-measurement value can be calculated based on the temporary gain parameter and the measurement value in the track measurement pair. To further ensure the accuracy of the temporary pseudo-measurement value, the temporary gain parameter can be numerically constrained based on the target type to prevent the temporary gain parameter from being too large or too small, causing the calculated temporary pseudo-measurement value to deviate from the target's true position. For example, the temporary gain parameter can be constrained to the range of [0.9, 1.1].

[0094] Among them, the temporary pseudo-measurement value can be a pseudo-measurement value constructed for target matching. Calculating the temporary pseudo-measurement value can make the measurement value close to the actual position of the target and compensate for the detection error of the sensor, so that it can be correctly matched to the track prediction value, ensuring the accuracy of target matching, avoiding the mistaken matching of the measurement values of different targets with the track prediction values, and at the same time, avoiding the failure to match the measurement value of the same target with the track prediction value.

[0095] For example, the temporary pseudo-measurement value can be calculated by the following formula:

[0096] x Pseudo =x m / Gain;

[0097] y Pseudo =y m / Gain;

[0098] V xPseudo =V xm *ScaleV x / Gain+V xm *(1-ScaleV x );

[0099] V yPseudo =V ym *ScaleV y / Gain+V ym *(1-ScaleV y );

[0100]

[0101] R vxPseudo =R mVx +V xm 2 *Gain;

[0102] R vyPseudo =R mVy +V ym *Gain;

[0103] Where x Pseudo 、y PseudoIndicates the vertical and horizontal position in the pseudo-measurement value, V xPseudo 、V yPseudo Indicates the longitudinal and transverse velocities in the pseudo-measurement value, R xPseudo 、R yPseudo represents the vertical and horizontal measurement noise in the pseudo-measurement value, R vxPseudo 、R vyPseudo represents the longitudinal and transverse velocity noise in the pseudo-measurement value; x m 、y m Indicates the vertical and horizontal position of the measured value, V xm 、V ym Indicates the longitudinal and transverse speed in the measured value, R mx 、R my Indicates the vertical and horizontal measurement noise in the measured value, R mVx 、R mVy Indicates the longitudinal and transverse velocity noise in the measured value; ScaleV x ScaleV y It represents the ratio between the longitudinal and lateral velocities (speed relative to the ground) in the measured value and the relative speed. The longitudinal and lateral velocities in the measured value can be converted into relative speed based on signals such as vehicle speed and yaw rate.

[0104] After constructing the temporary pseudo-measurement value, the probability that the track measurement pair belongs to the same target can be further measured based on the difference between the temporary pseudo-measurement value and the track prediction value, thereby obtaining an association cost. This association cost can be understood as a matching score. The higher the association cost, the lower the probability that the track measurement pair belongs to the same target. For example, the association cost can be determined based on the distance between the temporary pseudo-measurement value and the track prediction value. The closer the distance, the higher the probability that they belong to the same target, and the lower the association cost.

[0105] In order to ensure the accuracy of the association cost and thus improve the accuracy of target matching, considering that the difference between the measured speed and the track speed and the position difference of the same target are small, the association cost can be calculated by combining the speed and position.

[0106] In one example, determining the associated cost of a track measurement pair based on a temporary pseudo-measurement value and a track prediction value includes the following steps:

[0107] Step 125: Determine a speed-related cost based on the target speed in the temporary pseudo-measurement value and the target speed in the track prediction value;

[0108] Step 126: Determine a position association cost based on the target position in the temporary pseudo-measurement value and the target position in the track prediction value;

[0109] Step 127: Determine the association cost of the track measurement pair according to the speed association cost and the position association cost.

[0110] In step 125, the Mahalanobis distance in speed can be calculated based on the target speed in the temporary pseudo-measurement value and the target speed in the track prediction value as the speed association cost, so as to describe the difference between the temporary pseudo-measurement speed and the track prediction speed through the speed association cost.

[0111] Furthermore, in step 126, the Mahalanobis distance in position can be calculated based on the target position in the temporary pseudo-measurement value and the target position in the track prediction value as the position association cost, so as to describe the gap between the temporary pseudo-measurement position and the track prediction position through the position association cost.

[0112] Furthermore, in step 127, the speed-related cost and the position-related cost may be fused in a weighted manner, wherein the weights corresponding to the speed-related cost and the position-related cost, respectively, may be determined according to the vehicle speed or the target speed.

[0113] For example, if the target speed is higher, it means that the target is in a high-speed driving state. At this time, there may be a deviation in the target position in the track prediction value deduced based on the target speed, and the weight corresponding to the position-associated cost can be appropriately increased; or, if the vehicle speed is higher, it means that the vehicle is in a high-speed driving state. At this time, there may be a deviation in the position detected by the vehicle sensor, and the weight corresponding to the speed-associated cost can be appropriately increased.

[0114] Through the above steps 125 to 127, the possibility that the temporary pseudo-measurement value and the track prediction value belong to the same target can be calculated from the target speed and target position, thereby further improving the accuracy of target matching.

[0115] After obtaining the association cost for each track-measurement pair, in step 13, track-measurement pairs with an association cost below a preset cost threshold are selected as track-measurement matching pairs to achieve matching between the track prediction value and the measured value, i.e., target matching. For track-measurement matching pairs, the track prediction value and the measured value can be assumed to belong to the same target.

[0116] S140. For each track measurement matching pair, construct a pseudo measurement value of the corresponding target based on the track prediction value and the measurement value in the track measurement matching pair, and determine the track value of the corresponding target at the current moment based on the pseudo measurement value and the track prediction value.

[0117] Specifically, for each track-measurement pair, a pseudo-measurement value of the corresponding target can be constructed based on the track prediction value and the measured value. Pseudo-measurement values are virtual measurements constructed based on the track prediction value and the measured value. Unlike measurement information, which is the result of actual sensor measurements, pseudo-measurement values are the result of error compensation for the measured values.

[0118] It should be noted that the pseudo measurement value is different from the temporary pseudo measurement value. The temporary pseudo measurement value is used to calculate the association cost for target matching, while this pseudo measurement value is used for track filtering.

[0119] In a specific embodiment, constructing a pseudo measurement value of a corresponding target based on a track prediction value and a measurement value in a track measurement matching pair includes the following steps:

[0120] Step 21: Determine a measurement gain parameter based on the track prediction value and the measurement value in the track measurement matching pair;

[0121] Step 22: Determine the target's track gain parameter at the current moment based on the track gain parameter of the target at the previous moment corresponding to the track prediction value and the measurement gain parameter;

[0122] Step 23: Determine a pseudo measurement value of the target based on the measurement value in the track measurement matching pair and the track gain parameter at the current moment.

[0123] Among them, in step 21, the measurement gain parameter can be first calculated based on the track prediction value and the measurement value in the track measurement matching pair. The calculation of the measurement gain parameter is the same as the calculation method of the measurement gain parameter in the above-mentioned temporary pseudo-measurement value construction process. Please refer to the above-mentioned step 121 and will not be repeated here.

[0124] Furthermore, in step 22, the track gain parameter of the target corresponding to the track prediction value at the previous moment can be obtained based on the track information at the previous moment. The track gain parameter at the previous moment is then combined with the measurement gain parameter to determine the track gain parameter of the target at the current moment. The track gain parameter at the current moment can be stored together with the subsequently obtained track value at the current moment as the track information at the current moment.

[0125] Regarding step 22 above, in one example, determining the target's track gain parameter at the current moment based on the track gain parameter of the target corresponding to the track prediction value at the previous moment and the measurement gain parameter includes:

[0126] If the target corresponding to the track prediction value in the track measurement match is updated by the sensor corresponding to the measurement value for the first time, or the target corresponding to the track prediction value has not been updated by the sensor within a preset number of consecutive times, the track gain parameters of the target at the current moment are determined based on the measurement gain parameters. Otherwise, the track gain parameters of the target at the current moment are determined based on the track gain parameters of the target at the previous moment and the measurement gain parameters.

[0127] The track gain parameters include the track gain value and the track gain variance.

[0128] Specifically, if the sensor corresponding to the measurement value in the track measurement matching pair participates in the update of the target corresponding to the track prediction value in the track measurement matching pair for the first time, that is, the first measurement value is successfully matched with the track prediction value of the target, or the target corresponding to the track prediction value has not been updated by the sensor within a preset number of consecutive times, then in order to avoid missing the sensor by mistake, the measurement gain parameter can be directly used to initialize the track gain parameter at the current moment, that is, the measurement gain value is used as the track gain value at the current moment, and the measurement gain noise is used as the track gain variance at the current moment.

[0129] If the sensor corresponding to the measurement value in the track measurement matching pair is not participating in the update of the target corresponding to the track prediction value in the track measurement matching pair for the first time, and the target corresponding to the track prediction value has been updated by the sensor within a preset number of consecutive times, the track gain parameter at the previous moment can be fused with the measurement gain parameter to obtain the track gain parameter at the current moment.

[0130] For example, the target's track gain value at the current moment can be determined based on the target's track gain value and track gain variance at the previous moment, as well as the measurement gain value and measurement gain noise corresponding to the track prediction value. Furthermore, the current filter gain is determined based on the track gain variance and measurement gain noise at the previous moment. Finally, the target's track gain variance at the current moment is determined based on the current filter gain and the track gain variance at the previous moment. This is shown in the following formula:

[0131]

[0132] GainVar t =(IK)*GainVar t-1 +K*R Gain *K T ;

[0133] Where, Gain t Gain is the track gain value at the current moment. t-1 Gain is the track gain value at the previous moment; m is the measurement gain value, R Gain GainVar is the measurement of gain noise; t GainVar is the track gain variance at the current moment. t-1 is the track gain variance at the previous moment; K is the current filter gain (if the Kalman filter method is used later, K can specifically be the Kalman gain).

[0134] In the above example, by determining whether the target corresponding to the track prediction value in the track measurement match has been updated by the current sensor and whether it has been updated in the short term, multiple situations are divided and corresponding track gain determination methods are adopted in different situations, further ensuring the accuracy of the track gain parameters at the current moment.

[0135] After obtaining the track gain parameter at the current moment, further in step 23 , a pseudo measurement value may be calculated based on the measurement value in the track measurement matching pair and the track gain parameter at the current moment to obtain a result close to the true position of the target.

[0136] The calculation method of the pseudo-measurement value may refer to the calculation method of the temporary pseudo-measurement value mentioned above, and the temporary track gain parameter in the above formula may be replaced by the track gain parameter at the current moment.

[0137] In order to further ensure that the pseudo-measurement value is closer to the actual position of the target, the value range of the track gain parameter at the current moment may be restricted, and then the pseudo-measurement value may be calculated based on the restricted track gain parameter at the current moment.

[0138] In one example, before determining the pseudo measurement value of the target based on the measurement value in the track measurement matching pair and the track gain parameter at the current moment, the method further includes:

[0139] Based on the type of target corresponding to the track prediction value in the track measurement matching, the gain parameter range is determined; whether the track gain parameter at the current moment is within the gain parameter range is judged, and if not, the track gain parameter at the current moment is updated based on the gain parameter range.

[0140] The gain parameter ranges corresponding to different types can be pre-stored. For example, the target type can reflect the target's speed and size. For targets with high speed or small size, the ranging deviation is large, so the gain parameter range can be appropriately increased to compensate for more errors in the measured value. For targets with low speed or large size, the ranging deviation is small, so the gain parameter range can be appropriately reduced to avoid excessive error compensation in the measured value, which may cause further deviation from the target's true position.

[0141] Specifically, after querying the gain parameter range through the target type, it can be determined whether the track gain parameter at the current moment is less than the lower limit of the gain parameter range, and whether it is greater than the upper limit of the gain parameter range. If it is less than the lower limit, the lower limit is used to update the track gain parameter at the current moment; if it is greater than the upper limit, the upper limit is used to update the track gain parameter at the current moment.

[0142] Through the above example, it is possible to ensure that the error compensation of the measurement value is more accurate, avoid over-compensation or under-compensation, and make the pseudo-measurement value closer to the actual position of the target.

[0143] In the above steps 21 to 23, the track gain parameter at the current moment can be calculated, and then a pseudo-measurement value can be constructed by combining the track gain parameter with the measurement value, thereby ensuring the stability of target positioning.

[0144] After constructing the pseudo-measurement value, the pseudo-measurement value can be further fused with the track prediction value to obtain the track value at the current moment to achieve the track update of the target.

[0145] For example, the pseudo-measurement value and the track prediction value may be fused in a weighted manner, or the track prediction value may be subjected to Kalman filtering based on the pseudo-measurement value to obtain the track value at the current moment.

[0146] It should be noted that for track measurement pairs that have not been successfully matched, the attributes of the track values at the previous moment corresponding to the track prediction values can be updated, and the track generation processing can be performed after the measurement value decision is made. This method can obtain the track of each target during the positioning process at the first moment.

[0147] In the embodiment of the present application, the entire positioning process is also simulated and analyzed. First, the motion process of the vehicle and the target is simulated, and the speed, position, acceleration and other information at each time step are recorded. The initial parameters of the vehicle and the target include speed, position and acceleration. By setting the time step and motion model, the position and speed of the vehicle and the target are updated, and these data are recorded. The initial position of the target is 100 meters away from the longitudinal distance of the vehicle, and it moves at a constant speed of 15m / s. After running for 10 seconds, it moves at a speed of -9m / s. 2 The deceleration brake is at -4m / s when the running reaches 11 seconds. 2 The car starts braking at a deceleration of -4m / s at 10.5 seconds. 2 The deceleration brake is at -3m / s at 11.5 seconds. 2 The vehicle decelerates and brakes until it comes to a complete stop. Simulate the measurement information of two sensors. Sensor A is assumed to be a radar sensor. Random noise with a standard deviation of 3% is added to the position attribute, and the speed attribute noise is kept at the same location. Sensor B is assumed to be a vision sensor. Based on its characteristics, random deviation and random noise with a standard deviation of 10% are added to the position attribute, and the speed attribute noise is kept at the same location.

[0148] This example uses a standard Kalman filter, the motion model is a uniform acceleration model, and the process noise is The state transition matrix is The default value of the measurement noise of sensor A is where σ x =0.03*x,σ vx =0.03*v x ,σax =1, the upper and lower limits are [0.5,10]. The default value of the measurement noise of sensor B is where σ x =0.1*x,σ vx =0.1*v x ,σ ax =10, with upper and lower limits of [0.5, 10]. The association cost is represented by the standard Mahalanobis distance, and the filtering results with gain estimation are compared with the original filtering results.

[0149] Figure 2 This is a schematic diagram of a simulation result provided by an embodiment of the present application. Figure 2 As shown, the horizontal axis is time (unit is s), and the vertical axis is the target position (unit is m). Figure 2 The upper curve in the figure depicts the difference between the original filtered result (the result obtained by filtering the track value at the previous moment based on the measured value) and the true value, while the lower curve depicts the difference between the filtered result obtained by the method provided in the embodiment of the present application and the true value. As can be seen from the figure, the positioning result obtained by the method provided in the embodiment of the present application is closer to the target's true position.

[0150] Through simulation analysis, it can be obtained that the mean square error of sensor B is 91.7181, the mean square error of the original filtering method is 6.79, and the mean square error of the gain estimation filtering variance provided by the embodiment of the present application is 0.81, which significantly improves the reliability and accuracy of target positioning.

[0151] The multi-sensor fusion target positioning method provided in the embodiments of the present application obtains measurement information at the current moment and track information at the previous moment, converts the track information to the current moment, obtains track prediction information at the current moment, and then matches the measurement information with the track prediction information to obtain a track measurement matching pair. For each track measurement matching pair, a pseudo measurement value of the corresponding target is constructed based on the track prediction value and the measurement value in the track measurement matching pair. Based on the pseudo measurement value and the track prediction value, the track value of the corresponding target at the current moment is determined to achieve target positioning. Compared with directly fusing the measurement value and the track prediction value, this method calculates the pseudo measurement value by using the track prediction value and the measurement value in each track measurement matching pair, and can obtain a result close to the actual position of the target. Therefore, by combining the pseudo measurement value and the track prediction value, the track value at the current moment is determined, which can make the final positioning result closer to the actual position of the target, and solve the problems of track splitting, jitter, and pulling caused by large measurement gaps between sensors during fusion, thereby ensuring the stability of driving functions and driving safety.

[0152] Figure 3Schematic diagram of a multi-sensor fusion target positioning device provided in an embodiment of the present application. The multi-sensor fusion target positioning device provided in an embodiment of the present application can be applied to the multi-sensor fusion target positioning method provided in any embodiment of the present application.

[0153] like Figure 3 As shown, the device includes an information acquisition module 310, a spatiotemporal alignment module 320, a matching module 330, and a track update module 340, wherein:

[0154] An information acquisition module 310 is configured to acquire measurement information at a current moment and track information at a previous moment, wherein the measurement information includes the measurement value of each target, and the track information includes the track value of each target;

[0155] a spatiotemporal alignment module 320 for determining track prediction information at a current moment based on the track information, wherein the track prediction information includes track prediction values of each target;

[0156] a matching module 330 for matching the measurement information with the track prediction information to obtain a track measurement matching pair;

[0157] The track updating module 340 is configured to construct a pseudo-measurement value of the corresponding target for each track measurement matching pair based on the track prediction value and the measurement value in the track measurement matching pair, and determine the track value of the corresponding target at the current moment based on the pseudo-measurement value and the track prediction value.

[0158] The multi-sensor fusion target positioning device provided in the embodiment of the present application can execute the steps of the multi-sensor fusion target positioning method provided in the method embodiment of the present application, and the execution steps and beneficial effects are not repeated here.

[0159] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 4 As shown, the electronic device 400 includes one or more processors 401 and a memory 402 .

[0160] The processor 401 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 400 to perform desired functions.

[0161] The memory 402 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), a hard disk, a flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 401 may execute the program instructions to implement the multi-sensor fusion target positioning method of any embodiment of the present application described above and / or other desired functions. Various contents such as initial external parameters, thresholds, etc. may also be stored in the computer-readable storage medium.

[0162] In one example, electronic device 400 may further include an input device 403 and an output device 404, which are interconnected via a bus system and / or other connection mechanisms (not shown). Input device 403 may include, for example, a keyboard, a mouse, etc. Output device 404 may output various information to the outside, including warning information, braking force, etc. Output device 404 may include, for example, a display, a speaker, a printer, a communication network, and remote output devices connected thereto.

[0163] Of course, to simplify, Figure 4 Only some of the components related to the present application in the electronic device 400 are shown, and components such as buses, input / output interfaces, etc. are omitted. In addition, the electronic device 400 may further include any other appropriate components according to specific application scenarios.

[0164] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to perform the steps of the multi-sensor fusion target positioning method provided by any embodiment of the present application.

[0165] The computer program product may be written in any combination of one or more programming languages to implement the program code for performing the operations of the embodiments of the present application, including object-oriented programming languages such as Java, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0166] In addition, an embodiment of the present application may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, causes the processor to execute the steps of the multi-sensor fusion target positioning method provided by any embodiment of the present application.

[0167] The computer-readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can, for example, include but is not limited to a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0168] It should be noted that the terms used in this application are only for describing specific embodiments and are not intended to limit the scope of this application. As shown in the specification and claims of this application, unless the context clearly indicates an exception, the words "one", "an", "a kind of" and / or "the" do not specifically refer to the singular and may also include the plural. The terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method or device. In the absence of further restrictions, the elements defined by the sentence "comprise a..." do not exclude the presence of other identical elements in the process, method or device comprising the elements.

[0169] It should also be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on this application. Unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", etc. should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be a communication between the internal parts of two elements. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0170] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. The above is only the preferred implementation method of this application. It should be pointed out that due to the limitations of textual expression, there are objectively infinite specific structures. For ordinary technicians in this technical field, without departing from the principles of this application, they can also make several improvements, modifications or changes, and can also combine the above technical features in an appropriate manner; these improvements, modifications, changes or combinations, or the direct application of the inventive concept and technical solution to other occasions without improvement, should be regarded as the scope of protection of this application.

Claims

1. A target positioning method based on multi-sensor fusion, characterized in that: include: Acquire measurement information at a current moment and track information at a previous moment, wherein the measurement information includes a measurement value of each target, and the track information includes a track value of each target; Determining track prediction information at a current moment based on the track information, wherein the track prediction information includes track prediction values of each target; Matching the measurement information with the track prediction information to obtain a track measurement matching pair; For each track measurement matching pair, a pseudo measurement value of the corresponding target is constructed based on the track prediction value and the measurement value in the track measurement matching pair, and the track value of the corresponding target at the current moment is determined according to the pseudo measurement value and the track prediction value.

2. The method according to claim 1, characterized in that Matching the measurement information with the track prediction information to obtain a track measurement matching pair, including: Combining the measurement values in the measurement information with the track prediction values in the track prediction information in pairs to obtain track measurement pairs; For each track measurement pair, construct a temporary pseudo measurement value based on the track prediction value and the measurement value in the track measurement pair, and determine the associated cost of the track measurement pair according to the temporary pseudo measurement value and the track prediction value; A track measurement matching pair is determined according to the associated cost of each track measurement pair.

3. The method according to claim 2, characterized in that Constructing a temporary pseudo-measurement value based on the track prediction value and the measured value in the track measurement pair, including: Determining an estimated track gain parameter of the target at the current moment based on the track gain parameter of the target corresponding to the track prediction value at the previous moment; Determining a measurement gain parameter based on a track prediction value and a measurement value in the track measurement pair; determining a temporary gain parameter of the track-measurement pair based on the estimated track gain parameter and the measurement gain parameter; A temporary pseudo measurement value is determined according to the temporary gain parameter and the measurement value of the track measurement pair.

4. The method according to claim 3, characterized in that Determining a temporary gain parameter of the track-measurement pair based on the estimated track gain parameter and the measurement gain parameter includes: If the sensor corresponding to the track measurement centering measurement value has not participated in track updating, determining the temporary gain parameter based on the measurement gain parameter; If the sensor corresponding to the measurement value of the track measurement has participated in track update, and the target corresponding to the measurement value is different from the target updated by the sensor, then the temporary gain parameter is set to a preset value; If the sensor corresponding to the measurement value in the track measurement has participated in track update, and the target corresponding to the measurement value is the same as the target updated by the sensor, the temporary gain parameter is determined based on the estimated track gain parameter and the measurement gain parameter.

5. The method according to claim 2, characterized in that Determining an association cost of the track measurement pair according to the temporary pseudo measurement value and the track prediction value includes: determining a speed-associated cost based on the target speed in the temporary pseudo-measurement value and the target speed in the track prediction value; determining a position association cost based on the target position in the temporary pseudo-measurement value and the target position in the track prediction value; An association cost of the track measurement pair is determined according to the speed association cost and the position association cost.

6. The method according to claim 1, characterized in that Constructing a pseudo measurement value of the corresponding target based on the track prediction value and the measurement value in the track measurement matching pair, including: determining a measurement gain parameter based on a track prediction value and a measurement value in the track measurement matching pair; Determining a track gain parameter of the target at the current moment based on the track gain parameter of the target corresponding to the track prediction value at the previous moment and the measurement gain parameter; A pseudo measurement value of the target is determined based on the measurement values in the track measurement matching pair and a track gain parameter at a current moment.

7. The method according to claim 6, characterized in that Determining the track gain parameter of the target at the current moment based on the track gain parameter of the target corresponding to the track prediction value at the previous moment and the measurement gain parameter includes: If the target corresponding to the track prediction value in the track measurement matching pair is updated for the first time by the sensor corresponding to the measurement value, or the target corresponding to the track prediction value has not been updated by the sensor within a preset number of consecutive times, the track gain parameter of the target at the current moment is determined based on the measurement gain parameter; otherwise, the track gain parameter of the target at the current moment is determined based on the track gain parameter of the target at the previous moment and the measurement gain parameter.

8. The method according to claim 6, characterized in that Before determining the pseudo measurement value of the target based on the measurement value in the track measurement matching pair and the track gain parameter at the current moment, the method further includes: determining a gain parameter range based on a type of target corresponding to a track prediction value in the track measurement matching pair; It is determined whether the track gain parameter at the current moment is within the gain parameter range; if not, the track gain parameter at the current moment is updated based on the gain parameter range.

9. An electronic device, characterized in that: The electronic device comprises: processor and memory; The processor is configured to execute the steps of the multi-sensor fusion target positioning method according to any one of claims 1 to 8 by calling the program or instructions stored in the memory.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a program or instruction, and the program or instruction enables a computer to execute the steps of the target positioning method based on multi-sensor fusion according to any one of claims 1 to 8.

Images