Target positioning method, device, electronic device and storage medium

Abstract

Embodiments of the present invention provide a target positioning method, device, electronic device, and storage medium, relating to the field of autonomous driving positioning technology. The method includes: obtaining a target detection frame of multiple targets around a vehicle at the current moment in a second coordinate system, and obtaining a target tracking frame of multiple targets at the previous moment in a first coordinate system; obtaining the vehicle's current posture in the first coordinate system; converting the target detection frame of the multiple targets in the second coordinate system at the current moment into a target detection frame of the multiple targets in the first coordinate system at the current moment based on the vehicle's posture; matching the target detection frame in the first coordinate system at the current moment with the target tracking frame in the first coordinate system at the previous moment, using the successfully matched first target tracking frame as the vehicle state observation, optimizing the vehicle's posture, and obtaining the optimized vehicle posture in the first coordinate system at the current moment, thereby improving the accuracy of autonomous driving positioning.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of autonomous driving positioning technology, and in particular to a target positioning method, device, electronic device, and storage medium. Background Art

[0002] Currently, autonomous driving positioning technology is about achieving high-precision, stable and reliable positioning of autonomous driving vehicles at night or in severe weather conditions. This technology mainly involves the following aspects: sensor technology, map matching technology, communication technology and artificial intelligence technology. Among them, the development of sensor technology provides more diverse options for autonomous driving positioning.

[0003] However, current autonomous driving positioning technology still faces several challenges and issues. For example, some sensors and positioning systems may fail or distort data at night or in inclement weather, resulting in reduced or ineffective positioning accuracy. Therefore, improving the accuracy of autonomous driving positioning is a technical problem that this invention urgently seeks to address. Summary of the Invention

[0004] Based on the above technical problems, embodiments of the present invention provide a target positioning method, device, electronic device and storage medium to improve the accuracy of autonomous driving positioning.

[0005] An embodiment of the present invention provides a target positioning method, the method comprising:

[0006] Obtain target detection frames of multiple targets around the ego vehicle at the current moment in the second coordinate system, and obtain target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the ego vehicle at the previous moment;

[0007] According to the driving information of the vehicle, the vehicle's current position in the first coordinate system is obtained;

[0008] According to the current posture of the vehicle in the first coordinate system, the target detection frames of the multiple targets around the vehicle at the current moment in the second coordinate system are converted to obtain the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system;

[0009] Matching the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment, to obtain a first matching result corresponding to the target tracker;

[0010] The first target tracking frame successfully matched in the first matching result is used as the vehicle state observation quantity, and the vehicle posture is optimized to obtain the optimized vehicle posture of the vehicle in the first coordinate system at the current moment.

[0011] Optionally, the method further includes:

[0012] According to the optimized pose of the ego vehicle at the current moment in the first coordinate system, the target detection frames of the multiple targets around the ego vehicle at the current moment in the second coordinate system are converted to obtain the optimized target detection frames of the multiple targets around the ego vehicle at the current moment in the first coordinate system;

[0013] Matching the target detection frames optimized in the first coordinate system for the multiple targets surrounding the ego vehicle at the current moment with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment, to obtain a second matching result corresponding to the target tracker;

[0014] The second target tracking frame that is successfully matched in the second matching result is used as the target state observation, and the target tracker corresponding to the second target tracking frame is updated.

[0015] Optionally, the method further includes:

[0016] creating a new target tracker for the third target tracking frame that fails to match in the first matching result;

[0017] Obtaining first matching results corresponding to the new target tracker at multiple consecutive moments;

[0018] If the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all successful matches, retaining the new target tracker;

[0019] If the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all matching failures, the new target tracker is deleted.

[0020] Optionally, the method further includes:

[0021] Calculating a residual between the first target tracking frame and a target detection frame that successfully matches the first target tracking frame;

[0022] Determine the first target tracking frame corresponding to the residual smaller than a preset threshold as the first reference target tracking frame;

[0023] The method of using the first target tracking frame successfully matched in the first matching result as the vehicle state observation quantity and optimizing the vehicle posture to obtain the optimized vehicle posture of the vehicle in the first coordinate system at the current moment includes:

[0024] The first reference target tracking frame is used as the ego-vehicle state observation quantity, and the ego-vehicle posture is optimized to obtain the optimized ego-vehicle posture of the ego-vehicle at the current moment in the first coordinate system.

[0025] Optionally, obtaining the vehicle posture of the vehicle in the first coordinate system at the current moment based on the vehicle's driving information includes:

[0026] Obtain the current vehicle speed and displacement provided by the inertial measurement unit, and obtain the current actual travel distance provided by the wheel speed sensor;

[0027] The velocity and displacement are used as state variables, the actual movement distance is used as an observation value, and the vehicle posture of the vehicle at the current moment in the first coordinate system is obtained through Kalman filtering.

[0028] Optionally, the ego-vehicle posture includes: ego-vehicle posture information and ego-vehicle error value;

[0029] The converting, based on the current posture of the vehicle in the first coordinate system, the target detection frames of the multiple targets around the vehicle at the current moment in the second coordinate system to obtain the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system includes:

[0030] Converting the pose information and detection frame error values ​​of the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the second coordinate system according to the pose information and the ego vehicle error value of the ego vehicle at the current moment in the first coordinate system to obtain the pose information and detection frame error values ​​of the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system;

[0031] The step of matching the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment to obtain a first matching result corresponding to the target tracker includes:

[0032] Matching the pose information and detection frame error values ​​of the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system with the pose information and tracking frame error values ​​of the target tracking frames output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment in the first coordinate system to obtain a first matching result;

[0033] The method of using the first target tracking frame successfully matched in the first matching result as the vehicle state observation quantity and optimizing the vehicle posture to obtain the optimized vehicle posture of the vehicle in the first coordinate system at the current moment includes:

[0034] The pose information and tracking frame error value of the first target tracking frame that is successfully matched in the first matching result are used as vehicle state observation quantities, and the pose information and the error value of the vehicle are optimized to obtain the optimized pose information and error value of the vehicle in the first coordinate system at the current moment.

[0035] Optionally, the detection frame error value of the target detection frame of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system is: the error value after the detection frame error value of the target detection frame of the multiple targets surrounding the ego vehicle at the current moment in the second coordinate system is converted to the first coordinate system, and the sum of the error value caused by the ego vehicle error value in the first coordinate system at the current moment.

[0036] A second aspect of an embodiment of the present invention provides a target positioning device, the device comprising:

[0037] A target tracking and detection module is used to obtain target detection frames of multiple targets around the vehicle at the current moment in the second coordinate system, and to obtain target tracking frames in the first coordinate system output by the target tracker corresponding to multiple targets around the vehicle at the previous moment;

[0038] The ego vehicle posture inference module is used to obtain the ego vehicle posture at the current moment in the first coordinate system based on the ego vehicle's driving information;

[0039] a detection frame conversion module, configured to convert target detection frames of multiple targets surrounding the ego vehicle at the current moment in the second coordinate system according to the ego vehicle's current position in the first coordinate system, to obtain target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system;

[0040] a tracking, detection, and matching module configured to match target detection frames of the multiple targets surrounding the vehicle at the current moment in the first coordinate system with target tracking frames output by a target tracker corresponding to the multiple targets surrounding the vehicle at the previous moment in the first coordinate system, to obtain a first matching result corresponding to the target tracker;

[0041] The ego-vehicle posture optimization module is configured to optimize the ego-vehicle posture by using the first target tracking frame successfully matched in the first matching result as the ego-vehicle state observation quantity, and obtain the optimized ego-vehicle posture of the ego-vehicle at the current moment in the first coordinate system.

[0042] Optionally, the device further comprises:

[0043] a detection frame optimization module, configured to convert target detection frames of multiple targets surrounding the ego vehicle at the current moment in the second coordinate system according to the optimized ego vehicle posture at the current moment in the first coordinate system, to obtain target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system;

[0044] an optimization matching module, configured to match the target detection frames optimized in the first coordinate system for the multiple targets surrounding the ego vehicle at the current moment with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment, to obtain a second matching result corresponding to the target tracker;

[0045] The tracker update module is configured to use the successfully matched second target tracking frame in the second matching result as a target state observation and update the target tracker corresponding to the second target tracking frame.

[0046] Optionally, the device further comprises:

[0047] A tracker creation module, configured to create a new target tracker for the third target tracking frame that fails to match in the first matching result;

[0048] A matching result acquisition module, configured to acquire first matching results corresponding to the new target tracker at multiple consecutive moments;

[0049] a tracker retaining module, configured to retain the new target tracker if the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all successful matches;

[0050] The tracker deletion module is configured to delete the new target tracker if the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all matching failures.

[0051] Optionally, the device further comprises:

[0052] a residual determination module, configured to calculate a residual between the first target tracking frame and a target detection frame that successfully matches the first target tracking frame;

[0053] A tracking frame reference module, configured to determine a first target tracking frame corresponding to a residual smaller than a preset threshold as a first reference target tracking frame;

[0054] The vehicle posture optimization module includes:

[0055] The ego-vehicle posture optimization submodule is used to optimize the ego-vehicle posture using the first reference target tracking frame as the ego-vehicle state observation quantity to obtain the optimized ego-vehicle posture of the ego-vehicle at the current moment in the first coordinate system.

[0056] Optionally, the vehicle posture inference module includes:

[0057] The parameter determination submodule is used to obtain the current vehicle speed and displacement provided by the inertial measurement unit, and to obtain the current actual travel distance provided by the wheel speed sensor;

[0058] The Kalman filter determination submodule is used to use the speed and displacement as state variables and the actual movement distance as an observation value to obtain the vehicle posture of the vehicle in the first coordinate system at the current moment through Kalman filtering.

[0059] Optionally, the ego-vehicle posture includes: ego-vehicle posture information and ego-vehicle error value;

[0060] The detection frame conversion module includes:

[0061] a detection frame conversion submodule, configured to convert the target detection frame pose information and detection frame error values ​​of the multiple targets surrounding the ego vehicle at the current moment in the second coordinate system based on the ego vehicle's current position information and the ego vehicle error value in the first coordinate system, thereby obtaining the target detection frame pose information and detection frame error values ​​of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system;

[0062] The tracking, detection and matching module includes:

[0063] a tracking, detection, and matching submodule, configured to match the pose information and detection frame error values ​​of the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system with the pose information and tracking frame error values ​​of the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment, to obtain a first matching result;

[0064] The vehicle posture optimization module includes:

[0065] The ego-vehicle posture optimization submodule is used to use the posture information and tracking frame error value of the first target tracking frame that successfully matched in the first matching result as the ego-vehicle state observation quantity, optimize the ego-vehicle posture information and the ego-vehicle error value, and obtain the optimized ego-vehicle posture information and ego-vehicle error value of the ego-vehicle in the first coordinate system at the current moment.

[0066] Optionally, the detection frame error value of the target detection frame of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system is: the error value after the detection frame error value of the target detection frame of the multiple targets surrounding the ego vehicle at the current moment in the second coordinate system is converted to the first coordinate system, and the sum of the error value caused by the ego vehicle error value in the first coordinate system at the current moment.

[0067] A third aspect of an embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the target positioning method according to the first aspect of the embodiment of the present invention is implemented.

[0068] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the target positioning method of the first aspect of the embodiment of the present invention is implemented.

[0069] Through the target positioning method of the embodiment of the present invention, multiple targets around the ego-vehicle are detected at each moment, and the ego-vehicle posture and multiple targets around the ego-vehicle at each moment are tracked. In the first coordinate system, the target detection frames of the multiple targets around the ego-vehicle at the current moment are matched with the target tracking frames output by the target tracker corresponding to the multiple targets around the ego-vehicle at the previous moment to obtain a first matching result. The first target tracking frame that successfully matches in the first matching result is then used as the ego-vehicle state observation quantity, and the ego-vehicle posture at the current moment is optimized to obtain the optimized ego-vehicle posture at the current moment, thereby optimizing the ego-vehicle posture at each moment, achieving precise positioning of autonomous driving, and improving the accuracy of autonomous driving positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0071] Figure 1 is a flow chart of a target positioning method according to an embodiment of the present invention;

[0072] Figure 2 This is a flow chart of a method for multi-target positioning in a dynamic environment according to an embodiment of the present invention;

[0073] Figure 3 is a structural block diagram of a target positioning device provided by an embodiment of the present invention;

[0074] Figure 4 FIG. 1 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0075] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0076] Reference Figure 1 , Figure 1 This is a flow chart of a target positioning method according to an embodiment of the present invention.

[0077] like Figure 1 As shown, the method may include the following steps:

[0078] Step S11: Obtain target detection frames of multiple targets around the vehicle at the current moment in the second coordinate system, and obtain target tracking frames in the first coordinate system output by the target tracker corresponding to multiple targets around the vehicle at the previous moment.

[0079] In this embodiment, during the autonomous driving process, target detection is performed on multiple targets around the ego vehicle at each moment in the second coordinate system, resulting in target detection frames for the multiple targets around the ego vehicle at the current moment in the second coordinate system. The ego vehicle is the autonomous driving vehicle, and the second coordinate system is the ego vehicle coordinate system, the device coordinate system, or the BEV space coordinate system. In this embodiment, target detection for multiple targets around the ego vehicle may be performed using 3D target detection using the vehicle's sensors.

[0080] In particular, this embodiment does not impose any restrictions on the 3D target detection method. For example, 3D target detection can be single-mode (such as Lidar or Camera) 3D target detection based on the input type (sensor type), or multi-mode (such as Lidar+Camera, Radar+Camera) 3D target detection; it can also be the following four 3D target detection methods based on feature extraction methods: 1. Feature extraction based on original points; 2. Divide the point cloud into grids and then extract the features of the grids. For example, a voxel-based feature extraction method grids the point cloud to obtain regular features and then performs 3D convolution; 3. Use a graph to create a graph for points within a radius R and then extract features; 4. Project the 3D image onto a 2D plane, mostly using the BEV perspective, and then use 2D convolution to extract features. For example, through PointNet++, extract features based on original points and use the farthest point sampling (FPS) algorithm for sampling. Compared with random sampling, this sampling algorithm can better cover the entire sampling space.

[0081] In addition, in the process of autonomous driving, this embodiment will also track multiple targets around the ego vehicle at each moment in the first coordinate system, and obtain the target tracking frame in the first coordinate system output by the target tracker corresponding to the multiple targets around the ego vehicle at each moment. Among them, the first coordinate system is a coordinate system with the starting point of the absolute position of the ego vehicle in the physical world as the origin and the starting direction of the vehicle as the original direction (such as the y-axis). This embodiment can establish a target tracker for each target to track the target. Since the target tracking frame obtained by the target tracker at the current moment is actually a prediction of the position of the target at the next moment, in this embodiment, corresponding to obtaining the target detection frame in the second coordinate system for the multiple targets around the ego vehicle at the current moment is obtaining the target tracking frame in the first coordinate system output by the target tracker corresponding to the multiple targets around the ego vehicle at the previous moment.

[0082] Step S12: According to the driving information of the vehicle, the current position of the vehicle in the first coordinate system is obtained.

[0083] In this embodiment, track inference can be performed in the first coordinate system based on the vehicle's real-time driving information to obtain the vehicle's position in the first coordinate system at each moment. Track inference uses the vehicle's position at a given moment, along with heading and speed information, to infer the vehicle's current position. Based on this, this embodiment can obtain the vehicle's current position in the first coordinate system.

[0084] Step S13: According to the current posture of the vehicle in the first coordinate system, the target detection frames of the multiple targets around the vehicle at the current moment in the second coordinate system are converted to obtain the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system.

[0085] In this embodiment, since target detection is performed in the second coordinate system, the target detection frame in the second coordinate system needs to be converted to the first coordinate system before matching the target detection frame with the target tracking frame. Specifically, the target detection frames of multiple targets surrounding the vehicle at the current moment in the second coordinate system are converted based on the vehicle's current position in the first coordinate system to obtain the target detection frames of multiple targets surrounding the vehicle at the current moment in the first coordinate system.

[0086] Step S14: Match the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment to obtain a first matching result corresponding to the target tracker.

[0087] In this embodiment, after obtaining the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system, the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system can be matched with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment to obtain a first matching result. It will be understood that in this embodiment, each target tracker corresponds to a target tracking frame, and each target tracking frame can obtain a corresponding first matching result after matching the corresponding target detection frame. Therefore, this step can obtain the first matching results corresponding to each of the multiple target trackers.

[0088] For example, the Hungarian matching algorithm can be used to match the target detection frames of multiple targets around the vehicle at the current moment in the first coordinate system with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment to obtain a first matching result.

[0089] Step S15: Using the first target tracking frame that is successfully matched in the first matching result as the vehicle state observation quantity, optimizing the vehicle posture, and obtaining the optimized vehicle posture of the vehicle in the first coordinate system at the current moment.

[0090] In this embodiment, the first successfully matched target tracking frame in the first matching result can be used as the ego vehicle state observation to optimize the ego vehicle's current pose in the first coordinate system, thereby obtaining the optimized ego vehicle pose in the first coordinate system. This embodiment uses the successfully matched first target tracking frame to constrain the ego vehicle pose, thereby achieving more accurate autonomous driving positioning. The first target tracking frame in this embodiment refers to the successfully matched target tracking frame in the first matching result.

[0091] In this embodiment, multiple targets around the ego vehicle at each moment are detected, and the ego vehicle posture at each moment and the multiple targets around the ego vehicle are tracked. In the first coordinate system, the target detection frames of the multiple targets around the ego vehicle at the current moment are matched with the target tracking frames output by the target tracker corresponding to the multiple targets around the ego vehicle at the previous moment to obtain a first matching result. The first target tracking frame that is successfully matched in the first matching result is used as the ego vehicle state observation quantity, and the ego vehicle posture at the current moment is optimized to obtain the optimized ego vehicle posture at the current moment, thereby optimizing the ego vehicle posture at each moment, realizing accurate positioning of autonomous driving, and improving the accuracy of autonomous driving positioning.

[0092] In combination with the above embodiments, in one embodiment, the present invention further provides a target positioning method. In this method, in addition to the above steps, steps S21 to S23 may also be included:

[0093] Step S21: Based on the optimized posture of the vehicle at the current moment in the first coordinate system, the target detection frames of multiple targets around the vehicle at the current moment in the second coordinate system are converted to obtain the optimized target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system.

[0094] In this embodiment, after obtaining the optimized pose of the ego-vehicle at the current moment in the first coordinate system, the target detection frames of the multiple targets surrounding the ego-vehicle at the current moment in the second coordinate system can be transformed based on the optimized pose of the ego-vehicle at the current moment in the first coordinate system to obtain the optimized target detection frames of the multiple targets surrounding the ego-vehicle at the current moment in the first coordinate system. In other words, after obtaining the optimized pose of the ego-vehicle, this embodiment will re-transform the target detection frames of the multiple targets surrounding the ego-vehicle at the current moment in the second coordinate system based on the optimized pose of the ego-vehicle to obtain new target detection frames of the multiple targets surrounding the ego-vehicle at the current moment in the first coordinate system, that is, obtain the optimized target detection frames of the multiple targets surrounding the ego-vehicle at the current moment in the first coordinate system.

[0095] Step S22: Match the target detection frames optimized in the first coordinate system for the multiple targets around the vehicle at the current moment with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment, to obtain a second matching result corresponding to the target tracker.

[0096] In this embodiment, after obtaining the target detection frame optimized for multiple targets around the vehicle at the current moment in the first coordinate system, the target detection frame optimized for multiple targets around the vehicle at the current moment in the first coordinate system can be matched with the target tracking frame in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment to obtain a second matching result corresponding to the target tracker.

[0097] Among them, in this embodiment, the second matching result refers to the matching result obtained by matching the optimized target detection frame and the target tracking frame, the first matching result refers to the matching result obtained by matching the target detection frame and the target tracking frame, and step S22 is the same or similar to the above-mentioned step S14 and is not repeated here.

[0098] Step S23: using the successfully matched second target tracking frame in the second matching result as a target state observation, and updating the target tracker corresponding to the second target tracking frame.

[0099] In this embodiment, the second target tracking frame that is successfully matched in the second matching result can be used as the target state observation to update the target tracker corresponding to the second target tracking frame. In this embodiment, the second target tracking frame refers to the target tracking frame that is successfully matched in the second matching result.

[0100] In this embodiment, the target tracker for target tracking is updated based on the optimized ego-vehicle posture, thereby improving the accuracy of target tracking while improving the ego-vehicle positioning, thereby improving the accuracy of target trajectory prediction, improving the reliability of the intelligent driving system, and improving the safety of complex scenarios.

[0101] In combination with the above embodiments, in one implementation, the present invention further provides a target positioning method. In addition to the above steps, the method may further include steps S31 to S34:

[0102] Step S31: creating a new target tracker for the third target tracking frame that fails to match in the first matching result.

[0103] In this embodiment, if there is a third target tracking frame in the first matching result that fails to match the target detection frame, a new target tracker can be created for the third target tracking frame, thereby managing the new target tracker through the life management policy (steps S32 to S34). The third target tracking frame is the target tracking frame that failed to match in the first matching result. In an optional example, the target tracker originally corresponding to the third target tracking frame can be deleted, and a new target tracker can be created for the third target tracking frame.

[0104] In an optional embodiment, the new target tracker corresponding to the third target tracking frame is a Kalman filter, and the 12-dimensional data x = {θ, P, v, w} of the third target tracking frame's posture, velocity, and angular velocity are used as state quantities. The state transfer matrix is Where I is the 3*3 identity matrix, and the observation matrix is In some embodiments, the state quantity, state transfer matrix, observation matrix, and observation noise may be changed during tracking.

[0105] Step S32: Obtain the first matching results corresponding to the new target tracker at multiple consecutive moments.

[0106] In this embodiment, for a new target tracker, the target tracking frame output by the new tracker in the first coordinate system can be matched with the target detection frame in the first coordinate system at multiple moments in time to obtain a first matching result corresponding to the new target tracker. This step is the same as or similar to step S14 described above. In this embodiment, the first matching results corresponding to the new target tracker can be obtained at multiple consecutive moments in time.

[0107] Step S33: if the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all successful matches, retain the new target tracker.

[0108] In this embodiment, if it is determined that the first matching results corresponding to the new target tracker at multiple consecutive moments are all successful matches, the new target tracker can be retained.

[0109] Step S34: when the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all matching failures, deleting the new target tracker.

[0110] In this embodiment, when it is determined that the first matching results corresponding to the new target tracker at multiple consecutive moments are all matching failures, the new target tracker may be deleted.

[0111] In an optional example, the specific values ​​of multiple consecutive moments can be determined according to needs, for example, the multiple consecutive moments are two consecutive moments (i.e., two consecutive frames) or three consecutive moments (i.e., three consecutive frames). This embodiment does not impose any restrictions on this.

[0112] In this embodiment, a new target tracker can be established for the unmatched third target tracking frame, and the new target tracker can be managed by a life management strategy. The life management strategy can be controlled by many parameters. For example, after the new target tracker is created, it will change to a survival state only after matching 2 frames continuously. If it loses 2 frames continuously, it will change to a dead state, and the tracker will be deleted, etc.

[0113] In addition, the above steps can be restarted at the next moment after the current moment to optimize and update the vehicle posture and target tracker at the next moment.

[0114] In combination with the above embodiments, the present invention further provides a target positioning method. In addition to the above steps, the method may further include step S41 and step S42. Furthermore, the above step S15 may specifically include step S43:

[0115] Step S41: Calculate the residual between the first target tracking frame and the target detection frame that successfully matches the first target tracking frame.

[0116] In this embodiment, a successful match in the first matching result does not mean that the first target tracking frame and its corresponding target detection frame completely overlap, but rather that they are close within a certain range. After determining the first target tracking frame, the residual between the first target tracking frame and the target detection frame that successfully matched the first target tracking frame can be calculated.

[0117] Step S42: Determine the first target tracking frame corresponding to the residual smaller than a preset threshold as the first reference target tracking frame.

[0118] In this embodiment, a preset threshold value may be set. The value of the preset threshold value may be arbitrarily specified and is not limited thereto. After determining the residual between the first target tracking frame and the target detection frame that successfully matches the first target tracking frame, the first target tracking frame corresponding to the residual value that is less than the preset threshold value may be determined as the first reference target tracking frame.

[0119] Step S43: Using the first reference target tracking frame as the ego vehicle state observation quantity, optimizing the ego vehicle posture, and obtaining the optimized ego vehicle posture of the ego vehicle at the current moment in the first coordinate system.

[0120] In this embodiment, first target tracking frames corresponding to residuals greater than or equal to a preset threshold are removed, retaining only the first reference target tracking frame. The first reference target tracking frame is used as the ego vehicle state observation to optimize the ego vehicle pose, obtaining the optimized ego vehicle pose in the first coordinate system at the current moment. In an alternative embodiment, first target tracking frames with large residuals (e.g., residuals greater than or equal to a preset threshold) can be removed using a RANS algorithm.

[0121] In this embodiment, the tracking error of the first target tracking frame (i.e., the residual between the first target tracking frame and the target detection frame that successfully matches the first target tracking frame) is added to the algorithm. The better the convergence of the target tracking frame, the greater the constraint on the vehicle positioning, thereby improving the positioning accuracy.

[0122] In combination with the above embodiments, the present invention further provides a target positioning method. In this method, the above step S12 may specifically include step S51 and step S52:

[0123] Step S51: Obtain the current speed and displacement of the vehicle provided by the inertial measurement unit, and obtain the current actual travel distance provided by the wheel speed sensor.

[0124] In this embodiment, a method for calculating relative displacement (i.e., the vehicle's position) can be achieved by fusing an IMU (Inertial Measurement Unit) and a wheel speed sensor with a Kalman filter. First, the IMU can provide the vehicle's current acceleration and angular velocity information. By integrating the acceleration and angular velocity information, the vehicle's current speed and displacement can be obtained. Furthermore, the wheel speed sensor can directly measure the vehicle's wheel speed, and the actual distance traveled by the vehicle at the current moment can be obtained from the wheel radius and speed.

[0125] Step S52: Using the speed and displacement as state variables and the actual movement distance as an observation value, the vehicle posture of the vehicle at the current moment in the first coordinate system is obtained through Kalman filtering.

[0126] However, because the integration process of the inertial measurement unit (IMU) accumulates errors, the displacement information obtained after long-term integration may have significant deviations. Furthermore, while wheel speed sensors are relatively accurate, their measurements may be affected by conditions such as wheel slip and tire wear.

[0127] Based on this, in this embodiment, in order to comprehensively utilize the advantages of the IMU and wheel speed sensor, a Kalman filter algorithm can be used for data fusion. Kalman filtering is an efficient recursive filter that can obtain the optimal state estimate by fusing multiple observation data in the presence of uncertainty.

[0128] When calculating the ego-vehicle pose through fusion, the speed and displacement provided by the IMU can be used as the state variables of the Kalman filter, and the actual travel distance provided by the wheel speed sensor can be used as the observation value. Through the recursive process of the Kalman filter, the ego-vehicle pose of the current moment in the first coordinate system that integrates the information of the IMU and wheel speed sensors can be obtained.

[0129] In this embodiment, the vehicle's posture is determined by using the high-frequency data of the IMU to provide continuous displacement information, and the integral error of the IMU can be corrected by the measurement value of the wheel speed sensor, thereby obtaining a more accurate vehicle posture.

[0130] In combination with the above embodiments, the present invention further provides a target positioning method, in which the vehicle posture includes: the vehicle posture information and the vehicle error value; the above step S13 may specifically include step S61, the above step S14 may specifically include step S62, and the above step S15 may specifically include step S63:

[0131] Step S61: Based on the posture information and error value of the vehicle in the first coordinate system at the current moment, the posture information and detection frame error value of the target detection frame of multiple targets around the vehicle at the current moment are converted to obtain the posture information and detection frame error value of the target detection frame of multiple targets around the vehicle at the current moment in the first coordinate system.

[0132] In this embodiment, the current moment may be time t, and the position information of the vehicle in the first coordinate system at the current moment obtained by trajectory inference is T 1，t and the vehicle error value is δT 1，tThe pose information of the target detection frame of multiple targets around the vehicle at the current moment in the second coordinate system obtained by target detection can be expressed as: the pose information T of the target detection frame of the jth target 2，j The length, width and height of the target detection box are w, l and h, and the detection box error value of the target detection box of multiple targets around the vehicle in the second coordinate system at the current moment is δT according to the confidence setting (such as the smaller the confidence, the greater the error). 2，j .

[0133] In this embodiment, the point cloud position covariance calculation method in the voxelmap world system can be used, based on the current position information of the vehicle in the first coordinate system as T 1，t and the vehicle error value is δT 1，t , the pose information T of the target detection frame of multiple targets around the vehicle at the current moment in the second coordinate system 2，j And the detection box error value δT 2，j Perform the conversion to obtain the pose information T of the target detection frame of multiple targets around the vehicle at the current moment in the first coordinate system 1，j，t (T 1，j，t The pose information of the target detection frame representing the jth target in the first coordinate system) and the detection frame error value δT 1，j，t .

[0134] For example, the pose information T of the target detection frame of multiple targets around the vehicle at the current moment in the first coordinate system can be determined by the following formula: 1，j，t And the detection box error value δT 1，j，t :

[0135] T 1，j，t =T 1，t *T 2，j ,

[0136] δT 1，j，t =T 1，t *δT 2，j +δT 1，t *T 2，j +δT 1，t *δT 2，j ≈T 1，t *δT 2，j +δT 1，t *T 2，j

[0137] Among them, T 1，t Represents the current position information of the vehicle in the first coordinate system, T 2，j Represents the pose information of the target detection frame of multiple targets around the vehicle at the current moment in the second coordinate system, δT 2，jRepresents the detection frame error value of the target detection frame of multiple targets around the vehicle at the current moment in the second coordinate system, δT 1，t Represents the ego vehicle error value of the ego vehicle in the first coordinate system at the current moment.

[0138] Furthermore, in some embodiments, the pose may be converted into Lie algebra to reduce the amount of computation.

[0139] In an optional embodiment, the detection frame error value δT of the target detection frames of multiple targets around the vehicle at the current moment in the first coordinate system is 1，j，t is: the error value of the target detection frame of multiple targets around the vehicle at the current moment after the detection frame error value in the second coordinate system is converted to the first coordinate system (i.e. T 1，t *δT 2，j ), and the error value caused by the vehicle error value in the first coordinate system at the current moment (i.e., δT 1，t *T 2，j ) and.

[0140] For example, if the current moment corresponds to the first frame, after detecting the target detection frames of multiple targets around the vehicle at the current moment in the second coordinate system, the initial transformation matrix T between the second coordinate system and the first coordinate system can be directly used. 1，t , the target detection frames of multiple targets around the ego vehicle at the current moment in the second coordinate system are converted to obtain the target detection frames of multiple targets around the ego vehicle at the current moment in the first coordinate system, and then corresponding target trackers are established for all target detection frames to perform tracking prediction of the target tracking frames, and then the ego vehicle posture at the next moment (i.e., the second frame) is optimized through the above steps S11 to S15.

[0141] Step S62: Match the pose information and detection frame error values ​​of the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system with the pose information and tracking frame error values ​​of the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment to obtain the first matching result.

[0142] In this embodiment, the tracking result output by the target tracker at the previous moment is actually the prediction of the target detection frame at the current moment. Therefore, in this embodiment, the position information of the target tracking frame in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment is recorded as The tracking frame error value of the target tracking frame in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment is recorded as

[0143] In this embodiment, the position information T of the target detection frame of multiple targets around the vehicle at the current moment in the first coordinate system can be 1，j，t And the detection box error value δT 1，j，t , and, the pose information of the target tracking frame in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment and tracking box error The matching is performed to obtain a first matching result. In this embodiment, the detection frame error value is the error value corresponding to the target detection frame, and the tracking frame error value is the error value corresponding to the target tracking frame. The error in this embodiment may refer to the covariance.

[0144] Step S63: Using the pose information and tracking frame error value of the first target tracking frame that is successfully matched in the first matching result as the vehicle state observation quantity, optimizing the pose information and the error value of the vehicle, and obtaining the optimized pose information and error value of the vehicle in the first coordinate system at the current moment.

[0145] In this embodiment, the pose information of the first target tracking frame that is successfully matched in the first matching result is The tracking frame error value is used as the vehicle state observation value to calculate the vehicle posture information T 1，t and the vehicle error value δT 1，t Optimize and obtain the optimized posture information and error value of the vehicle in the first coordinate system at the current moment.

[0146] For example, the ego vehicle posture information and ego vehicle error value can be updated using the following formula:

[0147]

[0148] in,

[0149]

[0150] And the observation noise is:

[0151] in,

[0152] (θ)∧ is the antisymmetric matrix of θ;

[0153]

[0154]

[0155]

[0156] Among them, T1，t It is a state variable, the observation matrix is ​​H(x), and it can be updated based on the Kalman filter IEKF based on the observation matrix and observation noise.

[0157] In an optional embodiment, only the center point position of the target box can be used as an observation, with reference to the constraints of the laser point cloud on the posture.

[0158] In an optional embodiment, only the (x, y, θ) of the target box can be used as observations, with reference to the constraints of the cubes lam target on the pose.

[0159] In an optional embodiment, other observations may be added, such as lane lines in a high-precision map.

[0160] In combination with the above embodiments, the present invention further provides a target positioning method, in which step S21 may specifically include step S71, step S22 may specifically include step S72, and step S23 may specifically include step S73:

[0161] Step S71: Based on the optimized pose information and the error value of the ego vehicle in the first coordinate system at the current moment, the pose information and the error value of the target detection frame of the multiple targets around the ego vehicle at the current moment are converted in the second coordinate system to obtain the pose information T′ of the target detection frame of the multiple targets around the ego vehicle at the current moment in the first coordinate system. 1，j，t and the detection box error δT′ 1，j，t .

[0162] In this embodiment, after obtaining the posture information and error value of the vehicle after optimization in the first coordinate system at the current moment, the posture information and detection frame error value of the target detection frames of multiple targets around the vehicle at the current moment in the second coordinate system are converted based on the posture information and error value of the vehicle after optimization in the first coordinate system at the current moment, thereby obtaining the posture information and detection frame error value of the target detection frames of multiple targets around the vehicle at the current moment after optimization in the first coordinate system.

[0163] Step S72: Match the pose information and detection frame error value of the target detection frame optimized in the first coordinate system for the multiple targets around the vehicle at the current moment with the pose information and tracking frame error value of the target tracking frame in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment to obtain the second matching result.

[0164] In this embodiment, when performing the second matching, the pose information and detection frame error value of the target detection frame optimized in the first coordinate system for the multiple targets around the vehicle at the current moment can be matched with the pose information and tracking frame error value of the target tracking frame in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment, thereby obtaining a second matching result.

[0165] Step S73: Using the pose information and tracking frame error value of the second target tracking frame that is successfully matched in the second matching result as target state observations, and updating the target tracker corresponding to the second target tracking frame.

[0166] In this embodiment, the pose information of the second target tracking frame that is successfully matched in the second matching result and the tracking frame error value of the second target tracking frame are used as target state observations to update the target tracker corresponding to the second target tracking frame.

[0167] Currently, when performing autonomous driving positioning, some sensors and positioning systems may fail or distort data at night or in inclement weather, resulting in reduced or ineffective positioning accuracy. For example, lidar may be affected by interference in rain and fog, and cameras may lose visibility in darkness. Furthermore, at night or in inclement weather, road signs and lane markings may become blurred, necessitating even higher-precision maps and positioning systems to support autonomous driving. Furthermore, urban roads are often narrow and traffic conditions are complex, placing even higher demands on positioning accuracy. Furthermore, cities may contain numerous structures such as bridges, tunnels, and underpasses, which can affect the reception and transmission of positioning signals, thereby reducing positioning accuracy. Tall buildings can block positioning signals, especially in low-latitude regions, where the impact on GPS signals is more pronounced. Furthermore, tall buildings can cause multipath effects, where signals are reflected and refracted by buildings during propagation, reducing positioning accuracy. In traffic jams, vehicles are close together, increasing interference and obstructing traffic signs such as lane markings, which in turn affects the reception and interpretation of positioning signals, further reducing positioning accuracy.

[0168] On the other hand, with the development of technologies such as deep learning and hardware technologies such as millimeter wave radar and V2X, dynamic target detection technology will become more mature and stable in the future, and using dynamic target positioning under special conditions is also a good choice. Based on this, combined with the above embodiments, in one embodiment, if Figure 2 As shown, Figure 2 This is a flow chart of a multi-target positioning method in a dynamic environment shown in one embodiment of the present invention. This embodiment uses a target positioning method under specific conditions of autonomous driving, involving autonomous driving target tracking and positioning technology.

[0169] exist Figure 2 In the process, the first step is to set a first coordinate system at the starting position of the ego vehicle in the physical world (i.e., the coordinate system with the origin as the starting point of the ego vehicle's absolute position in the physical world and the original direction (e.g., the y-axis) as the starting direction). A vehicle state filter is then established, such as a Kalman filter, to predict and update the ego vehicle's position. For example, the IEKF Kalman filter can be used in the LiDAR and Inertial Measurement Unit (IMU) fusion positioning algorithm.

[0170] Secondly, the vehicle pose transformation and error are obtained based on the trajectory inference in the first coordinate system. It is assumed that the result of the trajectory inference at time t is the vehicle pose and error. In addition, target detection is performed in the second coordinate system at each moment, so that the target detection in the second coordinate system is performed at time t to obtain the pose and error of the target frame in the second coordinate system at time t. Among them, the vehicle coordinate system, device coordinate system or BEV space are collectively referred to as the second coordinate system. At the beginning, the transformation matrix between the first coordinate system and the second coordinate system is T 1，t Furthermore, when time t is the first frame, a tracker corresponding to the target frame detected in the first frame can be established, so that the tracker can predict the target frame of the tracker at each moment.

[0171] Then, based on the vehicle's position and error obtained by trajectory inference at time t, the position and error of the target frame in the second coordinate system at time t are transformed to obtain the position and error of the target frame in the first coordinate system at time t.

[0172] Then, based on the position and error of the target frame in the first coordinate system at time t, the target frame of the tracker at time t (i.e., the tracking frame output by the tracker at time t-1) is predicted and matched by the Hungarian algorithm to determine the first matching result between the target frame and the target frame of the tracker.

[0173] Finally, for the first target tracking frame that is successfully matched in the first matching result, the target frame pose residual is calculated, and the ego vehicle state filter is optimized and updated to obtain the optimized ego vehicle pose and error. Based on the optimized ego vehicle pose and error, the pose and error of the target frame at time t in the second coordinate system are re-converted to update the pose and error of the target frame at time t in the first coordinate system. Finally, based on the updated pose and error of the target frame at time t in the first coordinate system, the target frame tracker is updated.

[0174] In this embodiment, target detection technology is becoming increasingly mature. The target frame in the current frame is detected. The current detection frame is Hungarian matched with the tracker prediction frame. The matched target tracking prediction frame is used as an observation, and the odometry information of other sensors is integrated to optimize the vehicle positioning. The target frame tracker is then corrected. Unmatched detection frames generate new trackers. Unmatched trackers are managed through a life management strategy, forming a cycle that eventually reaches a convergence state. In this way, by continuously updating the vehicle position and the target tracker, not only can the reliability of positioning be improved through the positioning method of this complex scene, but the accuracy of target tracking can also be improved while improving the positioning itself, thereby improving the accuracy of target trajectory prediction, improving the reliability of the intelligent driving system, and improving the safety of complex scenes. In addition, in the case of map-free intelligent driving solutions, if there is no accurate prior information of high-precision maps and the accuracy of online vector map construction is poor, this embodiment will have a wide range of application scenarios. In addition, in one embodiment, the tracking error of the target frame is incorporated into the algorithm. The better the convergence of the target frame, the greater the constraint on the vehicle positioning, thereby improving the positioning accuracy.

[0175] It should be noted that for the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0176] Based on the same inventive concept, an embodiment of the present invention provides a target positioning device 300. Figure 3 , Figure 3 FIG is a structural block diagram of a target positioning device provided by an embodiment of the present invention. Figure 3 As shown, the device 300 includes:

[0177] The target tracking and detection module 301 is used to obtain target detection frames of multiple targets around the vehicle at the current moment in the second coordinate system, and obtain target tracking frames output by the target tracker corresponding to multiple targets around the vehicle at the previous moment in the first coordinate system;

[0178] The ego vehicle posture inference module 302 is used to obtain the ego vehicle posture in the first coordinate system at the current moment based on the ego vehicle's driving information;

[0179] A detection frame conversion module 303 is configured to convert target detection frames of multiple targets surrounding the vehicle at the current moment in the second coordinate system according to the vehicle's current position in the first coordinate system, thereby obtaining target detection frames of the multiple targets surrounding the vehicle at the current moment in the first coordinate system;

[0180] The tracking detection matching module 304 is configured to match the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment, to obtain a first matching result corresponding to the target tracker;

[0181] The ego-vehicle posture optimization module 305 is configured to optimize the ego-vehicle posture by using the first target tracking frame successfully matched in the first matching result as the ego-vehicle state observation, and obtain the optimized ego-vehicle posture of the ego-vehicle in the first coordinate system at the current moment.

[0182] Optionally, the apparatus 300 further includes:

[0183] a detection frame optimization module, configured to convert target detection frames of multiple targets surrounding the ego vehicle at the current moment in the second coordinate system according to the optimized ego vehicle posture at the current moment in the first coordinate system, to obtain target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system;

[0184] an optimization matching module, configured to match the target detection frames optimized in the first coordinate system for the multiple targets surrounding the ego vehicle at the current moment with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment, to obtain a second matching result corresponding to the target tracker;

[0185] The tracker update module is configured to use the successfully matched second target tracking frame in the second matching result as a target state observation and update the target tracker corresponding to the second target tracking frame.

[0186] Optionally, the apparatus 300 further includes:

[0187] A tracker creation module, configured to create a new target tracker for the third target tracking frame that fails to match in the first matching result;

[0188] A matching result acquisition module, configured to acquire first matching results corresponding to the new target tracker at multiple consecutive moments;

[0189] a tracker retaining module, configured to retain the new target tracker if the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all successful matches;

[0190] The tracker deletion module is configured to delete the new target tracker if the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all matching failures.

[0191] Optionally, the apparatus 300 further includes:

[0192] a residual determination module, configured to calculate a residual between the first target tracking frame and a target detection frame that successfully matches the first target tracking frame;

[0193] A tracking frame reference module, configured to determine a first target tracking frame corresponding to a residual smaller than a preset threshold as a first reference target tracking frame;

[0194] The vehicle posture optimization module 305 includes:

[0195] The ego-vehicle posture optimization submodule is used to optimize the ego-vehicle posture using the first reference target tracking frame as the ego-vehicle state observation quantity to obtain the optimized ego-vehicle posture of the ego-vehicle at the current moment in the first coordinate system.

[0196] Optionally, the vehicle posture inference module 302 includes:

[0197] The parameter determination submodule is used to obtain the current vehicle speed and displacement provided by the inertial measurement unit, and to obtain the current actual travel distance provided by the wheel speed sensor;

[0198] The Kalman filter determination submodule is used to use the speed and displacement as state variables and the actual movement distance as an observation value to obtain the vehicle posture of the vehicle in the first coordinate system at the current moment through Kalman filtering.

[0199] Optionally, the ego-vehicle posture includes: ego-vehicle posture information and ego-vehicle error value;

[0200] The detection frame conversion module 303 includes:

[0201] a detection frame conversion submodule, configured to convert the target detection frame pose information and detection frame error values ​​of the multiple targets surrounding the ego vehicle at the current moment in the second coordinate system based on the ego vehicle's current position information and the ego vehicle error value in the first coordinate system, thereby obtaining the target detection frame pose information and detection frame error values ​​of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system;

[0202] The tracking, detection and matching module 304 includes:

[0203] a tracking, detection, and matching submodule, configured to match the pose information and detection frame error values ​​of the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system with the pose information and tracking frame error values ​​of the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment, to obtain a first matching result;

[0204] The vehicle posture optimization module 305 includes:

[0205] The ego-vehicle posture optimization submodule is used to use the posture information and tracking frame error value of the first target tracking frame that successfully matched in the first matching result as the ego-vehicle state observation quantity, optimize the ego-vehicle posture information and the ego-vehicle error value, and obtain the optimized ego-vehicle posture information and ego-vehicle error value of the ego-vehicle in the first coordinate system at the current moment.

[0206] Optionally, the detection frame error value of the target detection frame of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system is: the error value after the detection frame error value of the target detection frame of the multiple targets surrounding the ego vehicle at the current moment in the second coordinate system is converted to the first coordinate system, and the sum of the error value caused by the ego vehicle error value in the first coordinate system at the current moment.

[0207] Based on the same inventive concept, another embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the target positioning method described in any of the above embodiments of the present invention are implemented.

[0208] Based on the same inventive concept, another embodiment of the present invention provides an electronic device 400, such as Figure 4 shown. Figure 4 4 is a schematic diagram of an electronic device according to an embodiment of the present invention. The electronic device includes a memory 402, a processor 401, and a computer program stored in the memory and executable by the processor. When executed by the processor, the computer program implements the steps of the target positioning method according to any of the above embodiments of the present invention.

[0209] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0210] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0211] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, apparatus, or computer program products. Thus, embodiments of the present invention may take the form of a fully hardware embodiment, a fully software embodiment, or an embodiment combining software and hardware. Furthermore, embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0212] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the process in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0213] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0214] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0215] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0216] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0217] The above is a detailed introduction to a target positioning method, device, electronic device and storage medium provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A target positioning method, characterized in that: The method comprises: Obtain target detection frames of multiple targets around the ego vehicle at the current moment in the second coordinate system, and obtain target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the ego vehicle at the previous moment; According to the driving information of the vehicle, the vehicle's current position in the first coordinate system is obtained; According to the current posture of the vehicle in the first coordinate system, the target detection frames of the multiple targets around the vehicle at the current moment in the second coordinate system are converted to obtain the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system; Matching the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment, to obtain a first matching result corresponding to the target tracker; The first target tracking frame is used as the vehicle state observation quantity, and the vehicle posture is optimized to obtain the optimized vehicle posture of the vehicle in the first coordinate system at the current moment. The first target tracking frame is the target tracking frame that is successfully matched in the first matching result.

2. The method according to claim 1, characterized in that The method further comprises: According to the optimized pose of the ego vehicle at the current moment in the first coordinate system, the target detection frames of the multiple targets around the ego vehicle at the current moment in the second coordinate system are converted to obtain the optimized target detection frames of the multiple targets around the ego vehicle at the current moment in the first coordinate system; Matching the target detection frames optimized in the first coordinate system for the multiple targets surrounding the ego vehicle at the current moment with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment, to obtain a second matching result corresponding to the target tracker; The second target tracking frame is used as the target state observation, and the target tracker corresponding to the second target tracking frame is updated, where the second target tracking frame is the target tracking frame that is successfully matched in the second matching result.

3. The method according to claim 1, wherein The method further comprises: creating a new target tracker for a third target tracking frame, where the third target tracking frame is the target tracking frame that fails to be matched in the first matching result; Obtaining first matching results corresponding to the new target tracker at multiple consecutive moments; If the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all successful matches, retaining the new target tracker; If the first matching results corresponding to the new target tracker at the plurality of consecutive moments are all matching failures, the new target tracker is deleted.

4. The method according to claim 1, wherein The method further comprises: Calculating a residual between the first target tracking frame and a target detection frame that successfully matches the first target tracking frame; Determine the first target tracking frame corresponding to the residual smaller than a preset threshold as the first reference target tracking frame; The first target tracking frame is used as the vehicle state observation quantity, and the vehicle posture is optimized to obtain the optimized vehicle posture of the vehicle in the first coordinate system at the current moment, including: The first reference target tracking frame is used as the ego-vehicle state observation quantity, and the ego-vehicle posture is optimized to obtain the optimized ego-vehicle posture of the ego-vehicle at the current moment in the first coordinate system.

5. The method according to claim 1, wherein The step of obtaining the current position of the vehicle in the first coordinate system based on the vehicle's driving information includes: Obtain the current vehicle speed and displacement provided by the inertial measurement unit, and obtain the current actual travel distance provided by the wheel speed sensor; The speed and displacement are used as state variables, the actual travel distance is used as an observation value, and the vehicle posture of the vehicle at the current moment in the first coordinate system is obtained through Kalman filtering.

6. The method according to any one of claims 1 to 5, characterized in that: The vehicle posture includes: the vehicle posture information and the vehicle error value; The converting, based on the current posture of the vehicle in the first coordinate system, the target detection frames of the multiple targets around the vehicle at the current moment in the second coordinate system to obtain the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system includes: Converting the pose information and detection frame error values ​​of the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the second coordinate system according to the pose information and the ego vehicle error value of the ego vehicle at the current moment in the first coordinate system to obtain the pose information and detection frame error values ​​of the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system; The step of matching the target detection frames of the multiple targets around the vehicle at the current moment in the first coordinate system with the target tracking frames in the first coordinate system output by the target tracker corresponding to the multiple targets around the vehicle at the previous moment to obtain a first matching result corresponding to the target tracker includes: Matching the pose information and detection frame error values ​​of the target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system with the pose information and tracking frame error values ​​of the target tracking frames output by the target tracker corresponding to the multiple targets surrounding the ego vehicle at the previous moment in the first coordinate system to obtain a first matching result; The first target tracking frame is used as the vehicle state observation quantity, and the vehicle posture is optimized to obtain the optimized vehicle posture of the vehicle in the first coordinate system at the current moment, including: The pose information and tracking frame error value of the first target tracking frame are used as the vehicle state observation quantity, and the pose information and the error value of the vehicle are optimized to obtain the optimized pose information and error value of the vehicle in the first coordinate system at the current moment.

7. The method according to claim 6, characterized in that The detection frame error value of the target detection frame of the multiple targets surrounding the vehicle at the current moment in the first coordinate system is: the error value after the detection frame error value of the target detection frame of the multiple targets surrounding the vehicle at the current moment in the second coordinate system is converted to the first coordinate system, and the sum of the error value caused by the vehicle error value of the vehicle at the current moment in the first coordinate system.

8. A target positioning device, characterized in that: The target positioning device comprises: A target tracking and detection module is used to obtain target detection frames of multiple targets around the vehicle at the current moment in the second coordinate system, and to obtain target tracking frames in the first coordinate system output by the target tracker corresponding to multiple targets around the vehicle at the previous moment; The ego vehicle posture inference module is used to obtain the ego vehicle posture in the first coordinate system at the current moment based on the ego vehicle's driving information; a detection frame conversion module, configured to convert target detection frames of multiple targets surrounding the ego vehicle at the current moment in the second coordinate system according to the ego vehicle's current position in the first coordinate system, to obtain target detection frames of the multiple targets surrounding the ego vehicle at the current moment in the first coordinate system; a tracking, detection, and matching module configured to match target detection frames of the multiple targets surrounding the vehicle at the current moment in the first coordinate system with target tracking frames output by a target tracker corresponding to the multiple targets surrounding the vehicle at the previous moment in the first coordinate system, to obtain a first matching result corresponding to the target tracker; The ego-vehicle posture optimization module is configured to optimize the ego-vehicle posture using the first target tracking frame as an ego-vehicle state observation to obtain the optimized ego-vehicle posture in the first coordinate system at the current moment, wherein the first target tracking frame is the target tracking frame that is successfully matched in the first matching result.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is executed by the processor, the target positioning method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the target positioning method according to any one of claims 1 to 7 is implemented.

Images