Perception target fusion method, device and equipment

Abstract

The application provides a target fusion method, device and equipment, and relates to the technical field of target detection and tracking. The method comprises the following steps: acquiring a target matching list set at time t; updating an i-1 global sensor priority sequence corresponding to a target object at time t-1 according to a sensor priority sequence in each target matching list, to obtain an i global sensor priority sequence corresponding to the target object at time t; determining a fusion position coordinate of the target object at time t by adding a position coordinate collected by a first sensor at time t and a correction value; determining a deviation value between the fusion position coordinate at time t and position coordinates collected by other sensors in the i global sensor priority sequence at time t, and correcting the position coordinates collected by the other sensors in the i global sensor priority sequence according to the deviation value. The scheme of the application can solve the problems of target position drift, backoff, jitter and target flicker.

Description

Technical Field

[0001] This invention relates to the field of target detection and tracking technology, and in particular to a method, apparatus and device for sensing target fusion. Background Technology

[0002] In complex intersection environments, to achieve the goal of road environment perception beyond line of sight at the entire intersection, multiple sensors need to be deployed on the roadside, including but not limited to multiple cameras, millimeter-wave radar, lidar, and ultrasonic sensors, to achieve all-weather, high-precision, and long-distance perception capabilities for road traffic participants. Effective data fusion between multiple sensors has become a key issue in intersection traffic perception technology.

[0003] Existing target-level fusion methods mainly include weighted averaging, Bayesian methods, and deep learning methods. However, due to significant differences in detection performance between different sensors at different distances—for example, cameras can achieve high accuracy and precision at close range, but at long distances, the target is too small, resulting in decreased positioning and detection accuracy—and as the target moves, the perception state of each sensor changes, easily leading to inconsistencies in the sensor combinations that perceive the target at different times. Whether using weighted averaging, filtering, or deep learning methods, this instability in the continuous motion of the target can easily result in problems such as target position drift, regression, jitter, or target flickering. Summary of the Invention

[0004] The purpose of this invention is to provide a perception target fusion method, apparatus, and device that solves the problems of target position drift, regression, jitter, or target flickering that easily occur in existing target fusion methods.

[0005] Firstly, to achieve the above objectives, embodiments of the present invention provide a perceptual target fusion method, comprising:

[0006] Obtain a target matching list set at time t, wherein the target matching list set includes at least one target pairing list, and each target pairing list includes the pairing relationship between multiple sensors that perceive the same target object and the target ID;

[0007] Based on the sensor priority sequence corresponding to multiple sensors in each target pairing list, update the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 to obtain the i-th global sensor priority sequence at time t, where i is a positive integer; wherein, the i-th global sensor priority sequence includes: the intersection of the target pairing list obtained at time t and the sensors included in the (i-1)th global sensor priority sequence, and the sensors added to the target pairing list at time t relative to the (i-1)th global sensor priority sequence;

[0008] The sum of the position coordinates collected by the first sensor at time t and the correction value is determined as the fused position coordinates of the target object at time t; wherein, the first sensor is the highest priority sensor in the i-th global sensor priority sequence, the correction value is the deviation between the position coordinates collected by the first sensor at time t-1 and the position coordinates collected by the second sensor at time t-1, and the second sensor is the highest priority sensor in the (i-1)-th global sensor priority sequence;

[0009] Determine the deviation between the fused position coordinates at time t and the position coordinates acquired by other sensors in the i-th global sensor priority sequence at time t, and calibrate the position coordinates acquired by other sensors in the i-th global sensor priority sequence based on the deviation.

[0010] In some embodiments of this application, the method for obtaining the sensor priority sequence corresponding to multiple sensors in the target pairing list includes:

[0011] According to the first priority rule, the highest priority sensor is determined from the multiple sensors in the target pairing list;

[0012] The other sensors in the target pairing list are prioritized according to the Euclidean distance between the first position and the second position to obtain the sensor priority sequence; wherein, the first position is the sensing position obtained by the highest priority sensor, and the second position is the sensing position obtained by any of the other sensors.

[0013] In some embodiments of this application, the first priority rule includes at least one of the following:

[0014] If the target object is detected by a single sensor at time t, then the single sensor is determined as the highest priority sensor;

[0015] If the target object is detected by multiple sensors at time t, the sensors are sorted according to their type to obtain the highest priority sensor.

[0016] If the target object is detected by multiple sensors at time t and the multiple sensors are of the same type, then the sensor with the highest confidence level of the detected target is determined as the highest priority sensor; wherein, the target confidence level is positively correlated with the priority level.

[0017] In some embodiments of this application, updating the first global sensor priority sequence corresponding to the target object based on the sensor priority sequence corresponding to multiple sensors in each target pairing list to obtain the second global sensor priority sequence includes:

[0018] If the historical highest priority sensor already saved in the (i-1)th global sensor priority sequence is located in the target pairing list, the historical highest priority sensor is taken as the highest priority sensor, and the sensor priority in the first sensor set is determined according to the second priority rule to obtain the i-th global sensor priority; wherein, the first sensor set includes multiple sensors in the target pairing list and other sensors in the (i-1)th global sensor priority sequence other than the highest priority sensor;

[0019] If the highest-priority historical sensor saved in the (i-1)th global sensor priority sequence is not in the target pairing list, the highest-priority sensor is determined from the second sensor set, and the priorities of other sensors in the second sensor set are determined according to the second priority rule to obtain the i-th global sensor priority; wherein, the second sensor set includes multiple sensors in the target pairing list and other sensors in the (i-1)th global sensor priority sequence other than the highest-priority historical sensor.

[0020] In some embodiments of this application, the second priority rule includes:

[0021] The sensors in the first sensor set are sorted according to the Euclidean distance between the third and fourth positions to determine the sensor priority in the first sensor set; wherein, the third position is the sensing position obtained by the highest priority sensor, and the fourth position is the sensing position obtained by any sensor in the first sensor set.

[0022] In some embodiments of this application, the above-described perceptual target fusion method further includes at least one of the following:

[0023] The fusion speed of the target object at time t is determined based on the average value of the motion speeds sensed by multiple sensors in the target pairing list.

[0024] The fused heading angle of the target object at time t is determined based on the average heading angle sensed by multiple sensors in the target pairing list.

[0025] In some embodiments of this application, the above-described perceptual target fusion method further includes:

[0026] Based on the filtered position coordinates of the target object at time t-1, the fused position coordinates of the target object at time t, and the motion trend coordinates of the target object at time t-1, determine the filtered position coordinates of the target object at time t; or,

[0027] The filtered position coordinates of the target object at time t are determined based on the filtered position coordinates of the target object at time t-1 and the motion trend coordinates of the target object at time t-1.

[0028] In some embodiments of this application, the above-described perceptual target fusion method further includes:

[0029] The historical heading angle of the target object is determined based on the filtered position coordinates of the target object at time t and the position coordinates in the first frame of the historical trajectory of the target object.

[0030] Based on the historical heading angle and the fused heading angle of the target object at time t, the corrected heading angle of the target object at time t is determined.

[0031] In some embodiments of this application, the above-described perceptual target fusion method further includes:

[0032] The historical velocity of the target object is determined based on the filtered position coordinates of the target object at time t, the position coordinates in the first frame of the historical trajectory of the target object, the timestamp corresponding to the first frame of the historical trajectory of the target object, and time t.

[0033] Based on the historical velocity and the fusion velocity of the target object at time t, the correction velocity of the target object at time t is determined.

[0034] Secondly, to achieve the above objectives, embodiments of the present invention provide a sensing target fusion device, comprising:

[0035] The acquisition module is used to acquire a target matching list set at time t. The target matching list set includes at least one target pairing list. Each target pairing list includes the pairing relationship between multiple sensors that perceive the same target object and the target ID.

[0036] An update processing module is used to update the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 based on the sensor priority sequence corresponding to multiple sensors in each target pairing list, to obtain the i-th global sensor priority sequence at time t, where i is a positive integer; wherein, the i-th global sensor priority sequence includes: the intersection of the target pairing list obtained at time t and the sensors included in the (i-1)th global sensor priority sequence, and the sensors added to the target pairing list at time t relative to the (i-1)th global sensor priority sequence;

[0037] The fusion processing module is used to determine the fused position coordinates of the target object at time t by summing the position coordinates collected by the first sensor at time t and the correction value; wherein, the first sensor is the highest priority sensor in the i-th global sensor priority sequence, the correction value is the deviation between the position coordinates collected by the first sensor at time t-1 and the position coordinates collected by the second sensor at time t-1, and the second sensor is the highest priority sensor in the (i-1)-th global sensor priority sequence;

[0038] The calibration module is used to determine the deviation between the fused position coordinates at time t and the position coordinates collected by other sensors in the i-th global sensor priority sequence at time t, and to calibrate the position coordinates collected by other sensors in the i-th global sensor priority sequence according to the deviation.

[0039] Thirdly, to achieve the above objectives, embodiments of the present invention provide a sensing target fusion device, including a processor and a transceiver, wherein the processor is used for:

[0040] Obtain a target matching list set at time t, wherein the target matching list set includes at least one target pairing list, and each target pairing list includes the pairing relationship between multiple sensors that perceive the same target object and the target ID;

[0041] Based on the sensor priority sequence corresponding to multiple sensors in each target pairing list, update the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 to obtain the i-th global sensor priority sequence at time t, where i is a positive integer; wherein, the i-th global sensor priority sequence includes: the intersection of the target pairing list obtained at time t and the sensors included in the (i-1)th global sensor priority sequence, and the sensors added to the target pairing list at time t relative to the (i-1)th global sensor priority sequence;

[0042] The sum of the position coordinates collected by the first sensor at time t and the correction value is determined as the fused position coordinates of the target object at time t; wherein, the first sensor is the highest priority sensor in the i-th global sensor priority sequence, and the correction value is the difference between the position coordinates collected by the first sensor at time t-1 and the position coordinates collected by the second sensor at time t-1, and the second sensor is the highest priority sensor in the (i-1)-th global sensor priority sequence.

[0043] Determine the deviation between the fused position coordinates at time t and the position coordinates acquired by other sensors in the i-th global sensor priority sequence at time t, and calibrate the position coordinates acquired by other sensors in the i-th global sensor priority sequence based on the deviation.

[0044] Fourthly, to achieve the above objectives, embodiments of the present invention provide a mobile terminal, including a transceiver, a processor, a memory, and a program or instructions stored in the memory and executable on the processor; when the processor executes the program or instructions, it implements the perception target fusion method as described above.

[0045] Fifthly, to achieve the above objectives, embodiments of the present invention provide a readable storage medium having a program or instructions stored thereon, which, when executed by a processor, implement the steps in the perception target fusion method as described above.

[0046] The beneficial effects of the above-mentioned technical solution of the present invention are as follows:

[0047] The method of this embodiment of the invention obtains a target matching list set at time t, which includes at least one target pairing list. Each target pairing list includes pairing relationships between multiple sensors sensing the same target object and the target ID. Based on the sensor priority sequence corresponding to the multiple sensors in each target pairing list, the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 is updated to obtain the i-th global sensor priority sequence at time t, where i is a positive integer. The i-th global sensor priority sequence includes: the intersection of the target pairing list obtained at time t and the sensors included in the (i-1)th global sensor priority sequence, and the target pairing list at time t relative to the (i-1)th global sensor priority sequence. The system adds a new sensor to the sensor priority sequence. The sum of the position coordinates acquired by the first sensor at time t and the correction value is used to determine the fused position coordinates of the target object at time t. Here, the first sensor is the highest priority sensor in the i-th global sensor priority sequence, and the correction value is the difference between the position coordinates acquired by the first sensor at time t-1 and the position coordinates acquired by the second sensor at time t-1. The second sensor is the highest priority sensor in the (i-1)-th global sensor priority sequence. The system determines the deviation between the fused position coordinates at time t and the position coordinates acquired by other sensors in the i-th global sensor priority sequence at time t, and calibrates the position coordinates acquired by the other sensors in the i-th global sensor priority sequence based on the deviation. In this embodiment, by maintaining the global sensor priority sequence of the target object, it is possible to switch to a higher priority sensor in a timely manner. Simultaneously, by calculating the position difference between the old and new highest priority sensors at time t-1, the current position acquired by the new highest priority sensor can be compensated when switching to the highest priority sensor. This eliminates or reduces position shifts caused by inherent sensor biases, thus avoiding the accumulation of drift; and enables a smooth transition based on historical data, avoiding sudden jumps, regressions, or jitters. On the other hand, by maintaining a dynamic global sensor priority sequence, even if a single sensor temporarily loses the target due to environmental occlusion, brief malfunction, or limited field of view, as long as other sensors still report the target in the target pairing list, the system can update the priority sequence and select the currently available best sensor to maintain tracking of the target object. This avoids the target object being mistakenly judged as disappeared by the system due to a temporary failure of a sensor, thus preventing flickering of the target trajectory. Furthermore, by using the fused position coordinates at time t to calibrate the position coordinates collected by other sensors at time t, when the highest priority sensor switches in the future, the position data provided by other sensors will be more consistent with the overall trajectory because they have been periodically calibrated with the system's current best estimate. This makes the switching of the highest priority sensor smoother, further reducing backoff and jitter. Attached Figure Description

[0048] Figure 1 This is a flowchart of the perception target fusion method according to an embodiment of the present invention;

[0049] Figure 2 This is a structural diagram of the perception target fusion device according to an embodiment of the present invention;

[0050] Figure 3 This is a structural diagram of the sensing target fusion device according to an embodiment of the present invention;

[0051] Figure 4 This is a structural diagram of a sensing target fusion device according to another embodiment of the present invention. Detailed Implementation

[0052] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0053] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of the invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0054] In various embodiments of the present invention, it should be understood that the sequence number of each process described below does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0055] In addition, the terms "system" and "network" are often used interchangeably in this article.

[0056] In the embodiments provided in this application, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean that B is determined solely based on A, but can also be determined based on A and / or other information.

[0057] like Figure 1 As shown, a perceptual target fusion method according to an embodiment of the present invention includes the following steps:

[0058] Step 101: Obtain the target matching list set at time t. The target matching list set includes at least one target pairing list. Each target pairing list includes the pairing relationship between multiple sensors that perceive the same target object and the target ID.

[0059] In some application scenarios, assuming an intersection has multiple sensors, before performing step 101, the intrinsic and extrinsic parameters of each sensor are first calibrated, and the coordinates of the sensor output data are converted to world physical coordinates. Based on the sensor's optimal sensing range, a Region of Interest (ROI) is defined for each sensor, ensuring that the image within the ROI is free from occlusion and shadows. For camera sensors, the distance within the ROI is within 80m to guarantee the detection and positioning accuracy of each sensor.

[0060] In specific implementation, the sensor acquires perception parameters in the following ways: For each frame of sensor data, deep learning methods are first used to detect targets, outputting target parameters such as timestamp, target type, target location coordinates, target confidence, and lane ID. Detected targets include, but are not limited to, common traffic participants such as cars, buses, trucks, motorcycles, bicycles, pedestrians, and obstacles. Detection models include, but are not limited to, deep learning network models such as the YOLO series, R-CNN series, and Transformer. Then, multi-target tracking algorithms (including but not limited to KCF, sort series, TransTrack, and DETR algorithms) are used to track the positions of each target, assigning a unique ID to each perceived target to form its motion trajectory. The target location coordinates are then transformed to the world coordinate system using camera intrinsic and extrinsic parameters. Based on the motion trajectory, the target's velocity and heading angle are calculated to obtain the target list information perceived by the sensor in the current frame. Each of the targets It includes the following perception parameters: target ID, timestamp, target type, target location coordinates, target confidence level, lane ID, target speed, and heading angle. That is, .

[0061] It should be noted that, due to the different timestamps of the targets output by each sensor, in practice, before target matching, all targets provided by all sensors are iterated, their timestamps are compared, and the largest timestamp is recorded as the unified time reference point for the entire sensor group, i.e., time t. The timestamps of all compensated targets are then uniformly updated to the determined latest timestamp t. Furthermore, for each target provided by a sensor, the difference between its own timestamp and the determined latest timestamp t is calculated; using the target's current position, velocity, and heading angle, combined with the time difference, the displacement of the target during this time is calculated using kinematic formulas, and its position is updated.

[0062] Furthermore, construct Matching pairs The target matching process links sensor data from different sensors that may describe the same physical target together to obtain a target matching list set. , Representing the target in the physical world Detected by multiple sensors, the target was detected. of Create a list. For example This indicates that there is a target in the physical world that is simultaneously detected by sensor 1 and sensor 2. The target ID in sensor 1 is 1, and the target ID in sensor 2 is 5.

[0063] In practice, the global target list is first set to be equal to the target list of any sensor. For example, the initial global target list set is the target list of sensor 1, i.e.:

[0064] Global target list collection .

[0065] The Hungarian matching algorithm is used to match the target lists of other sensors with the global target list in turn, and the matched targets are added to the corresponding matching pair list. For targets added by other sensors, create a new list of matching pairs.

[0066] For example, global goals If a target 1 from sensor 2 is matched, and sensor 2 adds a new target m, then the current global target list set is updated to... This process continues until all sensor target matching is completed, resulting in a list of matching pairs in the current frame containing information on all targets and sensors. ;

[0067] in, For target object 1, a list of target pairs;

[0068] This is the target pairing list for target object 2.

[0069] Step 102: Based on the sensor priority sequence corresponding to multiple sensors in each target pairing list, update the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 to obtain the i-th global sensor priority sequence at time t, where i is a positive integer; wherein, the i-th global sensor priority sequence includes: the intersection of the target pairing list obtained at time t and the sensors included in the (i-1)th global sensor priority sequence, and the sensors added to the target pairing list at time t relative to the (i-1)th global sensor priority sequence.

[0070] In this step, at each time step t, a latest, priority-sorted list of sensors is generated for the specific target object, i.e., the i-th global sensor priority sequence. Similarly, at each time step t-1, a latest, priority-sorted list of sensors is generated for the specific target object, i.e., the (i-1)-th global sensor priority sequence.

[0071] It is understandable that the intersection of the target pairing list obtained at time t and the sensors included in the (i-1)th global sensor priority sequence consists of sensors that continuously track the target object and were part of the global priority sequence before time t. Sensors added to the target pairing list at time t relative to the (i-1)th global sensor priority sequence do not exist in the (i-1)th global sensor priority sequence at time t-1.

[0072] For example, at time t-1, the priority sequence of the (i-1)th global sensor of target object A is: {S1 (high), S2 (medium), S3 (low)}; at time t, the system obtains the target pairing list corresponding to target object A as: {S1, S4, S2}, and the priority order is: [S4 (highest), S1 (second highest), S2 (lowest)]. Then the intersection of {S1 (high), S2 (medium), S3 (low)} and {S1, S4, S2} is {S1, S2}; the newly added sensor at time t is {S4}; at time t, the priority sequence of the i-th global sensor of target object A is: {S4, S1, S2}.

[0073] In this step, by maintaining a global sensor priority sequence for the target object, it can be ensured that when performing subsequent location fusion, the system can always prioritize the sensor with the best current state, the most reliable, and the most relevant, thus laying the foundation for the accuracy and stability of the target location.

[0074] Step 103: The sum of the position coordinates collected by the first sensor at time t and the correction value is determined as the fused position coordinates of the target object at time t; wherein, the first sensor is the highest priority sensor in the i-th global sensor priority sequence, the correction value is the deviation between the position coordinates collected by the first sensor at time t-1 and the position coordinates collected by the second sensor at time t-1, and the second sensor is the highest priority sensor in the (i-1)-th global sensor priority sequence.

[0075] In this step, by maintaining a global sensor priority sequence for the target object, the system can promptly switch to a higher-priority sensor. Simultaneously, by calculating the position difference between the old and new highest-priority sensors at time t-1, the system can compensate for the current position acquired by the new highest-priority sensor when switching. If a sensor experiences long-term error accumulation, the system can promptly switch to a higher-priority sensor if other more accurate sensors are available. Furthermore, by calculating correction values, the system can compensate for the current position of the new primary sensor when switching, eliminating or reducing positional shifts caused by inherent sensor biases, thereby preventing the accumulation of drift.

[0076] Furthermore, the correction value compensates for the inherent deviation between the old and new highest priority sensors at time t-1. By compensating based on the correction value, it can be ensured that when the highest priority sensor is switched, the position of the target object will not suddenly jump or regress due to the inconsistency of the measurement reference between the sensors.

[0077] In addition, by maintaining a dynamic global sensor priority sequence, even if a single sensor temporarily loses the target due to environmental occlusion, brief failure, or limited field of view, as long as other sensors still report the target in the target pairing list, the system can update the priority sequence and select the best available sensor to maintain tracking of the target object. This avoids the target object being mistakenly judged as disappeared by the system due to the temporary failure of a certain sensor, thereby preventing the target trajectory from flickering.

[0078] Step 104: Determine the deviation between the fused position coordinates at time t and the position coordinates acquired by other sensors in the i-th global sensor priority sequence at time t, and calibrate the position coordinates acquired by other sensors in the i-th global sensor priority sequence based on the deviation.

[0079] In specific implementation, after determining that the highest priority sensor sensing target object A at time t is the first sensor, the sum of the position coordinates collected by the first sensor at time t and the correction value is determined as the fused position coordinates of the target object at time t. The offset value is then determined between the fused position coordinates and the sensing positions collected by other sensors in the i-th global sensor priority sequence that detected target object A. The position coordinates collected by other sensors in the i-th global sensor priority sequence are then calibrated based on the offset value. The offset value includes:

[0080] ;

[0081] ;

[0082] ;

[0083] in, , , Sensors The offset values ​​relative to the first sensor in the x, y, and z directions.

[0084] In this embodiment, by calibrating the sensing positions of other sensors in the i-th global sensor priority sequence, if the current first sensor fails or becomes unavailable at time t+1, the system needs to select a new highest priority sensor from the other sensors. Since the other sensors have been periodically calibrated with the system's current best estimate, the position data they provide when promoted to the highest priority sensor will be more consistent with the overall trajectory, thus making the switching of the highest priority sensor smoother and further reducing backoff and jitter issues.

[0085] Moreover, by aligning the data from all sensors to a common and most reliable reference point through calibration, the data consistency of the entire sensor network can be improved, thereby reducing jitter caused by differences between sensors in target fusion.

[0086] In some embodiments of this application, the sensor priority sequence corresponding to multiple sensors in the target pairing list is obtained in the following ways:

[0087] According to the first priority rule, the highest priority sensor is determined from multiple sensors in the target pairing list;

[0088] The other sensors in the target pairing list are prioritized according to the Euclidean distance between the first position and the second position to obtain the sensor priority sequence; wherein, the first position is the sensing position obtained by the highest priority sensor, and the second position is the sensing position obtained by any of the other sensors.

[0089] In this embodiment, after determining the highest priority sensor from multiple sensors in the target pairing list, the priority of the remaining sensors can be sorted according to the Euclidean distance between their sensing positions and the sensing positions of the highest priority sensor, with sensors having higher priority due to smaller distances. Specifically, the Euclidean distance between the first position and the second position... for:

[0090] ;

[0091] in, The highest priority sensor in the target pairing list senses the location coordinates of the target object, i.e., the first location; , , Pair other sensors in the target list The position coordinates of the target object are perceived, i.e., the second position.

[0092] In some embodiments of this application, the first priority rule includes at least one of the following:

[0093] If the target object is detected by a single sensor at time t, then that single sensor is determined as the highest priority sensor.

[0094] If the target object is detected by multiple sensors at time t, the sensors are sorted according to their type to obtain the highest priority sensor.

[0095] If the target object is detected by multiple sensors at time t and the multiple sensors are of the same type, then the sensor with the highest confidence level of the detected target is determined as the highest priority sensor; wherein, the target confidence level is positively correlated with the priority level.

[0096] For example, when target object 3 first appears within the perception range of an intersection in the physical world, it is detected by the intersection's sensors at time t, meaning the target is detected for the first time. If target object 3 is detected by a single sensor, that single sensor is the highest priority sensor corresponding to target object 3. If target object 3 is detected by multiple sensors, the sensors are prioritized according to their type (e.g., LiDAR > millimeter-wave radar > camera; priority can be configured based on empirical values), and the highest priority sensor is determined. If target object 3 is detected by multiple sensors of the same type, they are prioritized according to the confidence level (conf) of each sensor detecting the target (sensors with higher confidence levels have higher priority), and the highest priority sensor is determined.

[0097] In some embodiments of this application, in step 102 above, updating the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 according to the sensor priority sequence corresponding to multiple sensors in each target pairing list to obtain the i-th global sensor priority sequence corresponding to time t includes the following two cases:

[0098] Case 1: If the historical highest priority sensor already saved in the (i-1)th global sensor priority sequence is in the target pairing list, the historical highest priority sensor is taken as the highest priority sensor, and the sensor priority in the first sensor set is determined according to the second priority rule to obtain the i-th global sensor priority; wherein, the first sensor set includes multiple sensors in the target pairing list and other sensors in the (i-1)th global sensor priority sequence except for the highest priority sensor.

[0099] In this scenario, the highest priority sensor in the (i-1)th global sensor priority sequence is already stored in the target pairing list. This means that the target object is still being tracked by the highest priority sensor at time t-1. By using the highest priority sensor at time t-1 as the highest priority sensor at time t, the highest priority sensor at time t is the same as that at time t-1, meaning that no highest priority sensor switching has occurred. This reduces the risk of drift caused by the accumulation of inherent system errors between different sensors and eliminates any potential position jumps or backtracking problems caused by the switching of the highest priority sensor.

[0100] It should be noted that, in the absence of a switch to the highest priority sensor, the correction value in step 103 is 0, and there will be no jump problem due to the difference in measurement references between different sensors.

[0101] Case 2: If the highest priority sensor in the saved historical sequence is not in the target pairing list, determine the highest priority sensor from the second sensor set, and determine the priorities of other sensors in the second sensor set according to the second priority rule to obtain the i-th global sensor priority; wherein the second sensor set includes multiple sensors in the target pairing list and other sensors in the i-1 global sensor priority sequence other than the highest priority sensor in the historical sequence.

[0102] In scenario two, the historical highest-priority sensor saved in the (i-1)th global sensor priority sequence is not in the target pairing list. This means that the historical highest-priority sensor at time t-1 cannot continuously track the target object at time t. By re-determining a new highest-priority sensor, it is ensured that the target object's position estimation is based on the best available data after the highest-priority sensor is replaced. By introducing a correction value to compensate for the position acquired by the switched highest-priority sensor, the potential rollback problem caused by the main sensor switch can be resolved, ensuring that the target object's position will not experience a sudden "jump" or "rollback" due to inconsistencies in the measurement benchmarks between the old and new highest-priority sensors when the highest-priority sensor is switched.

[0103] In some embodiments of this application, the second priority rule includes:

[0104] The sensors in the first sensor set are sorted according to the Euclidean distance between the third and fourth positions to determine the sensor priority in the first sensor set; wherein, the third position is the sensing position acquired by the highest priority sensor, and the fourth position is the sensing position acquired by any sensor in the first sensor set.

[0105] The Euclidean distance between the third and fourth positions. for:

[0106] ;

[0107] in, The position coordinates of the target object perceived by the highest priority sensor at time t, i.e., the third position; , , For any sensor in the first sensor set The position coordinates of the target object perceived at time t, i.e., the fourth position.

[0108] It should be noted that, regardless of whether a switch occurs in scenario one (no highest priority sensor switch) or scenario two (a switch occurs), the other sensors in the i-th global sensor priority sequence, excluding the new highest priority sensor, are ordered according to their distance from this new highest priority sensor. This ensures that even after a switch occurs, the subsequent backup sensors are matched with the new highest priority sensor. In this way, when a switch occurs, the accurate application of the correction value can be guaranteed, improving the trajectory stability of the target object.

[0109] Furthermore, by updating the priorities of other sensors based on their distance to the highest-priority sensor, even if the highest-priority sensor temporarily malfunctions or becomes obstructed at time t, it can still maintain a higher position in the priority sequence as long as other sensors can provide position information that is relatively consistent with the current highest-priority sensor. This helps maintain continuous tracking of the target, prevents the entire target from being lost due to the failure of a single sensor, and thus prevents target flickering.

[0110] In some embodiments of this application, the above-described perceptual target fusion method further includes at least one of the following:

[0111] The fusion speed of the target object at time t is determined based on the average motion speed sensed by multiple sensors in the target pairing list.

[0112] The fused heading angle of the target object at time t is determined based on the average heading angle sensed by multiple sensors in the target pairing list.

[0113] In this embodiment, by averaging the motion speed and lateral angle sensed by multiple sensors in the target pairing list, the speed and heading angle can be fused more accurately, improving the accuracy of target motion prediction. Stable speed and heading angle help to more accurately predict the approximate position of the target at the next moment, thereby improving the accuracy and stability of target matching.

[0114] In some embodiments of this application, the above-described perceptual target fusion method further includes:

[0115] Based on the filtered position coordinates of the target object at time t-1, the fused position coordinates of the target object at time t, and the motion trend coordinates of the target object at time t-1, determine the filtered position coordinates of the target object at time t; or,

[0116] Based on the filtered position coordinates of the target object at time t-1 and the motion trend coordinates of the target object at time t-1, determine the filtered position coordinates of the target object at time t.

[0117] It should be noted that at each time t, the system expects to receive new sensor data (i.e., the "current frame") and calculates the fused position, fused motion speed, fused heading angle, etc. at time t based on this data and the previous state.

[0118] As one implementation method, assuming no current frame is lost, the filtered position coordinates of the target object at time t are determined according to Formula 1. Formula 1 includes:

[0119] ;

[0120] ;

[0121] ;

[0122] in, Let be the filtered position coordinates of the target object at time t. Let be the fused position coordinates of the target object at time t. Let be the filtered position coordinates of the target object at time t-1. The coordinates represent the motion trend of the target object at time t-1. These are the position filtering coefficients. These are the motion trend filtering coefficients. and The value range is [0, 1.0].

[0123] Among them, the motion trend coordinates of the target object at time t-1 refer to the predicted position coordinates of the target object at time t based on the motion state and motion model at time t-1.

[0124] In this embodiment, "the current frame is not lost" means that at time t, the sensor successfully acquires a new data frame, and this data frame contains valid observation information of the current target object. This can be understood as the target object being successfully detected and matched by at least one sensor at time t. In this case, the fused position coordinates of the target object at time t can be obtained based on the current frame; then, based on the filtered position coordinates of the target object at time t-1, the fused position coordinates of the target object at time t, and the motion trend coordinates of the target object at time t, the filtered position coordinates of the target object at time t can be determined.

[0125] In some other implementations, when the current frame is lost, the filtered position coordinates of the target object at time t are determined according to Formula 2. Formula 2 includes:

[0126] ;

[0127] ;

[0128] ;

[0129] in, The predicted position coordinates of the target object at time t; The filtered position coordinates of the target object at time t-1; The coordinates represent the motion trend of the target object at time t-1; The coordinates of the target object's motion trend at time t; Let be the filtered position coordinates of the target object at time t; As the attenuation factor, The value range is [0, 1.0].

[0130] It should be pointed out that, It can be used to calculate the filtered position of the target object at time t+1.

[0131] In this embodiment, "current frame loss" refers to the situation where, at time t, the sensor fails to acquire a new data frame, or although a data frame is acquired, it does not contain valid observation information about the current target object. This means that at time t, the system cannot obtain any new valid sensing data about the target object from the current frame. Therefore, the fused position coordinates at time t cannot be calculated. In this case, the system lacks new sensor measurement data to correct the prediction and can only rely on the filtered position coordinates of the target object at time t-1 and the motion trend coordinates at time t to estimate the filtered position coordinates of the target object at time t.

[0132] In some embodiments of this application, the above-described perceptual target fusion method further includes:

[0133] The historical heading angle of the target object is determined based on the filtered position coordinates of the target object at time t and the position coordinates in the first frame of the historical trajectory of the target object.

[0134] Based on the historical heading angle and the fused heading angle of the target object at time t, the corrected heading angle of the target object at time t is determined.

[0135] For example, the corrected heading angle of the target object at time t is determined according to Formula 3;

[0136] Formula 3 includes:

[0137] ;

[0138] ;

[0139] in, It is the arctangent function. , These are the position coordinates in the first frame of data in the historical trajectory of the target object. , Let be the filtered position coordinates of the target object at time t. The corrected heading angle of the target object at time t. Let be the fused heading angle of the target object at time t. As a historical compass, This is the heading angle correction factor. The value range is [0, 1.0].

[0140] In this embodiment, the corrected heading angle of the target object at time t is obtained by weighted summation of the historical heading angle and the fused heading angle of the target object at time t.

[0141] In some embodiments of this application, the above-described perceptual target fusion method further includes:

[0142] The historical velocity of the target object is determined based on the filtered position coordinates of the target object at time t, the position coordinates in the first frame of the historical trajectory of the target object, the timestamp corresponding to the first frame of the historical trajectory of the target object, and time t.

[0143] Based on the historical velocity and the fusion velocity of the target object at time t, the correction velocity of the target object at time t is determined.

[0144] For example, the velocity of the target object at time t can be corrected according to Formula 4 to obtain the corrected velocity of the target object at time t. ;

[0145] Formula four includes:

[0146] ;

[0147] ;

[0148] in,( , , The position coordinates in the first frame of the target object's historical trajectory data, ( , , Let be the filtered position coordinates of the target object at time t. Let be the corrected velocity of the target object at time t. Let be the fusion speed of the target object at time t; The timestamp t corresponding to the current frame. This refers to the timestamp corresponding to the first frame of data in the historical trajectory of the target object. For speed correction factor, The value range is [0, 1.0].

[0149] In this embodiment, the corrected speed of the target object at time t is obtained by weighted summation of the historical speed and the fusion speed of the target object at time t.

[0150] like Figure 2 As shown, an embodiment of the present invention provides a sensing target fusion device, the device 200 including:

[0151] The acquisition module 201 is used to acquire a target matching list set at time t. The target matching list set includes at least one target pairing list, and each target pairing list includes the pairing relationship between multiple sensors that perceive the same target object and the target ID.

[0152] The update processing module 202 is used to update the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 according to the sensor priority sequence corresponding to multiple sensors in each target pairing list, so as to obtain the i-th global sensor priority sequence at time t, where i is a positive integer; wherein, the i-th global sensor priority sequence includes: the intersection of the target pairing list obtained at time t and the sensors included in the (i-1)th global sensor priority sequence, and the sensors added to the target pairing list at time t relative to the (i-1)th global sensor priority sequence;

[0153] The fusion processing module 203 is used to determine the fused position coordinates of the target object at time t by summing the position coordinates collected by the first sensor at time t and the correction value; wherein, the first sensor is the highest priority sensor in the i-th global sensor priority sequence, the correction value is the deviation between the position coordinates collected by the first sensor at time t-1 and the position coordinates collected by the second sensor at time t-1, and the second sensor is the highest priority sensor in the (i-1)-th global sensor priority sequence;

[0154] The calibration module 204 is used to determine the deviation between the fused position coordinates at time t and the position coordinates collected by other sensors in the i-th global sensor priority sequence at time t, and to calibrate the position coordinates collected by other sensors in the i-th global sensor priority sequence according to the deviation.

[0155] Optionally, the acquisition module 201 is also used for:

[0156] According to the first priority rule, the highest priority sensor is determined from the multiple sensors in the target pairing list;

[0157] The other sensors in the target pairing list are prioritized according to the Euclidean distance between the first position and the second position to obtain the sensor priority sequence; wherein, the first position is the sensing position obtained by the highest priority sensor, and the second position is the sensing position obtained by any of the other sensors.

[0158] Optionally, the first priority rule includes at least one of the following:

[0159] If the target object is detected by a single sensor at time t, then the single sensor is determined as the highest priority sensor;

[0160] If the target object is detected by multiple sensors at time t, the sensors are sorted according to their type to obtain the highest priority sensor.

[0161] If the target object is detected by multiple sensors at time t and the multiple sensors are of the same type, then the sensor with the highest confidence level of the detected target is determined as the highest priority sensor; wherein, the target confidence level is positively correlated with the priority level.

[0162] Optionally, update processing module 202 is specifically used for:

[0163] If the historical highest priority sensor already saved in the (i-1)th global sensor priority sequence is located in the target pairing list, the historical highest priority sensor is taken as the highest priority sensor, and the sensor priority in the first sensor set is determined according to the second priority rule to obtain the i-th global sensor priority; wherein, the first sensor set includes multiple sensors in the target pairing list and other sensors in the (i-1)th global sensor priority sequence other than the highest priority sensor;

[0164] If the highest-priority historical sensor saved in the (i-1)th global sensor priority sequence is not in the target pairing list, the highest-priority sensor is determined from the second sensor set, and the priorities of other sensors in the second sensor set are determined according to the second priority rule to obtain the i-th global sensor priority; wherein, the second sensor set includes multiple sensors in the target pairing list and other sensors in the (i-1)th global sensor priority sequence other than the highest-priority historical sensor.

[0165] Optionally, the second priority rule includes:

[0166] The sensors in the first sensor set are sorted according to the Euclidean distance between the third and fourth positions to determine the sensor priority in the first sensor set; wherein, the third position is the sensing position obtained by the highest priority sensor, and the fourth position is the sensing position obtained by any sensor in the first sensor set.

[0167] Optionally, the sensing target fusion device 200 further includes at least one of the following:

[0168] The second determining module is used to determine the fusion speed of the target object at time t based on the average value of the motion speeds sensed by multiple sensors in the target pairing list;

[0169] The third determining module is used to determine the fused heading angle of the target object at time t based on the average heading angle sensed by multiple sensors in the target pairing list.

[0170] Optionally, the sensing target fusion device 200 further includes:

[0171] The fourth determining module is used to determine the filtered position coordinates of the target object at time t based on the filtered position coordinates of the target object at time t-1, the fused position coordinates of the target object at time t, and the motion trend coordinates of the target object at time t-1; or,

[0172] The fifth determining module is used to determine the filtered position coordinates of the target object at time t based on the filtered position coordinates of the target object at time t-1 and the motion trend coordinates of the target object at time t-1.

[0173] Optionally, the sensing target fusion device 200 further includes:

[0174] The sixth determining module is used to determine the historical heading angle of the target object based on the filtered position coordinates of the target object at time t and the position coordinates in the first frame of the historical trajectory of the target object;

[0175] The seventh determining module is used to determine the corrected heading angle of the target object at time t based on the historical heading angle and the fused heading angle of the target object at time t.

[0176] Optionally, the sensing target fusion device 200 further includes:

[0177] The eighth determining module is used to determine the historical velocity of the target object based on the filtered position coordinates of the target object at time t, the position coordinates in the first frame of the historical trajectory of the target object, the timestamp corresponding to the first frame of the historical trajectory of the target object, and time t.

[0178] The ninth determining module is used to determine the correction speed of the target object at time t based on the historical speed and the fusion speed of the target object at time t.

[0179] like Figure 3 As shown, a perception target fusion device 300 according to an embodiment of the present invention includes a processor 310 and a transceiver 320, wherein,

[0180] The processor 310 is configured to acquire a target matching list set at time t, wherein the target matching list set includes at least one target pairing list, and each target pairing list includes pairing relationships between multiple sensors that perceive the same target object and the target ID; based on the sensor priority sequence corresponding to the multiple sensors in each target pairing list, update the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 to obtain the i-th global sensor priority sequence corresponding to time t, where i is a positive integer; wherein the i-th global sensor priority sequence includes: the intersection of the target pairing list acquired at time t and the sensors included in the (i-1)th global sensor priority sequence, and the priority sequence of the target pairing list at time t relative to the (i-1)th global sensor priority sequence. A new sensor is added to the priority sequence; the sum of the position coordinates collected by the first sensor at time t and the correction value is determined as the fused position coordinates of the target object at time t; wherein, the first sensor is the highest priority sensor in the i-th global sensor priority sequence, the correction value is the deviation between the position coordinates collected by the first sensor at time t-1 and the position coordinates collected by the second sensor at time t-1, and the second sensor is the highest priority sensor in the (i-1)-th global sensor priority sequence; the deviation between the fused position coordinates at time t and the position coordinates collected by other sensors in the i-th global sensor priority sequence at time t is determined, and the position coordinates collected by other sensors in the i-th global sensor priority sequence are calibrated according to the deviation value.

[0181] In some embodiments of this application, processor 310 is further configured to:

[0182] According to the first priority rule, the highest priority sensor is determined from the multiple sensors in the target pairing list;

[0183] The other sensors in the target pairing list are prioritized according to the Euclidean distance between the first position and the second position to obtain the sensor priority sequence; wherein, the first position is the sensing position obtained by the highest priority sensor, and the second position is the sensing position obtained by any of the other sensors.

[0184] In some embodiments of this application, the first priority rule includes at least one of the following:

[0185] If the target object is detected by a single sensor at time t, then the single sensor is determined as the highest priority sensor;

[0186] If the target object is detected by multiple sensors at time t, the sensors are sorted according to their type to obtain the highest priority sensor.

[0187] If the target object is detected by multiple sensors at time t and the multiple sensors are of the same type, then the sensor with the highest confidence level of the detected target is determined as the highest priority sensor; wherein, the target confidence level is positively correlated with the priority level.

[0188] In some embodiments of this application, the processor 310 is specifically used for:

[0189] If the historical highest priority sensor already saved in the (i-1)th global sensor priority sequence is located in the target pairing list, the historical highest priority sensor is taken as the highest priority sensor, and the sensor priority in the first sensor set is determined according to the second priority rule to obtain the i-th global sensor priority; wherein, the first sensor set includes multiple sensors in the target pairing list and other sensors in the (i-1)th global sensor priority sequence other than the highest priority sensor;

[0190] If the highest-priority historical sensor saved in the (i-1)th global sensor priority sequence is not in the target pairing list, the highest-priority sensor is determined from the second sensor set, and the priorities of other sensors in the second sensor set are determined according to the second priority rule to obtain the i-th global sensor priority; wherein, the second sensor set includes multiple sensors in the target pairing list and other sensors in the (i-1)th global sensor priority sequence other than the highest-priority historical sensor.

[0191] In some embodiments of this application, the second priority rule includes:

[0192] The sensors in the first sensor set are sorted according to the Euclidean distance between the third and fourth positions to determine the sensor priority in the first sensor set; wherein, the third position is the sensing position obtained by the highest priority sensor, and the fourth position is the sensing position obtained by any sensor in the first sensor set.

[0193] In some embodiments of this application, the processor 310 is also used for at least one of the following:

[0194] The fusion speed of the target object at time t is determined based on the average value of the motion speeds sensed by multiple sensors in the target pairing list.

[0195] The fused heading angle of the target object at time t is determined based on the average heading angle sensed by multiple sensors in the target pairing list.

[0196] In some embodiments of this application, processor 310 is further configured to:

[0197] Based on the filtered position coordinates of the target object at time t-1, the fused position coordinates of the target object at time t, and the motion trend coordinates of the target object at time t-1, determine the filtered position coordinates of the target object at time t; or,

[0198] The filtered position coordinates of the target object at time t are determined based on the filtered position coordinates of the target object at time t-1 and the motion trend coordinates of the target object at time t-1.

[0199] In some embodiments of this application, processor 310 is further configured to:

[0200] The historical heading angle of the target object is determined based on the filtered position coordinates of the target object at time t and the position coordinates in the first frame of the historical trajectory of the target object.

[0201] Based on the historical heading angle and the fused heading angle of the target object at time t, the corrected heading angle of the target object at time t is determined.

[0202] In some embodiments of this application, processor 310 is further configured to:

[0203] The historical velocity of the target object is determined based on the filtered position coordinates of the target object at time t, the position coordinates in the first frame of the historical trajectory of the target object, the timestamp corresponding to the first frame of the historical trajectory of the target object, and time t.

[0204] Based on the historical velocity and the fusion velocity of the target object at time t, the correction velocity of the target object at time t is determined.

[0205] Another embodiment of the present invention provides a sensing target fusion device, such as... Figure 4 As shown, it includes a transceiver 410, a processor 400, a memory 420, and a program or instructions stored in the memory 420 and executable on the processor 400; when the processor 400 executes the program or instructions, it implements the above-described method for fusion of perceived targets.

[0206] The transceiver 410 is used to receive and send data under the control of the processor 400.

[0207] Among them, Figure 4In this context, the bus architecture can include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by processor 400 and memory represented by memory 420 together. The bus architecture can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. Transceiver 410 can be multiple elements, including transmitters and receivers, providing a unit for communicating with various other devices over a transmission medium. For different user equipment, user interface 430 can also be an interface capable of connecting external or internal devices, including but not limited to keypads, displays, speakers, microphones, joysticks, etc.

[0208] The processor 400 is responsible for managing the bus architecture and general processing, while the memory 420 can store the data used by the processor 400 when performing operations.

[0209] This invention provides a readable storage medium storing a program or instructions. When executed by a processor, the program or instructions implement the steps of the perception target fusion method described above and achieve the same technical effect. To avoid repetition, further details are omitted here. The computer-readable storage medium may include read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0210] It should be further noted that the terminals described in this specification include, but are not limited to, smartphones, tablets, etc., and many of the functional components described are referred to as modules in order to emphasize the independence of their implementation.

[0211] In this embodiment of the invention, the module can be implemented in software so that it can be executed by various types of processors. For example, an identified executable code module may include one or more physical or logical blocks of computer instructions, which may be constructed as objects, procedures, or functions. Nevertheless, the executable code of the identified module does not need to be physically located together, but may include different instructions stored in different bits, which, when logically combined, constitute the module and achieve the module's intended purpose.

[0212] In practice, an executable code module can be a single instruction or many instructions, and can even be distributed across multiple different code segments, different programs, and across multiple memory devices. Similarly, operational data can be identified within the module and can be implemented in any suitable form and organized within any suitable data structure. This operational data can be collected as a single dataset or distributed across different locations (including different storage devices), and can exist, at least in part, solely as electronic signals within the system or network.

[0213] When a module can be implemented using software, considering the current level of hardware technology, modules that can be implemented in software can be implemented using hardware circuits by those skilled in the art to achieve the corresponding functions, without considering cost. These hardware circuits include conventional very-large-scale integrated circuits (VLSI) or gate arrays, as well as existing semiconductors such as logic chips and transistors, or other discrete components. Modules can also be implemented using programmable hardware devices, such as field-programmable gate arrays, programmable array logic, and programmable logic devices.

[0214] The exemplary embodiments described above are with reference to the accompanying drawings. Many different forms and embodiments are feasible without departing from the spirit and teachings of the invention. Therefore, the invention should not be construed as limiting the exemplary embodiments set forth herein. Rather, these exemplary embodiments are provided to make the invention complete and convey the scope of the invention to those skilled in the art. In these drawings, component dimensions and relative dimensions may be exaggerated for clarity. The terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. As used herein, unless clearly indicated otherwise, the singular forms “a,” “an,” and “the” are intended to include all such forms. It will be further understood that the terms “comprising” and / or “including”, when used in this specification, indicate the presence of the stated features, integers, steps, operations, components, and / or elements, but do not exclude the presence or addition of one or more other features, integers, steps, operations, components, and / or groups thereof. Unless otherwise indicated, when stated, a range of values ​​includes the upper and lower limits of the range and any subranges in between.

[0215] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for fusion of perceived targets, characterized in that, include: Obtain a target matching list set at time t, wherein the target matching list set includes at least one target pairing list, and each target pairing list includes the pairing relationship between multiple sensors that perceive the same target object and the target ID; Based on the sensor priority sequence corresponding to multiple sensors in each target pairing list, update the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 to obtain the i-th global sensor priority sequence at time t, where i is a positive integer; wherein, the i-th global sensor priority sequence includes: the intersection of the target pairing list obtained at time t and the sensors included in the (i-1)th global sensor priority sequence, and the sensors added to the target pairing list at time t relative to the (i-1)th global sensor priority sequence; The sum of the position coordinates collected by the first sensor at time t and the correction value is determined as the fused position coordinates of the target object at time t; wherein, the first sensor is the highest priority sensor in the i-th global sensor priority sequence, the correction value is the deviation between the position coordinates collected by the first sensor at time t-1 and the position coordinates collected by the second sensor at time t-1, and the second sensor is the highest priority sensor in the (i-1)-th global sensor priority sequence; Determine the deviation between the fused position coordinates at time t and the position coordinates acquired by other sensors in the i-th global sensor priority sequence at time t, and calibrate the position coordinates acquired by other sensors in the i-th global sensor priority sequence based on the deviation.

2. The method according to claim 1, characterized in that, The methods for obtaining the sensor priority sequence corresponding to multiple sensors in the target pairing list include: According to the first priority rule, the highest priority sensor is determined from the multiple sensors in the target pairing list; The other sensors in the target pairing list are prioritized according to the Euclidean distance between the first position and the second position to obtain the sensor priority sequence; wherein, the first position is the sensing position obtained by the highest priority sensor, and the second position is the sensing position obtained by any of the other sensors.

3. The method according to claim 2, characterized in that, The first priority rule includes at least one of the following: If the target object is detected by a single sensor at time t, then the single sensor is determined as the highest priority sensor; If the target object is detected by multiple sensors at time t, the sensors are sorted according to their type to obtain the highest priority sensor. If the target object is detected by multiple sensors at time t and the multiple sensors are of the same type, then the sensor with the highest confidence level of the detected target is determined as the highest priority sensor; wherein, the target confidence level is positively correlated with the priority level.

4. The method according to claim 1, characterized in that, The step of updating the (i-1)th global sensor priority sequence of the target object at time t-1 according to the sensor priority sequence corresponding to multiple sensors in each target pairing list, to obtain the ith global sensor priority sequence at time t, includes: If the historical highest priority sensor already saved in the (i-1)th global sensor priority sequence is located in the target pairing list, the historical highest priority sensor is taken as the highest priority sensor, and the sensor priority in the first sensor set is determined according to the second priority rule to obtain the i-th global sensor priority sequence; wherein, the first sensor set includes multiple sensors in the target pairing list and other sensors in the (i-1)th global sensor priority sequence except for the highest priority sensor; If the highest-priority historical sensor already saved in the (i-1)th global sensor priority sequence is not in the target pairing list, the highest-priority sensor is determined from the second sensor set, and the priorities of other sensors in the second sensor set are determined according to the second priority rule to obtain the i-th global sensor priority sequence; wherein, the second sensor set includes multiple sensors in the target pairing list and other sensors in the (i-1)th global sensor priority sequence other than the highest-priority historical sensor.

5. The method according to claim 4, characterized in that, The second priority rule includes: The sensors in the first sensor set are sorted according to the Euclidean distance between the third and fourth positions to determine the sensor priority in the first sensor set; wherein, the third position is the sensing position obtained by the highest priority sensor, and the fourth position is the sensing position obtained by any sensor in the first sensor set.

6. The method according to claim 1, characterized in that, The method further includes at least one of the following: The fusion speed of the target object at time t is determined based on the average value of the motion speeds sensed by multiple sensors in the target pairing list. The fused heading angle of the target object at time t is determined based on the average heading angle sensed by multiple sensors in the target pairing list.

7. The method according to claim 6, characterized in that, The method further includes: Based on the filtered position coordinates of the target object at time t-1, the fused position coordinates of the target object at time t, and the motion trend coordinates of the target object at time t-1, determine the filtered position coordinates of the target object at time t; or, The filtered position coordinates of the target object at time t are determined based on the filtered position coordinates of the target object at time t-1 and the motion trend coordinates of the target object at time t-1.

8. The method according to claim 7, characterized in that, The method further includes: The historical heading angle of the target object is determined based on the filtered position coordinates of the target object at time t and the position coordinates in the first frame of the historical trajectory of the target object. Based on the historical heading angle and the fused heading angle of the target object at time t, the corrected heading angle of the target object at time t is determined.

9. The method according to claim 7, characterized in that, The method further includes: The historical velocity of the target object is determined based on the filtered position coordinates of the target object at time t, the position coordinates in the first frame of the historical trajectory of the target object, the timestamp corresponding to the first frame of the historical trajectory of the target object, and time t. Based on the historical velocity and the fusion velocity of the target object at time t, the correction velocity of the target object at time t is determined.

10. A target fusion sensing device, characterized in that, include: The acquisition module is used to acquire a target matching list set at time t. The target matching list set includes at least one target pairing list. Each target pairing list includes the pairing relationship between multiple sensors that perceive the same target object and the target ID. An update processing module is used to update the (i-1)th global sensor priority sequence corresponding to the target object at time t-1 based on the sensor priority sequence corresponding to multiple sensors in each target pairing list, to obtain the i-th global sensor priority sequence at time t, where i is a positive integer; wherein, the i-th global sensor priority sequence includes: the intersection of the target pairing list obtained at time t and the sensors included in the (i-1)th global sensor priority sequence, and the sensors added to the target pairing list at time t relative to the (i-1)th global sensor priority sequence; The fusion processing module is used to determine the fused position coordinates of the target object at time t by summing the position coordinates collected by the first sensor at time t and the correction value; wherein, the first sensor is the highest priority sensor in the i-th global sensor priority sequence, the correction value is the deviation between the position coordinates collected by the first sensor at time t-1 and the position coordinates collected by the second sensor at time t-1, and the second sensor is the highest priority sensor in the (i-1)-th global sensor priority sequence; The calibration module is used to determine the deviation between the fused position coordinates at time t and the position coordinates collected by other sensors in the i-th global sensor priority sequence at time t, and to calibrate the position coordinates collected by other sensors in the i-th global sensor priority sequence according to the deviation.

11. A sensing target fusion device, comprising: A transceiver, a processor, a memory, and a program or instructions stored in the memory and executable on the processor; characterized in that, when the processor executes the program or instructions, it implements the perceptual target fusion method as described in any one of claims 1-10.

12. A readable storage medium having a program or instructions stored thereon, characterized in that, When the program or instructions are executed by the processor, they implement the steps in the perceptual target fusion method as described in any one of claims 1-10.

Images