Multi-sensor post-fusion method and device based on single-sensor time-series tracking results

Abstract

The application discloses a multi-sensor post-fusion method and device based on single-sensor time sequence tracking results, and belongs to the field of automatic driving. The method comprises the following steps: acquiring target detection results corresponding to target objects at each moment by using a single sensor of an automatic driving vehicle; preprocessing the target detection results to obtain target detection data; tracking the target detection data at each moment by using a first preset graph neural network model to determine a same target object; inputting the target detection data of the same target object into a second preset graph neural network model to re-determine the same target object, and obtaining an optimized same target. The graph neural network is used to realize the identification of targets such as vehicles and pedestrians in the surrounding environment, the data corresponding to the same target is determined according to the tracking result of a single sensor, the perception ability of the single sensor for the target is improved, and the data of the same target object of multiple sensors is fused, so that the identification of targets such as pedestrians and vehicles is more accurate.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to a multi-sensor post-fusion method and apparatus based on single-sensor time-series tracking results. Background Technology

[0002] Currently, with the continuous increase in the number of traffic participants, road condition information is becoming increasingly complex. In complex and dynamic traffic environments, enabling autonomous vehicles to make proactive decision-making capabilities, such as deciding on driving behaviors like lane changes, acceleration for overtaking, and braking based on the traffic environment, plays a crucial role in establishing safe and efficient intelligent transportation systems. Within the traffic environment in which the vehicle operates, the perception and detection of surrounding vehicles and pedestrians is a significant factor influencing the behavioral decisions of autonomous vehicles. Therefore, in multi-vehicle interaction environments, the perception and detection of surrounding vehicles and pedestrians is of paramount importance for the development of advanced driver assistance systems, intelligent transportation systems, and autonomous driving.

[0003] In the process of sensing and detecting vehicles in the surrounding environment, data on vehicles and pedestrians in the surrounding environment captured by multiple sensors such as cameras, millimeter-wave radar, and lidar equipped on the vehicle are usually fused together, and relevant data about vehicles and pedestrians are output through the fused data.

[0004] In existing technologies, two methods are commonly used for fusion: method (1) uses hard correlation for fusion, and method (2) trains a model using machine learning and deep learning algorithms and outputs the fusion result using the trained model. However, when using method (1) for fusion, it only complements the data between each other and does not utilize the data from other sensors. The resulting fused data is too singular, which affects the implementation of the perception algorithm and therefore does not have a good effect on the accuracy of perception. In method (2), CNN is generally used. This method can only acquire information within a limited perception range and cannot achieve global information fusion. Moreover, it only works well for densely regulated data and does not have a good effect on other data. Summary of the Invention

[0005] To address the problem of poor fusion results caused by single fusion data and limited sensing information in existing technologies, this application mainly provides a multi-sensor post-fusion method and apparatus based on single-sensor time-series tracking results.

[0006] To achieve the above objectives, one technical solution adopted in this application is: providing a multi-sensor post-fusion method based on single-sensor time-series tracking results, which includes: using multiple single sensors of an autonomous vehicle to acquire target detection results corresponding to target objects at various times in each single sensor; preprocessing the multiple target detection results to obtain target detection data corresponding to the target detection results; using a first preset graph neural network model to perform time-series tracking of the target detection data at various times in each single sensor to determine the same target object in each single sensor; inputting the target detection data of the same target object determined by multiple single sensors at the same time into a second preset graph neural network model to re-determine the same target object, thereby obtaining optimized same target corresponding to multiple single sensors.

[0007] Another technical solution adopted in this application is: providing a multi-sensor post-fusion device based on single-sensor time-series tracking results, comprising: a module for acquiring target detection results corresponding to target objects at various times in each single sensor using multiple single sensors of an autonomous vehicle; a module for preprocessing multiple target detection results to obtain target detection data corresponding to the target detection results; a module for performing time-series tracking of target detection data at various times in each single sensor using a first preset graph neural network model to determine the same target object in each single sensor; and a module for inputting the target detection data of the same target object determined by multiple single sensors at the same time into a second preset graph neural network model to re-determine the same target object and obtain optimized targets corresponding to multiple single sensors.

[0008] Another technical solution adopted in this application is to provide a computer-readable storage medium storing computer instructions that are operated to execute the multi-sensor post-fusion method based on single-sensor time-series tracking results in Solution 1.

[0009] Another technical solution adopted in this application is: providing a computer device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores computer instructions executable by the at least one processor, and the at least one processor operates the computer instructions to execute the multi-sensor post-fusion method based on single-sensor time-series tracking results in Solution 1.

[0010] The beneficial effects of the technical solution in this application are as follows: This application designs a multi-sensor post-fusion method and device based on single-sensor time-series tracking results. This application implements the entire technical solution entirely through data-driven methods, making perception fusion more convenient; it utilizes graph neural networks to identify targets such as vehicles and pedestrians in the surrounding environment, avoiding the influence of noise and improving robustness; it determines the data corresponding to the same target object based on the tracking results of a single sensor, improving the single sensor's ability to perceive targets; and it fuses the data of the same target object determined by multiple single sensors to achieve global information fusion, resulting in more accurate identification of targets such as pedestrians and vehicles. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a schematic diagram of one implementation of a multi-sensor post-fusion method based on single-sensor time-series tracking results according to this application;

[0013] Figure 2 This is a schematic diagram of one embodiment of a multi-sensor post-fusion device based on single-sensor timing tracking results according to this application.

[0014] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0015] The preferred embodiments of this application will now be described in detail with reference to the accompanying drawings, so that the advantages and features of this application can be more easily understood by those skilled in the art, thereby providing a clearer and more definite definition of the scope of protection of this application.

[0016] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0017] Currently, with the continuous increase in the number of traffic participants, road condition information is becoming increasingly complex. In complex and dynamic traffic environments, enabling autonomous vehicles to make proactive decision-making capabilities, such as deciding on driving behaviors like lane changes, acceleration for overtaking, and braking deceleration based on the traffic environment, plays a crucial role in establishing safe and efficient intelligent transportation systems. Within the traffic environment in which the vehicle operates, the perception and detection of surrounding vehicles and pedestrians is a significant factor influencing the behavioral decisions of autonomous vehicles. Therefore, in multi-vehicle interaction environments, the perception and detection of surrounding vehicles and pedestrians is crucial for the development of advanced driver assistance systems, intelligent transportation systems, and autonomous driving. In the process of perceiving and detecting vehicles in the surrounding environment, data on vehicles and pedestrians captured by multiple sensors such as cameras, millimeter-wave radar, and lidar installed on the vehicle are typically fused together. The fused data is then used to output relevant data about vehicles and pedestrians.

[0018] In existing technologies, two methods are commonly used for fusion: method (1) uses hard correlation for fusion, and method (2) trains a model using machine learning and deep learning algorithms and outputs the fusion result using the trained model. However, when using method (1) for fusion, it only complements the data between each other and does not utilize the data from other sensors. The resulting fused data is too singular, which affects the implementation of the perception algorithm and therefore does not have a good effect on the accuracy of perception. In method (2), CNN is generally used. This method can only acquire information within a limited perception range and cannot achieve global information fusion. Moreover, it only works well for densely regulated data and does not have a good effect on other data.

[0019] The inventive concept of this application is to provide a multi-sensor post-fusion method and apparatus based on single-sensor time-series tracking results. The post-fusion approach utilizes a graph neural network (GNN) model to identify targets such as vehicles and pedestrians in the surrounding environment, improving robustness to noise, reducing code writing, and simplifying the perception and recognition steps. It determines the data corresponding to the same target object based on the tracking results of a single sensor, improving the single sensor's target perception capability, and fuses the data of the same target object determined by multiple sensors to achieve global information fusion, resulting in more accurate identification of targets such as pedestrians and vehicles. Furthermore, this application implements the entire technical solution entirely through data-driven methods, making perception fusion more convenient.

[0020] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0021] Figure 1 This paper illustrates one implementation of a multi-sensor post-fusion method based on single-sensor timing tracking results according to this application.

[0022] exist Figure 1 In the embodiment shown, the multi-sensor post-fusion method based on single-sensor time-series tracking results mainly includes step S101, which uses multiple single sensors of the autonomous vehicle to obtain the target detection results corresponding to the target object at each time step in each single sensor.

[0023] In this embodiment, a single sensor refers to a single sensor, which may include, but is not limited to, a camera, LiDAR, or millimeter-wave radar; multiple single sensors may include multiple sensors of different types and / or multiple sensors of the same type. Based on the identification of the raw data collected by the sensor, multiple target detection results corresponding to the target object in the data detected by the sensor are obtained, providing a basis for subsequent fusion tasks. Since the sensor has already detected the target object, interference from irrelevant data is eliminated, reducing the computational power requirements of subsequent fusion tasks and accelerating the running speed of the fusion task.

[0024] exist Figure 1 In the embodiment shown, the multi-sensor post-fusion method based on single-sensor time-series tracking results further includes step S102, which preprocesses multiple target detection results to obtain target detection data corresponding to the target detection results.

[0025] In this embodiment, multiple target detection results are preprocessed, and the relevant data corresponding to the target detection results are integrated and spliced ​​to obtain the target detection data corresponding to the target detection results, thus providing conditions for subsequent fusion.

[0026] In an optional embodiment of this application, if the multiple single sensors are cameras and / or lidar, the target detection data includes: the detection result of the target object, current time information, sensor type information of the output target detection result, and feature vector of the target detection result; if the multiple sensors are millimeter-wave radars, the target detection data includes: the detection result of the target object, current time information, and sensor type information of the output target detection result.

[0027] In this optional embodiment, when multiple single sensors are all cameras, all are lidar, or both cameras and lidar are present, the target detection results of the target objects corresponding to each single sensor, the feature vector corresponding to the target detection results, the time information of the sensor outputting the target detection results, and the sensor type information of the sensor are integrated and stitched together to obtain the target detection data corresponding to each target object at each time and with each sensor type information; when multiple single sensors are all millimeter-wave radars, the target detection results of the target objects corresponding to each single sensor, the time information of the sensor outputting the target detection results, and the sensor type information of the sensor are integrated and stitched together to obtain the target detection data corresponding to each target object at each time and with each sensor type information.

[0028] exist Figure 1 In the embodiment shown, the multi-sensor post-fusion method based on single-sensor time-series tracking results further includes step S103, which uses a first preset graph neural network model to perform time-series tracking of target detection data at each moment in each single sensor to determine the same target object in each single sensor.

[0029] In this embodiment, the target detection data of each target object in a single sensor at each time is input into a first preset graph neural network model. The same target object in the single sensor is determined from the target detection data at each time through the first preset graph neural network model. In this application, the target detection data of each target object in a single sensor at each time is input into the first preset graph neural network model. The same target object in the single sensor is determined by using target detection data at different time sequences, which facilitates and effectively achieves the perception purpose.

[0030] In one optional embodiment of this application, the first preset graph neural network model is used to perform time-series tracking of target detection data at each time moment in each single sensor to determine the same target object in each single sensor. This includes: inputting the target detection data at each time moment in each single sensor into the first preset graph neural network model to obtain target fusion data corresponding to the target object in each single sensor at each time moment; judging the score of the target fusion data, and determining the target object corresponding to the target fusion data with a score greater than a threshold as the same target object in each single sensor.

[0031] In this optional embodiment, the target detection data obtained by integrating data from a single sensor is input into a first preset graph neural network model to determine the same target object detected by the single sensor using the detection results of the target object at different time sequences. The target fusion data obtained by fusing the detection results of the target object at different time sequences may contain target fusion errors. Therefore, a score threshold is used to distinguish the target objects corresponding to the target fusion data of the single sensor that are greater than the score threshold, and the target objects are determined to be the same target object. The score threshold can be set to 99 points.

[0032] exist Figure 1 In the embodiment shown, the multi-sensor post-fusion method based on single-sensor time-series tracking results further includes step S104, which inputs the target detection data of the same target object determined by multiple single sensors at the same time into the second preset graph neural network model to re-determine the same target object, thereby obtaining the optimized same target corresponding to multiple single sensors.

[0033] In this embodiment, the fused data corresponding to the same target object at each time point is input into the second preset graph neural network model, and the detection data of each target corresponding to the same target are fused to obtain the optimized same target object, thereby improving the accuracy of target object perception.

[0034] In one optional embodiment of this application, target detection data of the same target object determined by multiple single sensors at the same time are input into a second preset graph neural network model to re-determine the same target object and obtain optimized same target corresponding to multiple single sensors. This includes: inputting target detection data of the same target object determined by multiple single sensors at the same time into the second preset graph neural network model to determine the sensor type information in the target detection data; merging target detection data with the same sensor type information to obtain first fused data corresponding to the target object; and determining optimized same target based on the first fused data.

[0035] In this optional embodiment, by determining the sensor type information in the target detection data, the target detection data corresponding to the same type of sensor are fused, that is, the target detection data corresponding to multiple sensors of the same type at the same time are merged, so that the relevant information of the target object corresponding to the target detection data is more complete, and the identification of the target object is more accurate.

[0036] For example, if the sensor type information is all camera, all LiDAR, and all millimeter-wave radar, then the target detection data of the same time for camera type will be merged, the target detection data of the same time for LiDAR type will be merged, and the target detection data of the same time for millimeter-wave radar type will be merged.

[0037] In one optional embodiment of this application, target detection data of the same target object determined by multiple sensors at the same time are input into a second preset graph neural network model to re-determine the same target object and obtain optimized same target corresponding to multiple sensors. The method further includes: inputting target detection data of the same target object determined by multiple sensors at the same time into the second preset graph neural network model to determine the sensor type information in the target detection data; converting target detection data with different sensor type information and merging the converted target detection data to obtain second fused data corresponding to the target object; and determining optimized same target based on the first fused data.

[0038] In this optional embodiment, by determining the sensor type information in the target detection data, the target detection data corresponding to different types of sensors are fused, that is, the target detection data corresponding to multiple different types of sensors at the same time are fused, so that the relevant information of the target object corresponding to the target detection data is more complete, and the identification of the target object is more accurate.

[0039] Figure 2 This paper illustrates an embodiment of a multi-sensor post-fusion device based on single-sensor timing tracking results according to this application.

[0040] exist Figure 2In the illustrated embodiment, the multi-sensor post-fusion device based on single-sensor time-series tracking results mainly includes: a module 201 for acquiring target detection results corresponding to target objects at various times in each single sensor using multiple single sensors of an autonomous vehicle; a module 201 for preprocessing multiple target detection results to obtain target detection data corresponding to the target detection results; a module 203 for performing time-series tracking of target detection data at various times in each single sensor using a first preset graph neural network model to determine the same target object in each single sensor; and a module 204 for inputting the target detection data of the same target object determined by multiple single sensors at the same time into a second preset graph neural network model to re-determine the same target object and obtain optimized targets corresponding to multiple single sensors.

[0041] In an optional embodiment of this application, if the multiple sensors are cameras and / or lidar, the target detection data includes: the detection result of the target object, current time information, sensor type information of the output target detection result, and feature vector of the target detection result; if the multiple sensors are millimeter-wave radar, the target detection data includes: the detection result of the target object, current time information, and sensor type information of the output target detection result.

[0042] In one optional embodiment of this application, the first preset graph neural network model is used to perform time-series tracking of target detection data at each time moment in each single sensor to determine the same target object in each single sensor. This includes: inputting the target detection data at each time moment in each single sensor into the first preset graph neural network model to obtain target fusion data corresponding to the target object in each single sensor at each time moment; judging the score of the target fusion data, and determining the target object corresponding to the target fusion data with a score greater than a threshold as the same target object in each single sensor.

[0043] In one optional embodiment of this application, target detection data of the same target object determined by multiple single sensors at the same time are input into a second preset graph neural network model to re-determine the same target object and obtain optimized same target corresponding to multiple single sensors. This includes: inputting target detection data of the same target object determined by multiple sensors at the same time into the second preset graph neural network model to determine the sensor type information in the target detection data; merging target detection data with the same sensor type information to obtain first fused data corresponding to the target object; and determining optimized same target based on the first fused data.

[0044] In one optional embodiment of this application, target detection data of the same target object determined by multiple single sensors at the same time are input into a second preset graph neural network model to re-determine the same target object, thereby obtaining optimized same target corresponding to multiple single sensors. The method further includes: inputting target detection data of the same target object determined by multiple sensors at the same time into the second preset graph neural network model to determine sensor type information in the target detection data; converting target detection data with different sensor type information and merging the converted target detection data to obtain second fused data corresponding to the target object; and determining optimized same target based on the first fused data.

[0045] The multi-sensor post-fusion device based on single-sensor time-series tracking results provided in this application can be used to execute the multi-sensor post-fusion method based on single-sensor time-series tracking results described in any of the above embodiments. Its implementation principle and technical effect are similar, and will not be repeated here.

[0046] In another embodiment of this application, a computer-readable storage medium stores computer instructions that are operated to perform the multi-sensor post-fusion method based on single-sensor timing tracking results described in the above embodiments.

[0047] In an optional embodiment of this application, each functional module in the multi-sensor post-fusion method based on single-sensor timing tracking results can be directly in hardware, in a software module executed by a processor, or in a combination of both.

[0048] Software modules may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in this art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium.

[0049] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof. A general-purpose processor can be a microprocessor, but alternatively, it can be any conventional processor, controller, microcontroller, or state machine. The processor can also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors incorporating a DSP core, or any other such configuration. Alternatively, the storage medium can be integrated with the processor. The processor and storage medium can reside in an ASIC. The ASIC can reside in the user terminal. Alternatively, the processor and storage medium can reside as discrete components in the user terminal.

[0050] In one embodiment of this application, a computer device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores computer instructions executable by the at least one processor, and the at least one processor operates the computer instructions to execute the multi-sensor post-fusion method based on single-sensor timing tracking results described in the above embodiments.

[0051] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0052] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0053] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A multi-sensor post-fusion method based on single-sensor time-series tracking results, characterized in that, include: By utilizing multiple individual sensors of an autonomous vehicle, the target detection results corresponding to the target object at each time point are obtained from each individual sensor. The target detection results of multiple targets are preprocessed to obtain target detection data corresponding to the target detection results; The first preset graph neural network model is used to perform time-series tracking of the target detection data at each time in each single sensor to determine the same target object in each single sensor; The target detection data of the same target object determined by the multiple single sensors at the same time are input into a second preset graph neural network model to re-determine the same target object, thereby obtaining an optimized same target corresponding to the multiple single sensors. The step of inputting the target detection data of the same target object determined by the multiple single sensors at the same time into the second preset graph neural network model to re-determine the same target object and obtain an optimized same target corresponding to the multiple single sensors further includes: The target detection data of the same target object determined by multiple single sensors at the same time are input into the second preset graph neural network model to determine the sensor type information in the target detection data; The target detection data with different sensor type information are converted, and the converted target detection data are merged to obtain the second fused data corresponding to the target object; The optimization target is determined based on the second fused data.

2. The multi-sensor post-fusion method based on single-sensor time-series tracking results according to claim 1, characterized in that, If the plurality of single sensors are cameras and / or lidar, then the target detection data includes: The detection result of the target object, the current time information, the sensor type information that outputs the target detection result, and the feature vector of the target detection result; If the plurality of single sensors are millimeter-wave radars, then the target detection data includes: the detection result of the target object, the current time information, and the sensor type information that outputs the target detection result.

3. The multi-sensor post-fusion method based on single-sensor time-series tracking results according to claim 1, characterized in that, The step of using a first preset graph neural network model to perform time-series tracking of the target detection data at each time step in each individual sensor to determine the same target object in each individual sensor includes: The target detection data at each time step in each individual sensor is input into the first preset graph neural network model to obtain the target fusion data corresponding to the target object in each individual sensor. The score of the target fusion data is determined, and the target object corresponding to the target fusion data with a score greater than a threshold is identified as the same target object in each single sensor.

4. The multi-sensor post-fusion method based on single-sensor time-series tracking results according to claim 1, characterized in that, The step of inputting the target detection data of the same target object determined by the multiple single sensors at the same time into the second preset graph neural network model to re-determine the same target object and obtain the optimized same target corresponding to the multiple single sensors includes: The target detection data of the same target object determined by the multiple single sensors at the same time are input into the second preset graph neural network model to determine the sensor type information in the target detection data. The target detection data with the same sensor type information are merged to obtain the first fused data corresponding to the target object; The optimization target is determined based on the first fused data.

5. A multi-sensor post-fusion device based on single-sensor time-series tracking results, characterized in that, include: A module for using multiple single sensors of an autonomous vehicle to obtain the target detection results of the target object at each time point in each single sensor; A module for preprocessing multiple target detection results to obtain target detection data corresponding to the target detection results; A module for using a first preset graph neural network model to perform time-series tracking of the target detection data at each time moment in each single sensor, and to determine the same target object in each single sensor; The module for inputting target detection data of the same target object determined by the multiple single sensors at the same time into a second preset graph neural network model to re-determine the same target object and obtain the optimized same target corresponding to the multiple single sensors, wherein the module for inputting target detection data of the same target object determined by the multiple single sensors at the same time into a second preset graph neural network model to re-determine the same target object and obtain the optimized same target corresponding to the multiple single sensors further includes: A module for inputting the target detection data of the same target object determined by the multiple single sensors at the same time into the second preset graph neural network model, and determining the sensor type information in the target detection data; A module for converting target detection data with different sensor type information, and merging the converted target detection data to obtain second fused data corresponding to the target object; A module used to determine the same optimization objective based on the second fused data.

6. A computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are operated to perform the multi-sensor post-fusion method based on single-sensor timing tracking results as described in any one of claims 1-4.

7. A computer device, characterized in that, include: At least one processor; as well as A memory that is communicatively connected to the at least one processor; The memory stores computer instructions executable by the at least one processor, which operates the computer instructions to perform the multi-sensor post-fusion method based on single-sensor timing tracking results as described in any one of claims 1-4.

Images