Information fusion method and system based on Internet of Things technology

Abstract

The invention discloses an information fusion method and system based on the Internet of Things technology, and relates to the technical field of information fusion, and the method comprises the following steps: S1, preprocessing sensor measurement information; s2, multi-channel sensing information fusion is carried out; s3, acquiring a sensor measurement value under the historical track of the vehicle; s4, associating the sensor with the track of the vehicle; and S5, all-around perception of the vehicle is carried out. According to the method, the radar sensor is associated with the vehicle track, the tracking precision can be ensured while the automatic sensing requirement of the vehicle is met, the problem that the radar sensor is difficult to estimate a target in all directions in a complex environment is solved, all-direction 360-degree target tracking is carried out through the target-level fusion center, and the method is efficient, intelligent and high in practicability. And the track management module is used for managing and maintaining the moving tracks of all vehicles, so that all-directional and full-period target tracking is realized, the stability of the tracks is kept, the all-directional sensing requirement of vehicle automation function upgrading is met, and the development of social productivity is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of information fusion, and in particular to an information fusion method and system based on Internet of Things technology. Background Art

[0002] The Internet of Things refers to connecting any object to the network through information sensing devices according to the agreed protocol. Objects exchange and communicate information through information dissemination media to achieve intelligent identification, positioning, tracking, supervision and other functions. For autonomous driving technology, accurate perception of obstacles around unmanned vehicles can provide a basis for making correct decisions for autonomous driving control and ensure driving safety. In the Chinese patent application number 202110857243.7, "a radar sensor fusion method, device, computer equipment and storage medium are disclosed. On the one hand, two radar sensors are used for detection, which can reduce false detection and increase the possibility of obstacle detection compared with a single radar sensor. On the other hand, when the detection results of the two radar sensors are fused, since each obstacle detected by the two radar sensors is matched with representative points respectively, the target fused obstacle is determined, that is, multiple point clouds representing obstacles with one representative point are reduced. The number of point clouds to be matched is reduced, and the number of matching calculations to be reduced. The detection results of the two sensors corresponding to the target fused obstacle are fused."

[0003] The above documents can improve the fusion efficiency of sensor information. However, with the development of vehicle automation technology and the popularization of vehicle networking technology, the estimation of moving targets that integrate multi-source information also faces new challenges. On the one hand, the perception system needs to integrate more sensors to adapt to the all-round perception needs of vehicle automation function upgrades, but the rapid iterative update of vehicle automation requires the perception system to be compatible with the problem of multi-level target information mixing brought by sensors. On the other hand, vehicle networking technology provides intelligent connected vehicles with richer perception information to improve the inherent limitations of sensors such as perception distance and occlusion. However, the complex information sources also make the correlation between information very complicated, and there are also problems with the reference benchmark of information and time asynchronous processing, making it difficult to improve the utilization of information. Summary of the invention

[0004] The purpose of the present invention is to provide an information fusion method and system based on Internet of Things technology to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: an information fusion method based on Internet of Things technology, comprising the following steps:

[0006] S1. Preprocessing sensor measurement information: By combining N millimeter-wave radar sensors and the vehicle's own motion information, 360-degree target measurement is achieved. First, the radar sensor's measurement value is filtered by combining lane detection and the vehicle's own motion information, and each radar sensor is time-space aligned. A synchronization gateway is used to unify the clocks of each radar sensor to obtain measurement information under a unified time-space reference.

[0007] S2. Multi-channel sensor information fusion: A tree topology structure is used to reasonably allocate the routing node channels of the radar sensor to reduce the interference of parallel channels in the multi-channel sensor information transmission. Then, the fusion set of each information unit of the radar sensor is calculated, and the local distance entropy is obtained by a local algorithm, and the global distance entropy is obtained by a global algorithm. The weight value of information fusion is obtained through the local distance entropy and the global distance entropy, and finally the multi-channel fusion of the radar sensor network is completed;

[0008] S3, obtaining sensor measurement values ​​under the vehicle's historical track: using a conversion algorithm to convert the prior estimation value of the historical track in the vehicle's own coordinate system, i.e., the track state, into a measurement prediction value in the millimeter-wave radar sensor coordinate system, so as to realize the association between the sensor measurement and the vehicle track data;

[0009] S4, sensor and vehicle track association: Based on the radar sensor measurement and vehicle track data association results, a 360-degree target tracking is performed through the target-level fusion center. The target-level fusion center includes a centralized estimation method and a measurement fusion method. The centralized estimation method is used to update the status of the global historical track, and the measurement fusion method is used to initialize the unassociated measured historical tracks as new tracks for association, and then all tracks are fed back to the track management module;

[0010] S5. All-round perception of vehicles: The active tracks of all vehicles are managed and maintained through the track management module. All active tracks include updated historical tracks and newly generated tracks, so as to achieve all-round and full-cycle target tracking and maintain track stability, adapt to the all-round perception needs of vehicle automation function upgrades, and improve road safety and traffic efficiency.

[0011] Preferably, in step S2, the local algorithm is specifically:

[0012]

[0013] Among them, a i Indicates the information source, g' c represents the fusion set of information units, I represents the total number of information units, i represents the sequence number of the information unit, and g icrepresents the optimal sensor information unit function expression, J1 represents a large attribute set, J2 represents a small attribute set, j represents an attribute, max represents a maximum value function, min represents a minimum value function, g i represents the sensor information fusion object, represents the information fusion distance, represents the local sensing distance entropy.

[0014] Preferably, in step S2, the global distance entropy is calculated by a global algorithm, and the global weight value of each sensing channel information is obtained by the global distance entropy. The global algorithm includes a local algorithm, wherein the optimal sensing information unit function expression in the local algorithm is:

[0015]

[0016] Among them, g' c represents the fusion set of information units, g ic It represents the optimal sensing information unit function expression, I represents the total number of information units, i represents the sequence number of the information unit, and max represents the maximum value function.

[0017] Preferably, in step S3, the conversion algorithm is specifically:

[0018]

[0019] X k =[x k ,y k ] T

[0020]

[0021] Among them, Z k represents the output measurement vector of the radar sensor, ρ k , and They represent the distance, angle and radial velocity of the target object detected by the radar sensor, respectively. k represents the state vector, i.e., the prior estimate of the historical track in the vehicle coordinate system, x k and k They represent the longitudinal and lateral distances of the object at time k, respectively. represents the measured predicted value in the radar sensor coordinate system, α i Indicates the installation angle of the radar sensor, [·] T represents vector transpose, and i represents the radar sensor number.

[0022] Preferably, in step S4, the result of associating the radar sensor measurements with the vehicle track data will generate three types of tracks, namely, tracks associated with N radar sensor measurements, tracks associated with a single radar sensor measurement, and tracks with unassociated measurements.

[0023] Preferably, in step S4, the centralized estimation method is:

[0024] The target is tracked by fusing the measurement prediction values ​​of N radar sensors, and the target detected by the radar sensor is used as the input information of the measurement fusion method. The estimator is used to realize the centralized estimation in the coordinate system of N radar sensors to improve the scalability of the fusion.

[0025] The measurement fusion method is:

[0026] The existence of unassociated measured historical tracks indicates that a new target has entered the detection range of the radar sensor. The unassociated measured historical tracks are initialized as new tracks. The state vector of the new track is calculated based on the angle and distance of the measured values. The speed and acceleration are initialized to 0. A unique identifier is assigned to each track. The new tracks are then associated and fused based on the new targets detected by the radar sensor. A single radar sensor or multiple radar sensors with overlapping detection areas are used for association to achieve fusion of the new tracks.

[0027] Preferably, in step S5, the track management module includes a track generation process and a track elimination process;

[0028] Track generation process: A new track with M2 successful association cycles in M1 consecutive cycles is considered a stable track, M1 ≥ M2, and enters the track extinction process in the next cycle;

[0029] Track extinction process: The initial value of the track extinction flag variable is M d , each time there is an unsuccessful association, the flag variable value decreases by 1, each time the association is successful, and the track disappearance flag variable value is less than the initial value M d , the flag variable value increases by 1. If the flag variable value decreases to 0, the track is determined to be extinct.

[0030] The information fusion system based on Internet of Things technology includes:

[0031] A measurement information processing unit, which realizes 360-degree target measurement by combining N millimeter-wave radar sensors and the vehicle's own motion information. The measurement values ​​of the radar sensors are first filtered by combining lane detection and the vehicle's own motion information, and each radar sensor is time-space aligned. A synchronization gateway is used to unify the clocks of each radar sensor to obtain measurement information under a unified time-space reference;

[0032] A sensor information fusion unit, which is responsible for reducing the interference of parallel channels in multi-channel sensor information transmission and completing multi-channel fusion of the radar sensor network;

[0033] A measurement value acquisition unit, wherein the measurement value acquisition unit converts a priori estimated value of the historical track in the own vehicle coordinate system, that is, the track state, into a measurement prediction value in the millimeter wave radar sensor coordinate system by using a conversion algorithm, so as to realize the association between the sensor measurement and the vehicle track data;

[0034] An information association unit, wherein the information association unit performs 360-degree target tracking through a target-level fusion center based on the association result of radar sensor measurement and vehicle track data. The target-level fusion center includes a centralized estimation method and a measurement fusion method. The centralized estimation method is used to update the status of global historical tracks, and the measurement fusion method is used to initialize the historical tracks that have not been associated with the measurement as new tracks for association, and then all tracks are fed back to the track management module;

[0035] A comprehensive perception unit manages and maintains the active tracks of all vehicles through a track management module, thereby achieving all-round and full-cycle target tracking and maintaining track stability.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. The present invention realizes efficient target measurement by preprocessing sensor measurement information, can provide accurate environmental perception through multi-channel sensor information fusion, improves real-time performance and information utilization, uses a tree topology structure to allocate routing node channels of radar sensors, suppresses parallel channel interference, minimizes information transmission energy consumption, and obtains weight values ​​of local and global information fusion to complete multi-channel fusion, effectively reduces information fusion delay, realizes compatible expansion and efficient fusion of millimeter-wave radar sensors, improves the estimation accuracy of dynamic targets, and realizes the association of sensor measurement with vehicle track data by obtaining sensor measurement values ​​under the historical track of the vehicle;

[0038] 2. The present invention associates radar sensors with vehicle tracks, which can ensure tracking accuracy while meeting vehicle automation perception requirements, solving the problem that radar sensors are difficult to estimate targets in all directions in complex environments. All-round 360-degree target tracking is performed through a target-level fusion center, which is efficient and intelligent. An estimator is used to implement centralized estimation in the coordinate systems of N radar sensors to improve the scalability of fusion. The active tracks of all vehicles are managed and maintained through a track management module, thereby achieving all-round and full-cycle target tracking and maintaining track stability, adapting to the all-round perception requirements of vehicle automation function upgrades, improving road safety and traffic efficiency, and promoting the development of social productivity. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic diagram of the overall method flow provided by an embodiment of the present invention;

[0040] Figure 2 A schematic diagram of the overall system flow provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] See also Figure 1 to Figure 2 The present invention provides a technical solution: an information fusion method based on Internet of Things technology, comprising the following steps:

[0043] S1. Preprocessing sensor measurement information: By combining N millimeter-wave radar sensors and the vehicle's own motion information, 360-degree target measurement is achieved. First, the radar sensor's measurement value is filtered by combining lane detection and the vehicle's own motion information, and each radar sensor is time-space aligned. A synchronization gateway is used to unify the clocks of each radar sensor to obtain measurement information under a unified time-space reference.

[0044] S2. Multi-channel sensor information fusion: A tree topology structure is used to reasonably allocate the routing node channels of the radar sensor to reduce the interference of parallel channels in the multi-channel sensor information transmission. Then, the fusion set of each information unit of the radar sensor is calculated, and the local distance entropy is obtained by a local algorithm, and the global distance entropy is obtained by a global algorithm. The weight value of information fusion is obtained through the local distance entropy and the global distance entropy, and finally the multi-channel fusion of the radar sensor network is completed;

[0045] S3, obtaining sensor measurement values ​​under the vehicle's historical track: using a conversion algorithm to convert the prior estimation value of the historical track in the vehicle's own coordinate system, i.e., the track state, into a measurement prediction value in the millimeter-wave radar sensor coordinate system, so as to realize the association between the sensor measurement and the vehicle track data;

[0046] S4, sensor and vehicle track association: Based on the radar sensor measurement and vehicle track data association results, a 360-degree target tracking is performed through the target-level fusion center. The target-level fusion center includes a centralized estimation method and a measurement fusion method. The centralized estimation method is used to update the status of the global historical track, and the measurement fusion method is used to initialize the unassociated measured historical tracks as new tracks for association, and then all tracks are fed back to the track management module;

[0047] S5. All-round perception of vehicles: The active tracks of all vehicles are managed and maintained through the track management module. All active tracks include updated historical tracks and newly generated tracks, so as to achieve all-round and full-cycle target tracking and maintain track stability, adapt to the all-round perception needs of vehicle automation function upgrades, and improve road safety and traffic efficiency.

[0048] In step S2, the local algorithm is specifically:

[0049]

[0050] Among them, a i Indicates the information source, g' c represents the fusion set of information units, I represents the total number of information units, i represents the sequence number of the information unit, and g ic represents the optimal sensor information unit function expression, J1 represents a large attribute set, J2 represents a small attribute set, j represents an attribute, max represents a maximum value function, min represents a minimum value function, g i represents the sensor information fusion object, represents the information fusion distance, represents the local sensing distance entropy;

[0051] In step S2, the global distance entropy is calculated by a global algorithm, and the global weight value of each sensing channel information is obtained by the global distance entropy. The global algorithm includes a local algorithm, wherein the optimal sensing information unit function expression in the local algorithm is:

[0052]

[0053] Among them, g' c represents the fusion set of information units, g icrepresents the optimal sensing information unit function expression, I represents the total number of information units, i represents the information unit sequence number, and max represents the maximum value function;

[0054] The local weight value of each sensor channel information is obtained by local distance entropy and global distance entropy respectively, and the calculation formula is:

[0055]

[0056] Among them, ω c represents the weight value, k represents the total number of information units, c represents the sequence number of the information unit, represents local distance entropy or global distance entropy;

[0057] The local weight value is integrated with the global weight value using a linear weighted algorithm. The linear weighted algorithm is as follows:

[0058]

[0059] Among them, Z represents the weight value of information fusion, ω1 represents the local weight value, ω2 represents the global weight value, and z i represents the sensor information fusion object sequence, i represents the sensor information fusion object sequence number, i∈[1,q], q represents the total number of sensor information fusion objects;

[0060] In step S3, the conversion algorithm is specifically:

[0061]

[0062] X k =[x k ,y k ] T

[0063]

[0064] Among them, Z k represents the output measurement vector of the radar sensor, ρ k , and They represent the distance, angle and radial velocity of the target object detected by the radar sensor, respectively. k represents the state vector, i.e., the prior estimate of the historical track in the vehicle coordinate system, x k and k They represent the longitudinal and lateral distances of the object at time k, respectively. represents the measured predicted value in the radar sensor coordinate system, α i Indicates the installation angle of the radar sensor, [·] T represents vector transpose, i represents the radar sensor number;

[0065] In step S4, the result of associating the radar sensor measurements with the vehicle track data will generate three types of tracks, namely, tracks associated with N radar sensor measurements, tracks associated with a single radar sensor measurement, and tracks without associated measurements;

[0066] In step S4, the centralized estimation method is:

[0067] The target is tracked by fusing the measurement prediction values ​​of N radar sensors, and the target detected by the radar sensor is used as the input information of the measurement fusion method. The estimator is used to realize the centralized estimation in the coordinate system of N radar sensors to improve the scalability of the fusion.

[0068] The form of the estimator is:

[0069] Where P represents the representation of the estimator, represents the state vector estimation in the radar sensor coordinate system, (·) T represents the transposed vector, J i The Jacobian matrix representing the nonlinear measurement, R i represents the target matrix detected by N radar sensors, where N represents the number of radar sensors and i represents the radar sensor serial number;

[0070] The measurement fusion method is:

[0071] The existence of uncorrelated measured historical tracks indicates that a new target has entered the detection range of the radar sensor. The uncorrelated measured historical tracks are initialized as new tracks. The state vector of the new track is calculated based on the angle and distance of the measured values. The speed and acceleration are initialized to 0. A unique identifier is assigned to each track. The new track is then associated and fused based on the new target detected by the radar sensor. A single radar sensor or multiple radar sensors with overlapping detection areas are used for association, thereby realizing the fusion of the new track.

[0072] In step S5, the track management module includes a track generation process and a track elimination process;

[0073] Track generation process: A new track with M2 successful association cycles in M1 consecutive cycles is considered a stable track, M1 ≥ M2, and enters the track extinction process in the next cycle;

[0074] Track extinction process: The initial value of the track extinction flag variable is M d , each time there is an unsuccessful association, the flag variable value decreases by 1, each time the association is successful, and the track disappearance flag variable value is less than the initial value M d , the flag variable value increases by 1. If the flag variable value decreases to 0, the track is determined to be extinct;

[0075] The information fusion system based on Internet of Things technology includes:

[0076] The measurement information processing unit combines N millimeter-wave radar sensors and the vehicle's own motion information to achieve 360-degree target measurement. First, the radar sensor's measurement value is filtered by combining lane detection and the vehicle's own motion information, and each radar sensor is time-space aligned. A synchronization gateway is used to unify the clocks of each radar sensor to obtain measurement information under a unified time-space reference.

[0077] The sensor information fusion unit is responsible for reducing the interference of parallel channels in multi-channel sensor information transmission and completing the multi-channel fusion of the radar sensor network;

[0078] A measurement value acquisition unit, which uses a conversion algorithm to convert a priori estimated value of the historical track in the own vehicle coordinate system, that is, the track state, into a measurement prediction value in the millimeter wave radar sensor coordinate system, so as to realize the association between the sensor measurement and the vehicle track data;

[0079] Information association unit: The information association unit performs 360-degree target tracking through a target-level fusion center based on the association results of radar sensor measurements and vehicle track data. The target-level fusion center includes a centralized estimation method and a measurement fusion method. The centralized estimation method is used to update the status of global historical tracks, and the measurement fusion method is used to initialize historical tracks that have not been associated with measurements as new tracks for association. Then all tracks are fed back to the track management module.

[0080] Comprehensive perception unit: The comprehensive perception unit manages and maintains the active tracks of all vehicles through the track management module, thereby achieving all-round and full-cycle target tracking and maintaining track stability.

[0081] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0082] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The information fusion method based on Internet of Things technology is characterized by: The method comprises the following steps: S1. Preprocessing sensor measurement information: By combining N millimeter-wave radar sensors and the vehicle's own motion information, 360-degree target measurement is achieved. First, the radar sensor's measurement value is filtered by combining lane detection and the vehicle's own motion information, and each radar sensor is time-space aligned. A synchronization gateway is used to unify the clocks of each radar sensor to obtain measurement information under a unified time-space reference. S2. Multi-channel sensor information fusion: A tree topology structure is used to reasonably allocate the routing node channels of the radar sensor to reduce the interference of parallel channels in the multi-channel sensor information transmission. Then, the fusion set of each information unit of the radar sensor is calculated, and the local distance entropy is obtained by a local algorithm, and the global distance entropy is obtained by a global algorithm. The weight value of information fusion is obtained through the local distance entropy and the global distance entropy, and finally the multi-channel fusion of the radar sensor network is completed; S3, obtaining sensor measurement values ​​under the vehicle's historical track: using a conversion algorithm to convert the prior estimation value of the historical track in the vehicle's own coordinate system, i.e., the track state, into a measurement prediction value in the millimeter-wave radar sensor coordinate system, so as to realize the association between the sensor measurement and the vehicle track data; S4, sensor and vehicle track association: Based on the radar sensor measurement and vehicle track data association results, a 360-degree target tracking is performed through the target-level fusion center. The target-level fusion center includes a centralized estimation method and a measurement fusion method. The centralized estimation method is used to update the status of the global historical track, and the measurement fusion method is used to initialize the unassociated measured historical tracks as new tracks for association, and then all tracks are fed back to the track management module; S5. All-round perception of vehicles: The active tracks of all vehicles are managed and maintained through the track management module. All active tracks include updated historical tracks and newly generated tracks, so as to achieve all-round and full-cycle target tracking and maintain track stability, adapt to the all-round perception needs of vehicle automation function upgrades, and improve road safety and traffic efficiency.

2. The information fusion method based on Internet of Things technology according to claim 1 is characterized in that: In step S2, the local algorithm is specifically: Among them, a i represents the information source, g′ c represents the fusion set of information units, I represents the total number of information units, i represents the sequence number of the information unit, and g ic represents the optimal sensor information unit function expression, J1 represents a large attribute set, J2 represents a small attribute set, j represents an attribute, max represents a maximum value function, min represents a minimum value function, g i represents the sensor information fusion object, represents the information fusion distance, represents the local sensing distance entropy.

3. The information fusion method based on Internet of Things technology according to claim 2 is characterized in that: In step S2, the global distance entropy is calculated by a global algorithm, and the global weight value of each sensing channel information is obtained by the global distance entropy. The global algorithm includes a local algorithm, wherein the optimal sensing information unit function expression in the local algorithm is: Among them, g′ c represents the fusion set of information units, g ic It represents the optimal sensing information unit function expression, I represents the total number of information units, i represents the sequence number of the information unit, and max represents the maximum value function.

4. The information fusion method based on Internet of Things technology according to claim 1 is characterized in that: In step S3, the conversion algorithm is specifically: X k =[x k ,y k ] T Among them, Z k represents the output measurement vector of the radar sensor, ρ k , and They represent the distance, angle and radial velocity of the target object detected by the radar sensor, respectively. k represents the state vector, i.e., the prior estimate of the historical track in the vehicle coordinate system, x k and k They represent the longitudinal and lateral distances of the object at time k, respectively. represents the measured predicted value in the radar sensor coordinate system, α i Indicates the installation angle of the radar sensor, [·] T represents vector transpose, and i represents the radar sensor number.

5. The information fusion method based on Internet of Things technology according to claim 1 is characterized in that: In step S4, the association result of the radar sensor measurements with the vehicle track data will generate three types of tracks, namely, tracks associated with N radar sensor measurements, tracks associated with a single radar sensor measurement, and tracks without associated measurements.

6. The information fusion method based on Internet of Things technology according to claim 1 is characterized in that: In step S4, the centralized estimation method is: The target is tracked by fusing the measurement prediction values ​​of N radar sensors, and the target detected by the radar sensor is used as the input information of the measurement fusion method. The estimator is used to realize the centralized estimation in the coordinate system of N radar sensors to improve the scalability of the fusion. The measurement fusion method is: The existence of unassociated measured historical tracks indicates that a new target has entered the detection range of the radar sensor. The unassociated measured historical tracks are initialized as new tracks. The state vector of the new track is calculated based on the angle and distance of the measured values ​​of the radar sensor. The speed and acceleration are initialized to 0, and a unique identifier is assigned to each track. The new tracks are then associated and fused based on the new targets detected by the radar sensor. A single radar sensor or multiple radar sensors with overlapping detection areas are used for association, thereby realizing the fusion of the new tracks.

7. The information fusion method based on Internet of Things technology according to claim 1 is characterized in that: In step S5, the track management module includes a track generation process and a track elimination process; Track generation process: A new track with M2 successful association cycles in M1 consecutive cycles is considered a stable track, M1 ≥ M2, and enters the track extinction process in the next cycle; Track extinction process: The initial value of the track extinction flag variable is M d , each time there is an unsuccessful association, the flag variable value decreases by 1, each time the association is successful, and the track disappearance flag variable value is less than the initial value M d , the flag variable value increases by 1. If the flag variable value decreases to 0, the track is determined to be extinct.

8. The information fusion system based on Internet of Things technology is characterized by: The information fusion system is applicable to the information fusion method based on Internet of Things technology described in claims 1-7, including: A measurement information processing unit, which realizes 360-degree target measurement by combining N millimeter-wave radar sensors and the vehicle's own motion information. The measurement values ​​of the radar sensors are first filtered by combining lane detection and the vehicle's own motion information, and each radar sensor is time-space aligned. A synchronization gateway is used to unify the clocks of each radar sensor to obtain measurement information under a unified time-space reference; A sensor information fusion unit, which is responsible for reducing the interference of parallel channels in multi-channel sensor information transmission and completing multi-channel fusion of the radar sensor network; A measurement value acquisition unit, wherein the measurement value acquisition unit converts a priori estimated value of the historical track in the own vehicle coordinate system, that is, the track state, into a measurement prediction value in the millimeter wave radar sensor coordinate system by using a conversion algorithm, so as to realize the association between the sensor measurement and the vehicle track data; An information association unit, wherein the information association unit performs 360-degree target tracking through a target-level fusion center based on the association result of radar sensor measurement and vehicle track data. The target-level fusion center includes a centralized estimation method and a measurement fusion method. The centralized estimation method is used to update the status of global historical tracks, and the measurement fusion method is used to initialize the historical tracks that have not been associated with the measurement as new tracks for association, and then all tracks are fed back to the track management module; A comprehensive perception unit manages and maintains the active tracks of all vehicles through a track management module, thereby achieving all-round and full-cycle target tracking and maintaining track stability.

Images