Obstacle target detection method for unmanned underwater vehicle collision avoidance system

Abstract

The invention belongs to the technical field of underwater target detection, and particularly relates to an obstacle target detection method for an unmanned underwater vehicle collision avoidance system, which comprises the following steps of: 1) receiving beam data acquired by a collision avoidance sonar of an unmanned underwater vehicle; (2) background cancellation processing is carried out on adjacent N frames of beam data by adopting an inter-frame difference method; step 3) adopting a clustering method for the data after background offset processing to obtain scale information and position information of the target to be detected; and 4) data association is carried out, tracking filtering processing is carried out according to a data association result, and a target detection result is output. According to the method, the robustness of a detection algorithm in a complex environment is further enhanced, the obstacle target detection efficiency of the collision avoidance system is improved, and the early warning capability of the collision avoidance system and the target existence confidence coefficient are guaranteed.

Description

Technical Field

[0001] The present invention belongs to the technical field of underwater target detection, and in particular relates to an obstacle target detection method for an unmanned underwater vehicle collision avoidance system. Background Art

[0002] With the increasing use of underwater vehicles, safe navigation has become an important issue. Underwater vehicles, especially unmanned submersibles and automated underwater vehicles (AUVs), often face unknown and complex underwater environments when performing tasks such as ocean exploration, scientific research, and environmental monitoring, and there is a risk of collision with other underwater obstacles. In order to ensure the safety of underwater vehicles when performing tasks, the application of obstacle avoidance systems (OAS) is particularly important.

[0003] At present, researchers mainly detect obstacles in sonar images from the perspective of sonar images. However, mapping sonar data to sonar images will lead to loss of target information. When the target signal is masked by reverberation, image processing-based methods will usually find it difficult to detect obstacles. When using a fixed detection threshold to detect obstacles, it is easily affected by the background environment. If the detection threshold is too high, obstacles will be missed, and if the detection threshold is too low, there will be too many false alarms.

[0004] In the actual application of unmanned underwater vehicles, due to the complex and changeable underwater environment, the traditional collision avoidance detection algorithm processes the sonar image through image processing methods. However, in a strong reverberation environment, the target signal is masked by the reverberation environment, and the traditional method is difficult to detect the target signal; the detection method using a fixed detection threshold is difficult to achieve both high accuracy and low false alarm rate.

[0005] Therefore, in order to adapt to the complex underwater environment and complete the obstacle target detection task in the unmanned vehicle collision avoidance system, there is an urgent need for an obstacle target detection algorithm that has both background environment suppression and high accuracy, further enhancing the robustness of the detection algorithm in complex environments, improving the efficiency of the collision avoidance system in detecting obstacle targets, and ensuring the early warning capability of the collision avoidance system and the confidence in the existence of the target. Summary of the invention

[0006] The purpose of the present invention is to overcome the defects of the prior art and propose an obstacle target detection method for an unmanned underwater vehicle collision avoidance system.

[0007] In view of this, the present invention proposes an obstacle target detection method for an unmanned underwater vehicle collision avoidance system, comprising:

[0008] Step 1) receiving beam data collected by the collision avoidance sonar of the unmanned underwater vehicle;

[0009] Step 2) using the inter-frame difference method to perform background cancellation processing on the adjacent N frames of beam data;

[0010] Step 3) using a clustering method on the background-offset processed data to obtain scale information and position information of the target to be detected;

[0011] Step 4) Perform data association, perform tracking filtering processing based on the results of data association and output the target detection results.

[0012] Preferably, the N frame value of step 2) is determined according to the relative movement speed of the unmanned aerial vehicle and the obstacle target to be detected.

[0013] Preferably, the step 2) comprises:

[0014] Step 2-1) taking N adjacent frames of beam data as a group, performing difference processing on the sonar beam data at time T and the sonar beam data at time T-1 to obtain inter-frame sonar beam difference data;

[0015] Step 2-2) inputting the sonar beam differential data into a constant false alarm detector;

[0016] Step 2-3) inputting the sonar beam data at time T into the constant false alarm detector;

[0017] Step 2-4) combining the output of the constant false alarm detector in step 2-2) with the output of the constant false alarm detector in step 2-3) to obtain the output result of the inter-frame difference constant false alarm detector.

[0018] Preferably, the step 3) comprises:

[0019] For the output result of the inter-frame difference constant false alarm detector obtained in step 2), a clustering method is used to obtain the scale information of the target to be detected according to the number of occupied detection units, and the position information of the target to be detected is obtained by the row and column indexes of the beam data.

[0020] Preferably, the step 4) comprises:

[0021] Data association is performed on the clustering results obtained in step 3), and the measurement value of the tracked target is updated. After data association, Kalman filtering is performed on the tracked target.

[0022] Preferably, the data association strategy is:

[0023] Step S1) Calculate all target locations {(x 1,k ,y 1,k ),…,(x n,k ,y n,k )} and the jth target position in the tracking trajectory at time k-1 Distance Where n represents the number of targets in the clustering result at the current moment, j = 1, 2, ..., m, and m represents the number of tracking trajectories at the current moment;

[0024] Step S2) threshold determination: If If there is an element less than or equal to the set radius r in The smallest element d in i The corresponding target position (x i,k ,y i,k ) as the measurement value of the current moment of the tracking trajectory j, and the target position (x i,k ,y i,k ) mark, which does not participate in the calculation of subsequent new target trajectory generation; where i = 1, 2, ... n;

[0025] like If there is no element with a radius less than or equal to the set radius r in , it is further determined whether the tracking target j is dead: if the tracking target j has no updated measurement value for more than 4 frames, the tracking target is determined to be dead and the corresponding tracking trajectory is terminated;

[0026] Step S3) After j is incremented by 1, go to step S1) until j>m, i.e. the measurement value of each tracking trajectory is updated, and go to step S4);

[0027] Step S4) After all the tracking trajectories have been updated, all the unmarked targets in the clustering results are judged to generate new tracking target trajectories, and the positions (x t,k ,y t,k ) and all target positions (x) in the previous 5 frames of k t,k-1 ,y t,k-1 ),(x t,k-2 ,y t,k-2 ),…,(x t,k-5 ,y t,k-5 ) in They represent the positions of all targets and unmarked targets at time k-1, k-2, ... k-5 respectively (x t,k ,y t,k ) distance, if there are more than 3 frames If the radius is less than or equal to the set radius r, it is determined that a new target has appeared and a new tracking trajectory is established.

[0028] Preferably, the set radius r is determined by the size of the obstacle target and the movement speed of the aircraft.

[0029] Compared with the prior art, the advantages of the present invention are:

[0030] 1. The target detection method of first detecting and then tracking is applied to the obstacle target detection of the underwater unmanned vehicle collision avoidance system, in which the detection part and the tracking part can be optimized independently.

[0031] 2. The constant false alarm detector using the background cancellation method has a higher detection rate than the constant false alarm detector using the conventional method, and can detect targets at a farther distance, providing more operating space for underwater vehicles to avoid obstacles and perform corresponding avoidance actions.

[0032] 3. The method of the present invention further enhances the robustness of the detection algorithm in complex environments, improves the efficiency of the collision avoidance system in detecting obstacle targets, and ensures the early warning capability of the collision avoidance system and the confidence level of the target existence. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is the obstacle target detection algorithm that detects first and then tracks the framework;

[0034] Figure 2 It is an inter-frame "differential" constant false alarm detector;

[0035] FIG3 is the effect of the inter-frame “difference” operation, wherein FIG3(a) is a sound image, FIG3(b) is a target detection result image of a conventional constant false alarm detector, and FIG3(c) is a result image of an inter-frame “difference” constant false alarm detector;

[0036] Figure 4 It is the processing flow chart of the target detection module;

[0037] Figure 5 It is the processing flow chart of the tracking filter module;

[0038] Figure 6 is the processing result of the simulation experiment, where Figure 6(a) is the acoustic image, Figure 6(b) is the target detection result, Figure 6(c) is the tracking filter result, Figure 6(d) is the comparison between the real trajectory and the target tracking trajectory, Figure 6(e) is the comparison between the target tracking trajectory and the actual trajectory in the horizontal and vertical directions, and Figure 6(f) is the OSPA index;

[0039] Figure 7 is the result of the water pool experiment, where Figure 7(a) is the acoustic image, Figure 7(b) is the target detection result, Figure 7(c) is the tracking filter result, Figure 7(d) is the comparison between the true trajectory and the target tracking trajectory, Figure 7(e) is the comparison between the target tracking trajectory and the actual trajectory in the horizontal and vertical directions, and Figure 7(f) is the OSPA index. DETAILED DESCRIPTION

[0040] In the context of the increasing use of underwater vehicles, the application of collision avoidance systems is crucial to ensure the navigation safety of vehicles. The obstacle target detection part provides important information reference for the early warning and maneuvering of the collision avoidance system. In order to suppress reverberation and improve the probability of target detection, the present invention uses a method of first detection and then tracking in the target detection process of the collision avoidance system. This method has high execution efficiency while taking into account the detection performance and can meet the real-time requirements of mobile devices. In order to solve the problem that the echo signal of the obstacle target is masked by reverberation, the background cancellation method is introduced to suppress the reverberation interference and improve the detection probability of the obstacle target in the reverberation background environment. In the detection part, the OS-CFAR constant false alarm detector is used to adaptively adjust the threshold value to improve the target detection rate. In the tracking part, the Kalman filter is introduced to eliminate the reverberation interference, screen out the obstacle target, and stably monitor the obstacle target to form a continuous and stable target tracking trajectory relative to the sonar equipment.

[0041] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0042] Example

[0043] Conventional target detection algorithms inevitably lead to missed detections and false alarms. Therefore, in order to solve the problem of obstacle target detection in an underwater unmanned vehicle collision avoidance system, an embodiment of the present invention proposes an obstacle target detection method for an unmanned underwater vehicle collision avoidance system. A target detection algorithm of first detection and then tracking is introduced. The detection part and the tracking part can be independently optimized to adapt to different scenarios, which is suitable for the underwater vehicle collision avoidance system to work in a complex environment. In the target detection framework of first detection and then tracking, the detection part uses a constant false alarm detector with anti-interference ability to improve the detection probability of the target, and the tracking part uses a Kalman filter to eliminate reverberation interference and reduce false alarms.

[0044] The method comprises the following steps:

[0045] Step 1) receiving beam data collected by the collision avoidance sonar of the unmanned underwater vehicle;

[0046] Step 2) using the inter-frame difference method to perform background cancellation processing on the adjacent N frames of beam data;

[0047] Step 3) using a clustering method on the background-offset processed data to obtain scale information and position information of the target to be detected;

[0048] Step 4) Perform data association, perform tracking filtering processing based on the results of data association and output the target detection results.

[0049] 1. Obstacle target detection algorithm process of underwater unmanned vehicle collision avoidance system

[0050] The processing flow of the present invention is as follows Figure 1 As shown in the figure, the beam data in the collision avoidance sonar is used as input, and the background cancellation processing is first performed on the input multi-frame sonar data, and then the threshold judgment is performed on the processed data to determine the position of the target, and finally the tracking filtering processing is performed according to the result of the threshold judgment and the target detection result is output.

[0051] In the background offset part, the inter-frame "difference" method is used to suppress the background environment. The background offset can improve the contrast of the target signal in the background environment, making it easier for the constant false alarm detector to detect the target and improve the detection probability of the system. The target detection part is mainly composed of the inter-frame difference constant false alarm detector and the Dbscan clustering algorithm. The inter-frame difference constant false alarm detector can adaptively adjust the threshold according to the background environment, and has the ability to balance the background environment noise and improve the target detection while keeping the false alarm rate constant. The tracking filter part is mainly composed of data association and Kalman filter. In order to improve the confidence of the existence of obstacle targets, reduce the false alarm rate of target detection, obtain stable and continuous relative motion traces of the detection target, and eliminate reverberation interference, the target tracking algorithm is used. Finally, the target is identified according to the output of the tracking filter and the output of the target detection module to obtain the final detection result of the system.

[0052] Inter-frame "differential" constant false alarm detector

[0053] The processing flow of the inter-frame "difference" method is as follows Figure 2 As shown, several adjacent frames of data (need to be determined according to the relative movement speed of the unmanned aerial vehicle and the obstacle target, usually 3-5 frames) are taken as a group, and the sonar beam data at the current moment is subtracted from the data at the previous moment to obtain the inter-frame "differential" sonar beam data, and then the inter-frame "differential" data is input into the OS-CFAR detector; because the differential operation will eliminate stationary targets, in order to retain the information of stationary targets in the sonar data, the sonar beam data at the current moment also needs to be input into the OS-CFAR detector for processing; the outputs of the two are combined to obtain the output result of the inter-frame "differential" constant false alarm detector.

[0054] Figure 3 shows the benefits of the inter-frame difference operation, where Figure 3(a) is the forward-looking sonar beam intensity diagram, Figure 3(b) is the result after conventional CFAR detection, and Figure 3(c) is the result of the inter-frame "difference" constant false alarm detector detection. By comparison, it can be seen that the target intensity is not obvious under the interference of the reverberant background environment, and the target is submerged in the background environment. After processing using the inter-frame "difference" constant false alarm detector, the influence of the background environment can be significantly suppressed, and the detection of obstacle targets can be completed.

[0055] Object Detection Module

[0056] The target detection module consists of two parts: the inter-frame "difference" constant false alarm detector and the Dbscan clustering algorithm. Figure 4 As shown in the figure, the output of the inter-frame "difference" constant false alarm detector is used as the input of the clustering algorithm (Dbscan). Through the processing of the inter-frame "difference" constant false alarm detector and the clustering algorithm, the target detection module completes the detection of the sonar data, and its output contains the location information and scale information of the target. The reason for introducing the clustering algorithm is that the obstacle target is usually an extended target, which usually occupies multiple detection units (cells) in the inter-frame "difference" constant false alarm detector. Therefore, these detection units need to be attributed to one class. The scale information of the target can be judged according to the number of detection units occupied by the clustering result, and the row and column indexes of the beam data (usually a matrix) contain the location information of the real world. Therefore, the combination of the two can determine the location information and scale information of the sonar device from the obstacle target.

[0057] Tracking filter module

[0058] In actual sonar images, the sound waves returned by scatterers will reflect multiple times on the seabed and surface, which will show a "flickering" characteristic similar to the target on the sonar image, resulting in many suspected target bright spots still existing after the target detection module processes them. In order to obtain a higher detection probability and confidence of obstacle targets, reduce the false alarm rate of target detection, and eliminate reverberation interference, it is necessary to introduce a tracking algorithm.

[0059] The input of the tracking filter module is the target position information output by the target detection module, and the output is the target status information. The processing flow of the tracking filter module is as follows: Figure 5 As shown in Figure 1, it consists of two parts: data association and Kalman filter. The data association part includes distance threshold judgment, target birth, target death and target measurement value update. The distance judgment uses formula (1), which means that only the measurement value at a certain distance from the tracked target can be included in the measurement value of the corresponding trajectory. The threshold value r needs to be set according to the motion state of the target and the sonar device.

[0060]

[0061] where x i,k ,y i,k represents the horizontal and vertical positions of target i at time k in the rectangular coordinate system; represents the estimated values ​​of the horizontal and vertical positions of the target j at time k in the rectangular coordinate system; d represents the distance between the two; r represents the decision threshold, which is determined by the size of the obstacle target and the movement speed of the aircraft.

[0062] After the threshold judgment, the measurement value of the tracked target is updated. If the tracked target is not updated for many times, the target is considered dead and the tracking trajectory is terminated. If there is a measurement value that is not included in any tracked target, it enters the new target library. If the new target appears many times, it can be judged that the target exists and a new trajectory is established for it. After data association, the tracked target is Kalman filtered and finally output according to the state data after Kalman filtering and the scale information of the target in the detection module.

[0063] The specific data association strategy is as follows:

[0064] Calculate all target locations in the clustering result of the current time k {(x 1,k ,y 1,k ),…,(x n,k ,y n,k )} and the jth target position in the tracking trajectory at time k-1 Distance Where n represents the number of targets in the clustering result at the current moment, j = 1, 2, ..., m, and m represents the number of tracking trajectories at the current moment;

[0065] Perform threshold judgment, if If there is an element less than or equal to the set radius r in The smallest element d in i (i=1,2,…n) corresponding to the target position (x i,k ,y i,k ) as the measurement value of the current moment of the tracking trajectory j, and the target position (x i,k ,y i,k ) mark, and does not participate in the calculation of subsequent new target trajectory generation;

[0066] like If there is no element with a radius less than or equal to the set radius r in , it is necessary to determine whether the tracking target j is dead. If the tracking target j has no updated measurement value for more than 4 frames, it is determined that the tracking target is dead and the corresponding tracking trajectory is terminated;

[0067] After j is superimposed with j=j+1, the above operation is repeated until j>m, that is, the measurement value of each tracking trajectory is updated;

[0068] After all the tracking trajectories have been updated, the new tracking target trajectory generation is judged for all the unmarked targets in the clustering results, and the positions (x t,k ,y t,k ) and all target positions (x) in the previous 5 frames of k t,k-1 ,y t,k-1 ),(x t,k-2 ,y t,k-2),…,(x t,k-5 ,y t,k-5 ) in They represent the positions of all targets and unmarked targets at time k-1, k-2, ... k-5 respectively (x t,k ,y t,k ), if there are more than 3 frames with a distance less than or equal to the set radius r, it is determined that a new target appears and a new tracking trajectory is established.

[0069] Obstacle target detection algorithm results

[0070] In order to test the applicability of the method and the accuracy of detecting targets, simulation experiments and water tank experiments were carried out and the performance of the algorithm was qualitatively analyzed.

[0071] As shown in Figure 6(a), it is a simulated acoustic image of a single target under 10dB conditions. The target moves at a speed of 0.4m / s and gradually approaches the sonar from far to near. The active sonar transmits a CW signal with a frequency of 200KHz, a pulse width of 30us, and a range of 35m. Figure 6(b) is the result of the simulated acoustic image after differential CFAR processing. Figure 6(c) is the trajectory and azimuth of the target after processing by the detection algorithm. Figure 6(d) is the real trajectory of the target and the trajectory of the target detection. Figure 6(e) is a schematic diagram of the real trajectory and target state estimation value in the X and Y directions over time. Figure 6(f) is the OSPA performance analysis of the detection algorithm at 10dB, where the calculation method of OSPA is shown as follows (2). The smaller the value of OSPA, the higher the multi-target tracking accuracy.

[0072]

[0073] where X = {x1, x2, …, x m} and Y={y1,y2,…,y n} represent the true target set and the estimated target set respectively; m and n represent the number of true targets and estimated targets respectively; c represents the cutoff parameter, which is set to 30; p is the order parameter, which is set to 1; ||x i -y π(i) || represents the i-th true target x i With the estimated target y of the distribution π(i) The distance between p *|nm| indicates the quantitative error caused by the unmatched target.

[0074] The processing results of the water pool experiment are shown in Figure 7. The data after the forward-looking sonar scans the target is plotted as shown in Figure 7(a). The target is marked by a red frame. The echo signal of the target is submerged in the strong reverberation environment, and a single frame image is difficult to identify. After the inter-frame "difference" constant false alarm detection, the detection result is shown in Figure 7(b). The target is successfully detected, but the detection result contains a lot of reverberation interference. It is difficult to determine the direction and distance of the target based on the result of the constant false alarm detection alone. The constant false alarm rate of OS-CFAR is set to 10 -4 , N = 32. Figure 7(c) shows the result after being processed by the tracking filter module. By comparing with Figure 7(a), it can be found that after tracking processing, the target can be detected from a large number of reverberation interferences and the target's motion trajectory can be obtained. Figure 7(d) is the motion trajectory of the target's true motion trajectory and the target's state estimation. Figure 7(e) is a schematic diagram of the true trajectory in the X and Y directions and the estimated value of the tracking method over time. Figure 7(f) is a schematic diagram of the OSPA of the improved method and the traditional method in the pool test, where the horizontal axis is time and the vertical axis is the OSPA index.

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention is described in detail with reference to the embodiments, it should be understood by those skilled in the art that any modification or equivalent replacement of the technical solutions of the present invention does not depart from the spirit and scope of the technical solutions of the present invention and should be included in the scope of the claims of the present invention.

Claims

1. An obstacle target detection method for an unmanned underwater vehicle collision avoidance system, comprising: Step 1) receiving beam data collected by the collision avoidance sonar of the unmanned underwater vehicle; Step 2) using the inter-frame difference method to perform background cancellation processing on the adjacent N frames of beam data; Step 3) using a clustering method on the background-offset processed data to obtain scale information and position information of the target to be detected; Step 4) Perform data association, perform tracking filtering processing based on the results of data association and output the target detection results.

2. The obstacle target detection method for an unmanned underwater vehicle collision avoidance system according to claim 1, characterized in that: The N frame value of step 2) is determined according to the relative movement speed between the unmanned aerial vehicle and the obstacle target to be detected.

3. The obstacle target detection method for an unmanned underwater vehicle collision avoidance system according to claim 2, characterized in that: The step 2) comprises: Step 2-1) taking N adjacent frames of beam data as a group, performing difference processing on the sonar beam data at time T and the sonar beam data at time T-1 to obtain inter-frame sonar beam difference data; Step 2-2) inputting the sonar beam differential data into a constant false alarm detector; Step 2-3) inputting the sonar beam data at time T into the constant false alarm detector; Step 2-4) combining the output of the constant false alarm detector in step 2-2) with the output of the constant false alarm detector in step 2-3) to obtain the output result of the inter-frame difference constant false alarm detector.

4. The obstacle target detection method for an unmanned underwater vehicle collision avoidance system according to claim 3, characterized in that: The step 3) comprises: For the output result of the inter-frame difference constant false alarm detector obtained in step 2), a clustering method is used to obtain the scale information of the target to be detected according to the number of occupied detection units, and the position information of the target to be detected is obtained by the row and column indexes of the beam data.

5. The obstacle target detection method for an unmanned underwater vehicle collision avoidance system according to claim 1, characterized in that: The step 4) comprises: Data association is performed on the clustering results obtained in step 3), and the measurement value of the tracked target is updated. After data association, Kalman filtering is performed on the tracked target.

6. The obstacle target detection method for an unmanned underwater vehicle collision avoidance system according to claim 5, characterized in that: The data association strategy is: Step S1) Calculate all target locations {(x 1,k ,y 1,k ),…,(x n,k ,y n,k )} and the jth target position in the tracking trajectory at time k-1 Distance Where n represents the number of targets in the clustering result at the current moment, j = 1, 2, ..., m, and m represents the number of tracking trajectories at the current moment; Step S2) threshold determination: If If there is an element less than or equal to the set radius r in The smallest element d in i The corresponding target position (x i,k ,y i,k ) as the measurement value of the current moment of the tracking trajectory j, and the target position (x i,k ,y i,k ) mark, which does not participate in the calculation of subsequent new target trajectory generation; where i = 1, 2, ... n; like If there is no element with a radius less than or equal to the set radius r in , it is further determined whether the tracking target j is dead: if the tracking target j has no updated measurement value for more than 4 frames, the tracking target is determined to be dead and the corresponding tracking trajectory is terminated; Step S3) After j is incremented by 1, go to step S1) until j>m, i.e. the measurement value of each tracking trajectory is updated, and go to step S4); Step S4) After all the tracking trajectories have been updated, all the unmarked targets in the clustering results are judged to generate new tracking target trajectories, and the positions (x t,k ,y t,k ) and all target positions (x) in the previous 5 frames of k t,k-1 ,y t,k-1 ),(x t,k-2 ,y t,k-2 ),…,(x t,k-5 ,y t,k-5 ) in They represent the positions of all targets and unmarked targets at time k-1, k-2, ... k-5 respectively (x t,k ,y t,k ) distance, if there are more than 3 frames If the radius is less than or equal to the set radius r, it is determined that a new target has appeared and a new tracking trajectory is established.

7. The obstacle target detection method for an unmanned underwater vehicle collision avoidance system according to claim 6, characterized in that: The set radius r is determined by the size of the obstacle target and the movement speed of the aircraft.

Images