A global cooperative collision avoidance method in a mixed traffic environment of manned and unmanned ships

Abstract

The present invention provides a global cooperative collision avoidance method in a mixed traffic environment of manned ships and unmanned ships, including: obtaining the navigation information of all ships within the global scope; calculating the ship motion parameters according to the navigation information of the ships, where the ship motion parameters include the distances between ships, the minimum encounter distance, and the shortest encounter time; using the five-factor weighting method to calculate the collision risk between ships according to the distances between ships, the minimum encounter distance, the shortest encounter time, the relative bearing, and the ship speed ratio; determining whether the global collision risk is greater than a threshold value. If it is greater than the threshold value, the improved NSGA-II algorithm is used to calculate the global collision avoidance decision. If it is less than the threshold value, no collision avoidance operation is required and the ship sails straight normally. The present invention can solve the problem of multi-ship cooperative collision avoidance in a mixed traffic environment of manned ships and unmanned ships, and improves the navigation economy on the premise of ensuring the navigation safety of ships.

Description

Technical Field

[0001] The present invention relates to the technical field of ship traffic management. Specifically, it particularly relates to a global collaborative collision avoidance method in a mixed traffic environment of manned ships and unmanned ships. Background Art

[0002] In recent years, with the development of intelligent shipping technology, unmanned ships have been increasingly widely used in the field of water transportation, and the scenario of coexistence of manned ships and unmanned ships has been increasing. To ensure maritime traffic safety, it is urgent to solve the problem of ship collision avoidance in the mixed traffic environment of manned ships and unmanned ships. Currently, the research on multi-ship collision avoidance in the mixed traffic environment of manned ships and unmanned ships in the industry mainly focuses on intelligent collision avoidance decision-making methods. Most of these methods are centered around the ship itself and use different algorithms to achieve intelligent ship collision avoidance. Each ship can sense the surrounding environment in real time and make collision avoidance decisions independently according to collision avoidance rules.

[0003] Currently, ship autonomous collision avoidance decision-making methods mainly include three categories. One is the deterministic collision avoidance decision-making method, the second is the collision avoidance decision-making method based on machine learning, and the third is the heuristic collision avoidance decision-making method.

[0004] The deterministic collision avoidance decision-making method refers to a method that realizes ship autonomous collision avoidance through relevant knowledge of mathematics and nautical science. This type of method usually needs to rely on the subjective understanding of relevant experts, scholars or nautical practitioners to form an expert system. For example, some scholars have proposed a collision avoidance algorithm based on local normal distribution. This algorithm can generate the best path that follows collision avoidance rules according to the normal distribution of the bell curve and can also be used in scenarios with static and dynamic obstacles to ensure the safe navigation of ships. Other scholars have proposed a trajectory-based algorithm to calculate the best path for ships to ensure navigation safety. The algorithm can be applied to the ship decision-making system. This method solves the problem of ship collision avoidance path planning and takes into account static and dynamic obstacles in the environment, maritime traffic rules, and the dynamic characteristics of ships. The deterministic collision avoidance decision-making algorithm involves many human factors, and the obtained collision avoidance results are usually only feasible solutions rather than optimal solutions, and there may even be errors in the results due to analysis mistakes.

[0005] The collision avoidance decision-making method based on machine learning uses machine learning algorithms to solve the ship collision avoidance problem. For example, some scholars, aiming at the automatic collision avoidance problem of unmanned ships under complex sea navigation conditions, proposed an intelligent collision avoidance navigation method for unmanned ships considering navigation experience and collision avoidance rules based on the deep competitive Q-learning algorithm and the A* algorithm. This method maps collision avoidance experience into dynamic navigation restriction lines, formulates avoidance strategies that conform to collision avoidance rules and navigation experience, and optimizes the performance of model training and path planning in the calculations of reinforcement learning and A* local dynamic obstacle avoidance to improve the efficiency of the method. There are also scholars who proposed an autonomous ship collision avoidance method based on attention and deep reinforcement learning (ADRL) to solve the collision avoidance problem in the case of multi-ship encounters. Starting from the attention distribution of the driver, this method encodes ship information through a local map and calculates the collision avoidance decision of each ship in real time in the form of attention scores under the constraints of the International Regulations for Preventing Collisions at Sea. Such collision avoidance decision-making methods based on machine learning are difficult in the selection of the training set and the setting of the cost function and parameters, and there is also a problem that a set of parameters is applicable to some specific ship collision avoidance scenarios but not to general scenarios.

[0006] The heuristic collision avoidance decision-making method refers to using heuristic algorithms to solve the ship collision avoidance problem. The heuristic algorithm is a method of continuously searching for feasible solutions in the solution space. The collision avoidance process is essentially an optimization process of feasible solutions, so many experts and scholars choose to use heuristic algorithms to solve the ship collision avoidance problem. For example, some scholars used a genetic algorithm with multiple populations to make auxiliary decisions for ship collision avoidance, taking the course deviation of the ship and the safety of the ship as the fitness function, and taking the minimum retention generation of the optimal individual in the elite population as the termination condition of the algorithm. The simulation results show that the genetic algorithm with multiple populations can effectively make auxiliary decisions for ship collision avoidance. Ship collision avoidance is a complex problem, and multiple factors such as the collision avoidance responsibilities of each ship, the timing of collision avoidance, intention judgment, behavior interaction, and the timing of resuming navigation need to be fully considered to comprehensively obtain the steering angle and steering action of the ship. The heuristic collision avoidance decision-making method can effectively solve the ship collision avoidance problem. By optimizing the objective function, it can select the optimal strategy from the feasible solutions and obtain the optimal collision avoidance strategy in the multi-ship encounter scenario. However, such algorithms need to design appropriate fitness functions and have problems such as insufficient global search ability and slow convergence speed.

[0007] In the sea area scenario where manned ships and unmanned ships coexist, different ship drivers may have inconsistent identifications of ship collision risks. If each ship adopts its own way to perform collision avoidance operations, it is easy to cause uncoordinated collision avoidance actions, resulting in collision avoidance failure or too long deviation distance, reducing the safety and economy of ship navigation. Summary of the Invention

[0008] The present invention proposes a global cooperative collision avoidance method in a mixed traffic environment of manned and unmanned ships, which solves the problems of increased ship collision risk or excessive deviation distance caused by differences in ship collision risk recognition among different ship drivers in a mixed traffic environment of manned and unmanned ships.

[0009] The technical means adopted by the present invention are as follows:

[0010] A global cooperative collision avoidance method in a mixed traffic environment of manned and unmanned ships, comprising:

[0011] Obtaining the navigation information of all ships within the global scope;

[0012] Calculating ship motion parameters based on the navigation information of the ships, where the ship motion parameters include the distance between ships, the minimum distance of approach, and the shortest time to encounter;

[0013] Adopting a five-factor weighting method to calculate the collision risk between ships based on the distance between ships, the minimum distance of approach, the shortest time to encounter, the relative bearing, and the speed ratio;

[0014] Judging whether the global collision risk is greater than the threshold. If it is greater than the threshold, use the improved NSGA-II algorithm to calculate the global collision avoidance decision. If it is less than the threshold, no collision avoidance operation is required, and the ship sails straight normally.

[0015] Further, obtaining the navigation information of all ships within the global scope includes obtaining the navigation information of all ships within the global scope through environmental perception devices, where the navigation information includes ship position, course, and speed.

[0016] Further, the minimum distance of approach is calculated according to the following formula:

[0017] DCPA = Dsin(θ AB -α B -π)

[0018] where DCPA represents the minimum distance of approach, θ AB represents the relative velocity direction of ship A with respect to ship B, α B represents the true bearing of ship B relative to ship A, and D represents the distance between ship A and ship B;

[0019] The shortest time to encounter is calculated according to the following formula:

[0020]

[0021] where TCPA represents the shortest time to encounter, v AB represents the relative velocity of ship A with respect to ship B, θ AB represents the relative velocity direction of ship A with respect to ship B, α BLet \(\theta\) represent the true azimuth of ship B relative to ship A, and D represent the distance between ship A and ship B.

[0022] Furthermore, the five-factor weighting method is adopted to calculate the collision risk degree between ships according to the distance between ships, the minimum distance of approach, the shortest time to collision, the relative azimuth, and the ship speed ratio, including:

[0023] Solve the risk membership function of the minimum distance of approach according to the following method:

[0024]

[0025] Among them, \(U(DCPA)\) represents the risk membership function of the minimum distance of approach, \(d_1\) represents the dangerous distance of approach, and \(d_2\) represents the safe distance of approach;

[0026] Solve the risk membership function of the shortest time to collision according to the following method:

[0027]

[0028]

[0029]

[0030] Among them, \(U(TCPA)\) represents the risk membership function of the shortest time to collision, \(d_1\) represents the dangerous distance of approach, and \(V\) R represents the magnitude of the relative speed;

[0031] Solve the risk membership function of the distance between two ships according to the following method:

[0032]

[0033] Among them, \(U(D)\) represents the risk degree function of the distance between two ships, \(d_1\) represents the dangerous distance of approach, and \(d_2\) represents the safe distance of approach;

[0034] Solve the risk membership function of the azimuth of the approaching ship according to the following method:

[0035]

[0036] Among them, \(U(B)\) represents the risk membership function of the azimuth of the approaching ship, and B represents the azimuth of the approaching ship;

[0037] Solve the risk membership function of the ship speed ratio according to the following method:

[0038]

[0039] Among them, \(U(K)\) represents the risk membership function of the ship speed ratio, K represents the ship speed ratio, and C represents the course of the own ship.

[0040] Further, the five-factor weighting method is adopted to calculate the collision risk degree between ships according to the distance between ships, the minimum encounter distance, the shortest encounter time, the relative bearing, and the ship speed ratio. It also includes:

[0041] Solve the global collision risk degree according to the following method:

[0042]

[0043] Among them, GlobalCRI represents the global collision risk degree, N represents the number of ships, and CRI(i,j) represents the collision risk degree between two ships.

[0044] Further, the improved NSGA-II algorithm improves the traditional NSGA-II algorithm from two aspects: population initialization and population iterative evolution:

[0045] The population initialization algorithm based on Latin hypercube sampling performs stratification and sampling operations according to the number of samples within the optional steering angle range of each ship, and after merging the sampling results of different ships, a uniformly distributed initial population is formed;

[0046] Select the solution with the largest crowding degree in the population as the brightest firefly individual, calculate the attraction degree of the brightest individual to other individuals and move the positions of other individuals to form a new population after firefly perturbation.

[0047] Further, using the improved NSGA-II algorithm to calculate the global collision avoidance decision also includes:

[0048] After iterative optimization and solution through the improved NSGA-II algorithm, the final non-dominated solution set will be obtained;

[0049] Screen the global collision avoidance decisions in the non-dominated solution set according to the avoidance angle and the relative velocity direction, so as to obtain the final global collision avoidance strategy.

[0050] Compared with the existing technology, the present invention has the following advantages:

[0051] In the collision avoidance process of the method of the present invention, the collision avoidance action decisions of each ship are uniformly formulated. The global collision risk degree is calculated according to the navigation parameters of the ships in the sea area. After reaching the threshold, the multi-ship cooperative collision avoidance decision is calculated and sent to each ship. There is no need for communication between ships, and they only need to navigate according to the instructions. This method can solve the problem of multi-ship cooperative collision avoidance in the mixed traffic environment of manned ships and unmanned ships, and improves the navigation economy on the premise of ensuring the navigation safety of ships. Description of the Drawings

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0053] Figure 1 This is a flowchart of the global collaborative collision avoidance method in a mixed traffic environment of manned ships and unmanned ships according to the present invention.

[0054] Figure 2 This is the initial ship position diagram in the embodiment of the present invention.

[0055] Figure 3 This is the collaborative collision avoidance effect diagram in the embodiment of the present invention.

[0056] Figure 4 This is the Pareto front comparison diagram in the embodiment of the present invention.

[0057] Figure 5 This is the comparison diagram of the collision avoidance effects before and after improving the NSGA-II algorithm in the embodiment of the present invention. Among them, (a) is the collision avoidance effect of the method of the present invention in scenario 1, (b) is the collision avoidance effect of the method of the present invention in scenario 2, (c) is the collision avoidance effect of the method of the present invention in scenario 3, and (d) is the comparison of the global avoidance angles. Detailed implementation manners

[0058] In order to enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0059] This embodiment provides a global collaborative collision avoidance method in a mixed traffic environment of manned ships and unmanned ships. This method solves the problem that different ship drivers have different understandings of ship collision risks in a mixed traffic environment of manned ships and unmanned ships, resulting in an increase in ship collision risk or an excessive detour distance. The flow of this method is as Figure 1 shown, and the specific steps are as follows.

[0060] Step 1: Obtain the navigation information of all ships within the global scope.

[0061] Obtain the navigation information of all ships within the global scope through environmental perception devices such as AIS ship automatic identification system, GPS global positioning system, electronic compass, magnetic compass, etc., mainly including ship position, course, speed, etc.

[0062] Step 2: Calculate the ship motion parameters.

[0063] Calculate the distance (D), minimum passing distance (DCPA), and shortest passing time (TCPA) between ships according to the parameters between ships. Assume the position of ship A is (x A , y A ), speed is v A , and course is θ A , and the position of ship B is (x B , y B ), speed is v B , and course is θ B .

[0064] The components of the speeds of the two ships on the X and Y axes are shown in formula (1).

[0065]

[0066] The components of the relative speed of the two ships on the X and Y axes are shown in formula (2).

[0067]

[0068] The direction of the relative speed θ AB is shown in formula (3).

[0069]

[0070] The relative distance between the two ships is shown in formula (4).

[0071]

[0072] The true azimuth of ship B relative to ship A is shown in formula (5).

[0073]

[0074] The calculation method of DCPA between the two ships is shown in formula (6).

[0075] DCPA = Dsin(θ AB - α B - π) (6)

[0076] The calculation method of TCPA between the two ships is shown in formula (7).

[0077]

[0078] Step 3: Calculate the global collision risk.

[0079] The calculation method of the collision risk between ships adopts the five-factor weighting method, comprehensively considering DCPA, TCPA, the distance between the two ships (D), the relative bearing (B), and the speed ratio (K). The target factor set U = [DCPA, TCPA, D, B, K] is composed of the parameters of the target ship. Considering the above five factors, the collision risk CRI between ships is finally calculated as CRI = a DCPA U DCPA + a TCPA U TCPA + a D U D + a B U B + a K U K , where a DCPA , a TCPA , a D , a B , a K are the weights of different parameters, and the sum of the five weights is 1.

[0080] The degree of danger caused by DCPA to the own ship is obvious. The larger this value is, the smaller the degree of danger. The danger membership function of DCPA is shown in formula (8), where d1 is the dangerous encounter distance and d2 is the safe encounter distance.

[0081]

[0082] The degree of danger caused by TCPA to the own ship is also obvious. The larger this value is, the smaller the degree of danger. The danger membership function of TCPA is shown in formula (9).

[0083]

[0084] Among them, U(TCPA) represents the danger membership function of the shortest encounter time, d1 represents the dangerous encounter distance, and V R represents the magnitude of the relative speed.

[0085] The distance between the two ships will also cause the ship's risk. The closer to the target, the greater the degree of danger of the target ship to the own ship. The risk function of the distance between the two ships is shown in formula (10).

[0086]

[0087] Among them, d1 represents the dangerous encounter distance and d2 represents the safe encounter distance.

[0088] Oncoming vessels in different directions pose different levels of danger to the own ship. Generally speaking, the danger level on the starboard side is greater than that on the port side, and the danger level ahead of the beam is greater than that behind the beam. When the oncoming vessel is at a 19° angle off the bow of the own ship, it is the most dangerous, and when it is at a 199° angle off the bow, it is the safest. The danger membership function of the oncoming vessel's bearing can be obtained as shown in formula (11).

[0089]

[0090] Among them, U(B) represents the danger membership function of the oncoming vessel's bearing, and B represents the oncoming vessel's bearing.

[0091] If other conditions of the encountering vessels are the same, only the ship speed ratio is different, the resulting collision danger will also be different. At the same distance between the two ships and the same collision angle, when taking the same turning amplitude, if the ship speed ratio is larger, the new DCPA obtained after turning is smaller. At the same time, the larger the ship speed ratio, that is, the slower the own ship, the worse the effect of the rudder turning, and the earlier the timing of the action taken. Therefore, it can be concluded that the larger the ship speed ratio, the greater the collision danger level. The danger membership function of the ship speed ratio is established as shown in formula (12).

[0092]

[0093] Among them, K represents the ship speed ratio, and C represents the own ship's course.

[0094] The global collision danger level is the average value of the danger levels of each ship. Since the ship danger level matrix is a symmetric matrix, the global collision danger level can be calculated based on the upper half of the ship danger level matrix, and the calculation method is as shown in formula (13).

[0095]

[0096] Step 4: Determine whether the global collision danger level is greater than the threshold value.

[0097] Determine whether the global collision danger level calculated in step 3 is greater than the threshold value. If it is greater than the threshold value, the improved NSGA-II algorithm needs to be used to calculate the global collision avoidance decision. If it is less than the threshold value, no collision avoidance operation is required, and the ship sails straight normally.

[0098] Step 5: Calculate the ship cooperative collision avoidance decision based on the improved NSGA-II algorithm.

[0099] Ship collision avoidance itself is a complex problem that requires comprehensive consideration of various factors to obtain the final collision avoidance decision. Therefore, multi-objective optimization algorithms are more suitable for solving ship collision avoidance problems. The NSGA-II algorithm, namely the non-dominated sorting genetic algorithm, is a multi-objective optimization algorithm based on the dominance relationship and Pareto optimal solutions. Pareto optimal solutions are also called non-dominated solutions. The performance of these solutions in different optimization objectives is better than that of other feasible solutions, and these non-dominated solutions constitute the Pareto front. Compared with traditional genetic algorithms, the NSGA-II algorithm can solve multi-objective optimization problems. In addition, a fast non-dominated sorting algorithm is introduced into the algorithm, which reduces the time complexity of the algorithm.

[0100] This invention designs two optimization objectives: the global avoidance angle and the ship encounter distance. The algorithm generates an intermediate population through selection, crossover, and mutation operations, and selects excellent individuals to enter the next generation population according to the dominance level and crowding distance. Through iterative operations, multi-objective optimization is achieved. However, there are a large number of random operations in the traditional NSGA-II algorithm during the initialization of the population and the iterative evolution of the population, and there are certain deficiencies in the global search ability and convergence speed.

[0101] This step improves the traditional NSGA-II algorithm from two aspects: population initialization and population iterative evolution, and proposes a population initialization method based on Latin hypercube sampling and a population iterative evolution method based on the firefly perturbation strategy, which improves the global search ability and convergence speed of the algorithm; then uses the improved NSGA-II algorithm to solve the ship collision avoidance problem and obtains an optimized collision avoidance strategy.

[0102] (1) Population initialization algorithm based on Latin hypercube sampling.

[0103] The traditional NSGA-II algorithm adopts a population initialization method based on a random function, that is, randomly selects individuals between the variable upper and lower bounds as the initialized population. However, random sampling has problems such as uneven distribution and inability to cover the entire feasible domain. Therefore, this invention proposes a population initialization algorithm based on Latin hypercube sampling, which performs stratification and sampling operations according to the number of samples within the optional steering angle range of each ship, and combines the sampling results of different ships to form a uniformly distributed initialized population. The population initialization algorithm based on Latin hypercube sampling is shown in Algorithm 1.

[0104]

[0105]

[0106] (2) Population iterative evolution algorithm based on the firefly perturbation strategy.

[0107] In the population iterative evolution operation of the traditional NSGA-II algorithm, the mutation operation is relatively random. Usually, the polynomial mutation method is used to mutate individuals, which easily mutates the already better solutions into sub-optimal solutions, reducing the convergence speed of the algorithm. The present invention proposes a population iterative evolution algorithm based on the firefly perturbation strategy to solve this problem. Firefly perturbation simulates the flashing behavior of fireflies in nature. Individuals with better performance have higher luminous brightness, which can attract other individuals to move towards them to optimize the individual performance. In this algorithm, the solution with the largest crowding degree in the population is selected as the brightest firefly individual, the attraction degree of the brightest individual to other individuals is calculated, and the positions of other individuals are moved to form a new population after firefly perturbation. The population iterative evolution algorithm based on the firefly perturbation strategy is shown in Algorithm 2.

[0108]

[0109]

[0110] Step 6: Determine the final collision avoidance decision according to the ship domain knowledge and the non-dominated solution set.

[0111] After iterative optimization and solution by the improved NSGA-II algorithm, the final non-dominated solution set will be obtained. However, some solutions in the non-dominated solution set obtained by the multi-objective optimization algorithm will be biased towards a certain objective function, and it is necessary to find a suitable optimal solution according to the actual problem to be solved. After step 5, there are some solutions in the non-dominated solution set with too small avoidance angles, that is, the problem of overemphasizing navigation economy and ignoring navigation safety. During the ship collision avoidance process, it is necessary to select a smaller avoidance angle to improve navigation economy on the premise of ensuring navigation safety. At this time, it is necessary to combine ship domain knowledge to determine the final collision avoidance decision to ensure the safety and economy of ship navigation.

[0112] In maritime collision avoidance, a ship is generally regarded as a particle. Taking the position of each ship as the center and the ship's safety distance as the radius, the ship domain can be determined. The conical area formed by the tangents of each ship to the ship domains of other ships in the environment is called the ship domain range. The ship domain knowledge stipulates that the relative velocity direction of the ship shall not fall within the ship domain range. If the relative velocity direction of a ship relative to another ship in the environment is within the ship domain range, this navigation angle cannot ensure navigation safety and a steering avoidance is required. Therefore, when determining the final collision avoidance decision, first sort the non-dominated solution set according to the avoidance angle, and it is also necessary to filter out unsafe solutions according to whether the relative velocity direction of the ship falls within the ship domain range. Finally, select the solution with the smallest avoidance angle on the premise of ensuring navigation safety as the final collision avoidance decision.

[0113] Step 7: The ship performs a steering operation.

[0114] After each ship receives the final collision avoidance decision from the shore-based control center, it performs a steering operation respectively.

[0115] Step 8: Determine whether the ship has reached the destination.

[0116] If the ship has not reached the destination, continue to monitor the navigation status in real time. If the ship has reached the destination, the process ends.

[0117] As can be seen from the above steps, in this embodiment, during the collision avoidance process, the ship collision avoidance action decision is uniformly made by the shore-based control center. The shore-based control center calculates the global collision risk based on the navigation parameters of the ships in the sea area. After reaching the threshold, it calculates the multi-ship cooperative collision avoidance decision and issues it to each ship. There is no need for communication between ships, and they only need to navigate according to the instructions of the shore-based control center. This method can solve the problem of multi-ship cooperative collision avoidance in a mixed traffic environment of manned and unmanned ships, and improves the navigation economy while ensuring the safety of ship navigation.

[0118] Next, through specific application examples, the solutions and effects of the present invention will be further described.

[0119] The present invention uses MATLAB to conduct simulation experiments to verify the effectiveness of the method of the present invention. First, a multi-ship encounter scenario that conforms to the actual sea navigation situation is designed. The initial parameters of the ships are set as shown in Table 1, and the initial ship positions are as Figure 2 shown.

[0120] Table 1 Initial parameters of the ships

[0121]

[0122] Figure 2 In the encounter scenario shown, there are 4 ships, namely Ship-A, Ship-B, Ship-C, and Ship-D. Ship-A sails from south to north, Ship-B sails from northeast to southwest, Ship-C sails from southeast to northwest, and Ship-D sails from northwest to southeast.

[0123] When the global risk reaches the threshold, based on the improved NSGA-II algorithm, calculate the multi-ship cooperative collision avoidance decision, and adopt the global cooperative collision avoidance decision of turning Ship-A 13 degrees to the right, Ship-D turning 21 degrees to the right, and Ship-B and Ship-C sailing straight. After the collision avoidance ends, Ship-A and Ship-D turn to the left to return to the original route, and the four ships sail towards their respective destinations. As Figure 3 shown, after sailing to the destination, the entire collision avoidance process ends.

[0124] The comparison diagram of the Pareto fronts of the improved NSGA-II algorithm and the traditional NSGA-II algorithm in the current scenario is as Figure 4As shown in the figure. The horizontal and vertical axes are the fitness function values of the two optimization objectives. Since this method sets the NGSA-II algorithm to optimize in the direction of minimization, the solutions closer to the lower left corner perform better in the two optimization objectives. Figure 4 The orange points in Figure 4 are the Pareto fronts calculated by the improved NSGA-II algorithm. It can be seen that the individual performance in the Pareto front after algorithm improvement is more excellent than that before improvement, with a smaller avoidance angle and a larger ship encounter distance, which improves the navigation economy on the premise of ensuring safety.

[0125] In addition, to verify the effectiveness of the global cooperative collision avoidance method based on the improved NSGA-II algorithm proposed in the present invention, the improved NSGA-II algorithm is compared with the traditional NSGA-II algorithm in three typical scenarios. Among them, Scenario 1 is a four-ship encounter scenario with a combination of port-starboard crossing encounters, Scenario 2 is a four-ship encounter scenario with a combination of overtaking and crossing encounters, and Scenario 3 simulates a four-ship encounter scenario with ships in the anchored state. The improved and unimproved NSGA-II algorithms are respectively used to conduct multi-ship collision avoidance experiments and compare the global ship avoidance angles. The collision avoidance effects and global avoidance angles are compared as Figure 5 shown.

[0126] In Figure 5 , Figures 5(a)-5(c) are schematic diagrams of the collision avoidance effects of the improved NSGA-II algorithm in three typical scenarios. In Scenario 1, the global cooperative collision avoidance decision is calculated as Ship-A turns right by 31 degrees, Ship-B turns right by 3 degrees, Ship-C turns right by 11 degrees, and Ship-D sails straight. In Scenario 2, the global cooperative collision avoidance decision is calculated as Ship-A turns right by 22 degrees, Ship-B and Ship-C sail straight, and Ship-D turns right by 17 degrees. In Scenario 3, the global cooperative collision avoidance decision is calculated as Ship-A turns right by 27 degrees, Ship-B turns right by 12 degrees, Ship-C turns right by 9 degrees, and Ship-D sails straight. Figure 5(d) is a comparison chart of the global avoidance angles of the improved NSGA-II algorithm and the traditional NSGA-II algorithm in three typical scenarios. From Figure 5 (d), it can be seen that the collision avoidance decisions calculated by the improved NSGA-II algorithm in the three typical scenarios can all achieve safe collision avoidance, and the avoidance angles of each ship are smaller than those of the traditional NSGA-II algorithm. This makes the turning amplitude of each ship smaller when performing collision avoidance operations as a whole, and the deviation distance caused by collision avoidance is shorter. Therefore, this method improves the navigation economy of ships while ensuring safe collision avoidance. The global cooperative collision avoidance decision method based on the improved NSGA-II algorithm proposed in the present invention is effective.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A global cooperative collision avoidance method in a mixed traffic environment of manned and unmanned ships, characterized in that, Including: Obtain the navigation information of all ships within the global scope; Calculate the ship motion parameters according to the navigation information of the ships, where the ship motion parameters include the distance between ships, the minimum encounter distance, and the shortest encounter time; Adopt the five-factor weighting method to calculate the collision risk degree between ships according to the distance between ships, the minimum encounter distance, the shortest encounter time, the relative bearing, and the ship speed ratio; Judge whether the global collision risk degree is greater than the threshold. If it is greater than the threshold, use the improved NSGA-II algorithm to calculate the global collision avoidance decision. If it is less than the threshold, no collision avoidance operation is required, and the ship sails straight normally. The global collision risk degree is obtained according to the following formula: ， Among them, represents the global collision risk, represents the number of ships, represents the collision risk between two ships, The improved NSGA-II algorithm improves the traditional NSGA-II algorithm from two aspects: population initialization and population iterative evolution: The population initialization algorithm based on Latin hypercube sampling performs stratification and sampling operations according to the number of samples within the optional steering angle range of each ship. After merging the sampling results of different ships, a uniformly distributed initial population is formed. Select the solution with the largest crowding degree in the population as the brightest firefly individual, calculate the attraction degree of the brightest individual to other individuals, and move the positions of other individuals to form a new population after firefly perturbation.

2. The global cooperative collision avoidance method in a mixed traffic environment of manned and unmanned ships according to claim 1, characterized in that Obtain the navigation information of all ships within the global scope, including obtaining the navigation information of all ships within the global scope through environmental perception devices, where the navigation information includes ship position, heading, and speed.

3. The global collaborative collision avoidance method in a mixed traffic environment of manned and unmanned vessels according to claim 2, wherein, The minimum encounter distance is calculated according to the following formula: Among them, DCPA represents the closest distance of approach, represents the relative velocity direction of ship A with respect to ship B, represents the true bearing of ship B relative to ship A, represents the distance between ship A and ship B; The shortest encounter time is calculated according to the following formula: Among them, TCPA represents the shortest time to closest point of approach, represents the relative speed of ship A with respect to ship B, represents the direction of the relative speed of ship A with respect to ship B, represents the true bearing of ship B relative to ship A, represents the distance between ship A and ship B.

4. A global collaborative collision avoidance method in a mixed traffic environment of manned and unmanned ships, characterized in that, Adopt the five-factor weighting method to calculate the collision risk degree between ships according to the distance between ships, the minimum encounter distance, the shortest encounter time, the relative bearing, and the ship speed ratio, including: Solve the risk membership function of the minimum encounter distance according to the following method: Among them, represents the membership function of the risk of the minimum distance of closest approach, represents the dangerous distance of closest approach, represents the safe distance of closest approach; Solve the risk membership function of the shortest encounter time according to the following method: Among them, represents the membership function of the shortest encounter time for danger, represents the dangerous encounter distance, represents the magnitude of the relative speed; Solve the risk membership function of the distance between two ships according to the following method: Among them, represents the risk function of the distance between two ships, represents the dangerous encounter distance, represents the safe encounter distance; Solve the risk membership function of the oncoming ship bearing according to the following method: Among them, The dangerous membership function indicating the bearing of the approaching ship, Indicates the bearing of the approaching ship; Solve the risk membership function of the ship speed ratio according to the following method: Among them, represents the dangerous membership function of the ship speed ratio, represents the ship speed ratio, represents the course of the ship itself.

5. A global cooperative collision avoidance method in a mixed traffic environment of manned and unmanned ships, characterized in that, Using the improved NSGA-II algorithm to calculate the global collision avoidance decision also includes: After iterative optimization and solution through the improved NSGA-II algorithm, the final non-dominated solution set will be obtained; Screen the global collision avoidance decision in the non-dominated solution set according to the avoidance angle and the relative speed direction, so as to obtain the final global collision avoidance strategy.

Images