An AGV obstacle avoidance path planning method in unmanned warehousing system

Abstract

The present invention discloses a method for AGV obstacle avoidance path planning in an unmanned warehousing system. The method replaces the population generation method of the traditional differential evolution algorithm by generating a chaotic initialization population using nested logistic mapping and skew tent mapping, and performs a Levy flight operation after the selection operation to perform a global path search on the environmental map, thereby obtaining a global feasible path for the AGV that avoids obstacles in the environmental map. When an unknown obstacle exists on the feasible path of the AGV, a dynamic window method is used to re-plan the local path so that the AGV can achieve the obstacle avoidance process within the interval where the unknown obstacle is located. The improved differential evolution algorithm enhances the global search capability through chaotic initialization and Levy flight strategy, while the dynamic window rule enables the AGV to quickly respond to unknown obstacles. The adaptability and accuracy of path planning are improved, while the obstacle avoidance capability and safety of the AGV in complex environments are enhanced.

Description

Technical Field

[0001] This invention relates to the field of warehousing and logistics technology, and more specifically to an obstacle avoidance path planning method for AGVs in an unmanned warehousing system. Background Technology

[0002] Unmanned warehousing systems improve logistics efficiency and accuracy through automated equipment. Among them, Automated Guided Vehicles (AGVs), also known as automated guided vehicles or automated guided transport vehicles, are key automated handling tools. Their path planning and obstacle avoidance capabilities directly affect the operational efficiency and safety of the warehousing system.

[0003] However, in practical applications, the warehousing environment is not static. The complexity of the warehousing environment affects the efficiency and accuracy of AGV path planning. Traditional AGV path planning methods often react slowly when facing complex warehousing environments and lack effective global search capabilities, resulting in unsatisfactory path planning efficiency and obstacle avoidance. Therefore, in order to enable AGV path planning to meet various warehousing environments and ensure that AGVs can still operate accurately and efficiently in complex warehousing environments, it is necessary to improve the relevant algorithms in the AGV operation process to ensure the efficiency and safety of AGVs in transportation work. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide an AGV obstacle avoidance path planning method in an unmanned warehouse system, in order to solve the problems that existing AGV path planning methods are slow to respond when facing complex warehouse environments and lack effective global search capabilities, resulting in unsatisfactory path planning efficiency and obstacle avoidance effects.

[0005] To achieve the above objectives, this invention provides a method for obstacle avoidance path planning in an unmanned warehousing system. Specifically, the method includes the following steps:

[0006] An obstacle avoidance path planning method for AGVs in an unmanned warehouse system, characterized by the following steps:

[0007] 1) Construct an environmental map of the warehouse where the AGV is located;

[0008] 2) An improved differential evolution algorithm is used to perform a global path search on the environment map to obtain a globally feasible path for the AGV that avoids obstacles in the environment map, so as to control the AGV to run along this globally feasible path.

[0009] The improved differential evolution algorithm is as follows: a chaotic initial population is generated by using nested Logistic mapping and SkewTent mapping, and multiple iterations of mutation, crossover, selection and Lévy flight operations are performed on the chaotic initial population until the termination condition is met and the result is output.

[0010] 3) When the AGV moves along the global feasible path, it is detected in real time whether there are unknown obstacles on the feasible path. If there are unknown obstacles, the dynamic window method is used to re-plan the local path in the interval where the unknown obstacle is located in the global feasible path, so as to obtain the local path planning result, so that the AGV runs along the local path planning result in the interval where the unknown obstacle is located.

[0011] The method of the present invention has the following advantages: The difference between the improved differential evolution algorithm of the present invention and the existing traditional differential evolution algorithm is that the present invention generates a chaotic initial population based on nested Logistic mapping and Skew Tent mapping, while the traditional one generates a randomly generated initial population. In addition, compared with the traditional differential evolution algorithm, the present invention also performs a Lévy fly operation after the selection operation.

[0012] First, the Logistic mapping is widely used due to its simple mathematical form and controllable chaotic properties. When initializing a population, it can generate a sequence with randomness and ergodicity, which helps the algorithm to effectively explore a wide search space and avoid getting trapped in local optima. The Skew Tent mapping, as a chaotic mapping, has a fast iteration speed and good statistical performance. By generating chaotic sequences through piecewise linear functions, it can provide more diversity to the population during initialization. At the same time, the chaotic properties of the Skew Tent mapping help the algorithm avoid premature convergence and maintain population diversity. Therefore, generating a chaotic initial population through the above Logistic mapping and Skew Tent mapping fully utilizes the advantages of the two mappings, making the generated population have better randomness and uniformity, thereby improving the algorithm's global search capability and local search accuracy, and helping the algorithm find the global optimum in complex optimization problems.

[0013] Secondly, after the selection operation, the present invention applies the Levy flight strategy to the selected individuals, which enhances the global search capability and local search capability of the algorithm by simulating the foraging behavior of animals and combining long-distance and short-distance random movements.

[0014] Furthermore, in the method of this invention, after obtaining the global path planning based on the improved differential evolution algorithm, the AGV is also monitored in real time during operation to detect whether there are unknown obstacles on the global path planning route. These unknown obstacles are obstacles that do not appear in the constructed environment map, such as moving operators, other running AGVs, or other objects on the feasible path that may affect the passage of the AGV. These unknown obstacles will affect the normal operation of the AGV. Therefore, this invention performs real-time detection on the running AGV, and when it detects that unknown obstacles are affecting the operation of the AGV, it uses the dynamic window method to adjust the local path to ensure the safe operation of the AGV.

[0015] In summary, the method of the present invention effectively avoids local optima, improves the adaptability and accuracy of path planning, and enhances the obstacle avoidance capability and safety of AGVs in complex environments.

[0016] Further, in step 2), the steps of generating a chaotic initialization population using nested Logistic mapping and SkewTent mapping include: setting the mapping parameters within a certain interval, selecting initial values ​​to perform iterative calculations of Logistic mapping and SkewTent mapping to obtain multiple values, arranging the multiple values ​​according to the set rules to form a new sequence, and using the new sequence as the chaotic initialization population after the new sequence satisfies the chaotic characteristics.

[0017] Furthermore, in the step of generating the chaotic initial population, if the new sequence does not satisfy the chaotic properties, the initial value is changed and the Logistic mapping and Skew Tent mapping are iteratively calculated again until the new sequence satisfies the chaotic properties.

[0018] Furthermore, the mapping formula for the Logistic mapping is:

[0019] X k+1 =μx k (1-x k );

[0020] Among them, x k The population proportion is represented at iteration step k; μ is a system parameter.

[0021] Furthermore, the mapping formula for the Skew Tent mapping is:

[0022]

[0023] Where 'a' is the parameter of the mapping.

[0024] Its beneficial effects are as follows: Within a certain interval, the parameters of the mapping are set to ensure its chaotic characteristics. An initial value is selected for the chaotic mapping, and iterative calculations are performed using the aforementioned Logistic mapping formula and Skew Tent mapping formula to generate a series of values ​​as part of the population. The generated values ​​are arranged according to a set rule to form a new sequence, fully utilizing the advantages of both mappings to make the generated population more random and uniform. The merged sequence is checked to see if it satisfies the chaotic characteristics. If the check fails, a new initial value needs to be selected. If the check passes, this chaotic sequence is used as the initial population for improving the algorithm, increasing the population's diversity and distribution uniformity.

[0025] Further, in step 3), the dynamic window method is as follows: obtain the current status information of the AGV and the operating speed range of the AGV, simulate the trajectory of the AGV at each operating speed within the operating speed range, predict the AGV operating speed range trajectory under multiple current status information, and select a path with no collision point with the unknown obstacle as the local path planning result according to the selection conditions.

[0026] Furthermore, in step 3), the AGV's operation is controlled according to the AGV's running speed corresponding to the selected path that has no collision point with the unknown obstacle, so that the AGV can run along the local path planning result within the area where the unknown obstacle is located.

[0027] Furthermore, the selection criteria are as follows: the AGV running speed range trajectory under each current state information is comprehensively evaluated based on running safety and running efficiency, and the selection is made according to the comprehensive evaluation results.

[0028] The beneficial effects are as follows: The dynamic window method in this invention obtains the current status of the AGV and its operating speed range. The operating speed range refers to the range of speed states that the AGV can achieve during actual operation. In the current state, the local operating path obtained by the AGV at different operating speeds within this range is different. Therefore, this invention simulates the local operating paths of the AGV at various operating speeds under the current situation. Based on the simulated local operating paths, a suitable local operating path is selected according to selection criteria. Then, the AGV is controlled using the operating speed corresponding to this suitable local operating path, enabling the AGV to run along this suitable local operating path. Furthermore, the selection of local operating paths at various operating speeds in this invention simultaneously considers operational safety (distance from obstacles) and operational efficiency (degree of approach to the target), that is, it simultaneously considers the AGV's own operational safety and operational efficiency to select the final local path planning result. Therefore, this local path planning result ensures the efficient and safe operation of the AGV on this local operating path.

[0029] Furthermore, in step 2), the Levy flight formula is:

[0030] x′ (t) = x(t) + s·levy(β);

[0031] Where s is the step size parameter; levy(β) represents Levi's flight.

[0032] Furthermore, in step 1), a grid map is used to construct an environmental map of the warehouse where the AGV is located.

[0033] Its beneficial effects are as follows: In this invention, the operation of the AGV is guaranteed by real-time detection of the AGV's operation process after global path planning and dynamic obstacle avoidance. Therefore, based on this dynamic obstacle avoidance process, when generating the warehouse environment map, there is no need to consider the influence of dynamic or unknown factors. Thus, by using a grid map to construct the environment map of the warehouse where the AGV is located, this high-precision and high-efficiency method of constructing the environment map efficiently provides accurate environmental information for path planning.

[0034] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0035] Figure 1 This is a schematic diagram illustrating the implementation process of the obstacle avoidance path planning method for AGVs in an unmanned warehouse system according to an embodiment of the present invention.

[0036] Figure 2 This is a schematic diagram illustrating the specific process of the improved differential evolution algorithm according to an embodiment of the present invention.

[0037] Figure 3 This is a schematic diagram illustrating the specific process of chaos initialization in an embodiment of the present invention.

[0038] Figure 4 This is a schematic diagram illustrating the specific process of the Levi flight strategy in an embodiment of the present invention.

[0039] Figure 5 This is a schematic diagram illustrating the specific process of the dynamic window method in an embodiment of the present invention. Detailed Implementation

[0040] The technical solution of the present invention will be clearly and completely described below with reference to specific embodiments. However, those skilled in the art should understand that the embodiments described below are only for illustrating the present invention and should not be regarded as limiting the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Example of AGV obstacle avoidance path planning method in unmanned warehousing system

[0042] The traditional differential evolution (DE) algorithm process includes:

[0043] (1) Determine the control parameters of the differential evolution algorithm and the fitness function. The control parameters of the differential evolution algorithm include: population size NP, scaling factor F and crossover probability CR.

[0044] (2) Randomly generate the initial population.

[0045] (3) Evaluate the initial population, i.e. calculate the fitness value of each individual in the initial population.

[0046] (4) Determine whether the termination condition has been met or the number of generations has reached its maximum. If yes, terminate the evolution and output the best individual as the optimal solution; otherwise, continue.

[0047] (5) Perform mutation and crossover operations to obtain an intermediate population.

[0048] (6) Select individuals from the original population and the intermediate population to obtain a new generation of population.

[0049] (7) Evolutionary generation g = g + 1, go to step (4).

[0050] The method of this invention replaces the random generation of the initial population in the traditional differential evolution algorithm with the generation of a chaotic initial population using nested Logistic and Skew Tent mappings. Furthermore, a Lévy flight operation is performed after the selection step in the traditional differential evolution algorithm. The Logistic mapping effectively generates sequences with randomness and ergodicity, helping the algorithm to explore effectively in a broad search space and avoid getting trapped in local optima. The Skew Tent mapping, as a chaotic mapping, has fast iteration speed and good statistical performance. Generating chaotic sequences through piecewise linear functions provides greater diversity to the population during initialization. Simultaneously, the chaotic characteristics of the Skew Tent mapping help the algorithm avoid premature convergence and maintain population diversity. After the selection operation, the Lévy flight strategy is applied to the selected individuals. By simulating animal foraging behavior and combining long-distance and short-distance random movements, the algorithm's local search capability is enhanced, thereby improving its global search capability, avoiding local optima, and improving the adaptability and accuracy of path planning.

[0051] Furthermore, the method of this invention performs real-time detection on the AGV while it is running along the global path planning result. For the intervals where unknown obstacles affecting the AGV's operation are located, a Dynamic Window Approach (DWA) is used for local path planning to ensure the AGV's safe operation within these intervals, thereby guaranteeing the AGV's overall operational safety. Therefore, the improved differential evolution algorithm enhances the global search capability through chaotic initialization and the Lévy flight strategy, while the dynamic window approach enables the AGV to respond quickly to unknown obstacles. This method effectively avoids local optima, improves the adaptability and accuracy of path planning, and enhances the AGV's obstacle avoidance capability and safety in complex environments. The obstacle avoidance path planning method of this invention can meet the needs of various complex warehousing environments.

[0052] like Figure 1 As shown, the obstacle avoidance path planning method for AGVs in the unmanned warehouse system in this embodiment specifically includes the following steps:

[0053] 1) Construct an environment map.

[0054] In this embodiment, a high-precision warehouse environment map is constructed using a grid map. The specific steps include: constructing a map of the warehouse environment using a grid map method; dividing the entire warehouse environment into several grids of equal size based on the warehouse size, shelf layout, and passable area information; each grid represents a passable or inaccessible unit.

[0055] In this embodiment, obstacle information from the real environment is projected onto a grid map. The size of the obstacle is determined as the grid size. The obstacles projected onto the grid map are expanded to ensure that the AGV can avoid obstacles when planning its path. A q×q grid map is constructed, where white grids are assigned a value of "0" to represent passable areas, and black grids are assigned a value of "1" to represent obstacle areas (i.e., in this embodiment, grids corresponding to passable areas are assigned a value of "0", and impassable areas are treated as obstacle areas, with grids corresponding to obstacle areas assigned a value of "1"). The position of each grid on the map is represented by coordinates (x, y) in a Cartesian coordinate system, with each grid corresponding to one coordinate.

[0056] 2) An improved differential evolution algorithm is used to perform a global path search on the environmental grid map to obtain the globally feasible path for the AGV.

[0057] like Figure 2 As shown, the improved differential evolution algorithm used in this embodiment includes the following steps:

[0058] 2.1) A chaotic initialization strategy is constructed using nested Logistic mappings and Skew Tent mappings to obtain a chaotic initialization population. The chaotic initialization strategy process is as follows: Figure 3 As shown.

[0059] The Logistic chaotic map is widely used due to its simple mathematical form and controllable chaotic properties. When initializing a population, it can generate sequences with randomness and ergodicity, which helps the algorithm to effectively explore a large search space and avoid getting trapped in local optima. Its definition is as follows:

[0060] X k+1 =μx k (1-x k );

[0061] Among them, x k The population proportion at iteration step k is usually within the interval (0,1); μ is a system parameter that controls the dynamic behavior of the system and usually takes values ​​between [0,4].

[0062] The SkewTent mapping, as a chaotic mapping, possesses fast iteration speed and good statistical performance. It generates chaotic sequences through piecewise linear functions, providing greater diversity to the population during initialization. Furthermore, the chaotic properties of the SkewTent mapping help the algorithm avoid premature convergence and maintain population diversity. Its definition is as follows:

[0063]

[0064] Where a is the mapping parameter, and when a = 0.5, it is a Tent mapping.

[0065] Based on the characteristics of the Logistic chaotic mapping and the Skew Tent mapping described above, this embodiment first sets the mapping parameters within a certain interval to ensure its chaotic characteristics. Simultaneously, an initial value is selected as the starting point for the chaotic mapping iteration. Iterative calculations are then performed using the aforementioned Logistic mapping formula and Skew Tent mapping formula to generate a series of values ​​as part of the population. The generated values ​​are arranged according to a set rule to form a new sequence. The rule used in this embodiment is to alternately place the results of the Logistic mapping and the Skew Tent mapping of the initial value to form a new sequence, fully utilizing the advantages of both mapping methods to make the generated population more random and uniform. Other implementation methods could also include randomly arranging the results of the Logistic mapping and the Skew Tent mapping of the initial value to form a new sequence.

[0066] The merged sequence is checked to see if it satisfies the chaotic property. If the check fails, the initial value needs to be selected again. If the check passes, the chaotic sequence is used as the initial population for the improved algorithm to increase the diversity and distribution uniformity of the population.

[0067] 2.2) Perform mutation, crossover, and selection operations on the chaotic initial population.

[0068] Three distinct individuals are selected from the population, their difference vectors are calculated, and a mutation factor is introduced. These difference vectors are then used to generate new candidate solutions. Specifically, the mutation strategy of the differential evolution algorithm is used, combined with a population generated through chaotic initialization, to perform the mutation operation. The mutation operation formula is:

[0069] v i =x r1 +F(x r2 -x r3 )

[0070] Among them, x r1 ,x r2 ,x r3 These are three different individuals randomly selected from the population, where F is the mutation factor.

[0071] Based on the crossover probability, it is determined whether to replace certain components of the mutated vector with corresponding components of the target vector, thereby generating new experimental individuals for the next selection operation. Specifically, the individuals generated by the mutation operation are crossed with the current individuals to generate experimental individuals. Through a certain probability selection, combined with the characteristics of the parent and mutated individuals, new individuals are generated. The crossover operation formula is:

[0072]

[0073] Here, rand(0,1) is a random number between 0 and 1; CR is the crossover probability.

[0074] The fitness of parent and experimental individuals is compared, and individuals with better fitness are selected for the next generation. Specifically, through a selection operation, the fitness of experimental and current individuals is compared, and individuals with better fitness are selected for the next generation. Individuals with high fitness are more likely to be selected for crossover and mutation, thus preserving their characteristics in offspring. The selection operation formula is as follows:

[0075]

[0076] Here, function f is the fitness function.

[0077] 2.3) Perform Levi flight operations.

[0078] like Figure 4 As shown, a Lévy distribution step size is generated to simulate bird flight behavior. Based on the current best solution, long-distance random jumps are performed to explore new potential solutions, enhancing the algorithm's global search capability and avoiding getting trapped in local optima. Specifically, after the selection operation, the Lévy flight strategy is applied to the selected individual. By simulating animal foraging behavior and combining long-distance and short-distance random movements, the algorithm's local search capability is enhanced. The Lévy flight formula is:

[0079] x′ (t) = x(t) + s·levy(β);

[0080] Where s is the step size parameter; levy(β) represents Levi's flight.

[0081] 2.4) Repeat the above mutation, crossover, selection and Levy flight operations to update the population until the maximum number of iterations or other termination conditions are met (i.e. until the termination conditions are met), while checking whether the fitness has converged to the preset threshold; select and output the individual with the highest fitness from the final population as the optimal solution, output the result of the algorithm iteration, and select the optimal path as the global path planning result of the AGV accordingly.

[0082] In this invention, the existing algorithm is improved by using chaotic initialization and Lévy flight operation, which enhances the algorithm's global search capability, avoids local optima, and improves the adaptability and accuracy of path planning.

[0083] 3) Real-time detection of unknown obstacles along the global feasible path of the AGV. If an unknown obstacle is detected, the dynamic window method is used to avoid the obstacle locally before continuing to run along the next global feasible path.

[0084] like Figure 5 As shown, the specific steps of the Dynamic Window (DWA) method for local path planning and obstacle avoidance when an AGV detects an unknown obstacle include: When the AGV's sensors detect an unknown obstacle, the DWA algorithm is activated for obstacle avoidance. Real-time acquisition of the AGV's position, speed, acceleration, and other state information provides an accurate starting point for path planning. Simultaneously, based on the AGV's dynamic characteristics, such as maximum speed and maximum acceleration, the range of the speed sampling space (i.e., the range of speed states that the AGV can achieve during actual operation, including maximum speed, minimum speed, and acceleration) is determined, including sampling of linear velocity and angular velocity. For each sampled speed, considering the AGV's dynamic constraints, the possible trajectory of the AGV within a certain time period at that speed is predicted. Each simulated trajectory is evaluated based on safety (distance from the obstacle) and efficiency (degree of approach to the target). The trajectory with the lowest cost is selected from all simulated trajectories as the optimal driving path for the AGV. Based on the selected optimal trajectory, speed and steering control commands for the AGV are generated and sent to the AGV to guide its movement and achieve obstacle avoidance. The AGV checks whether it has successfully bypassed the obstacle using sensor feedback. If it has, it continues with the original path planning. This process is repeated continuously during the AGV's operation to cope with dynamically changing environments and potential new obstacles.

[0085] In this invention, a dynamic window method is introduced, which enables AGVs to respond quickly to unknown obstacles in local path planning, enhances obstacle avoidance capabilities, and improves the efficiency and safety of AGV path planning in complex dynamic environments.

[0086] The method of this invention utilizes an improved differential evolution algorithm for global path planning and a dynamic window method for local path planning to achieve real-time obstacle avoidance. The improved differential evolution algorithm enhances global search capabilities through chaotic initialization and the Lévy flight strategy, while the dynamic window method enables the AGV to respond quickly to unknown obstacles. This method effectively avoids local optima, improves the adaptability and accuracy of path planning, and enhances the obstacle avoidance capability and safety of the AGV in complex environments.

[0087] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for obstacle avoidance path planning in an unmanned warehousing system, characterized in that, The steps include: 1) Construct an environmental map of the warehouse where the AGV is located; 2) An improved differential evolution algorithm is used to perform a global path search on the environment map to obtain a globally feasible path for the AGV that avoids obstacles in the environment map, so as to control the AGV to run along this globally feasible path. The improved differential evolution algorithm is as follows: a chaotic initial population is generated by using nested Logistic mapping and Skew Tent mapping, and multiple iterations of mutation, crossover, selection and Lévy flight operations are performed on the chaotic initial population until the termination condition is met and the result is output. The steps for generating a chaotic initialization population using nested Logistic mapping and Skew Tent mapping include: setting the mapping parameters within a certain interval, selecting initial values ​​for iterative calculation of Logistic mapping and Skew Tent mapping to obtain multiple values, arranging the multiple values ​​according to the set rules to form a new sequence, and using the new sequence as the chaotic initialization population after the new sequence satisfies the chaotic characteristics. In the step of generating the chaotic initial population, if the new sequence does not satisfy the chaotic properties, the initial value is changed and the Logistic mapping and SkewTent mapping are iterated again until the new sequence satisfies the chaotic properties. 3) When the AGV moves along the global feasible path, it is detected in real time whether there are unknown obstacles on the feasible path. If there are unknown obstacles, the dynamic window method is used to re-plan the local path in the interval where the unknown obstacle is located in the global feasible path, so as to obtain the local path planning result, so that the AGV runs along the local path planning result in the interval where the unknown obstacle is located.

2. The AGV obstacle avoidance path planning method in the unmanned warehousing system according to claim 1, characterized in that, The mapping formula for the Logistic mapping is: X k+1 =μx k (1-x k ); Among them, x k The population proportion is represented at iteration step k; μ is a system parameter.

3. The AGV obstacle avoidance path planning method in the unmanned warehousing system according to claim 1, characterized in that, The mapping formula for SkewTent mapping is: Where a is the parameter of the mapping, x is the current iteration value, and f a (x) is the next iteration value.

4. The AGV obstacle avoidance path planning method in the unmanned warehousing system according to claim 1, characterized in that, In step 3), the dynamic window method is as follows: obtain the current status information of the AGV and the operating speed range of the AGV, simulate the trajectory of the AGV at each operating speed within the operating speed range, predict the AGV operating speed range trajectory under multiple current status information, and select a path with no collision point with the unknown obstacle as the local path planning result according to the selection conditions.

5. The AGV obstacle avoidance path planning method in the unmanned warehousing system according to claim 4, characterized in that, In step 3), the AGV's operation is controlled according to the selected AGV running speed corresponding to the path with no collision point with the unknown obstacle, so that the AGV can run along the local path planning result in the area where the unknown obstacle is located.

6. The AGV obstacle avoidance path planning method in the unmanned warehousing system according to claim 4, characterized in that, The selection criteria are as follows: the AGV running speed range trajectory under each current state information is comprehensively evaluated based on running safety and running efficiency, and the selection is based on the comprehensive evaluation results.

7. The AGV obstacle avoidance path planning method in the unmanned warehousing system according to claim 1, characterized in that, In step 2), the Levy flight formula is: x′ (t) =x(t)+s·levy(β); Where s is the step size parameter; levy(β) represents the Levy flight, x(t) is the current position of the individual at the t-th iteration, and x′ (t) It is the path solution updated by Lévy flight, and β is the dimension of the objective function.

8. The AGV obstacle avoidance path planning method in the unmanned warehousing system according to claim 1, characterized in that, In step 1), a grid map is used to construct an environmental map of the warehouse where the AGV is located.

Images