Multi-robot cooperative path planning method and system and storage medium

Abstract

The invention provides a multi-robot cooperative path planning method, a multi-robot cooperative path planning system and a computer readable storage medium. The multi-robot cooperative path planning method comprises the following steps: S1, acquiring laser radar data of each robot to form environment observation data; s2, according to the environment observation data and the target point position of each robot, first action values of multiple actions in the action space of each robot are determined through a DQN algorithm and a full-connection network, and second action values of the multiple actions are determined through an artificial potential field method; s3, according to the first action value and the second action value, determining a complete action value of the plurality of actions through a linear addition mode; s4, determining and executing an action strategy with the highest action value according to the complete action value of the plurality of actions; and S5, in response to the situation that all the robots do not arrive at the target point position, repeating the steps S1 to S4 until all the robots arrive at the target point position.

Description

Technical Field

[0001] The present invention relates to the field of path planning, and in particular to a multi-robot collaborative path planning method, a multi-robot collaborative path planning system, and a computer-readable storage medium. Background Art

[0002] In recent years, with the rapid development of artificial intelligence, mobile robots with embedded sensing and computing capabilities have been widely used in many fields such as transportation, warehousing and daily life. Multi-robot collaborative path planning is a hot research topic, which can independently plan safe paths for multiple mobile robots in a shared workspace so that multiple mobile robots do not collide when moving.

[0003] Traditional multi-robot collaborative path planning methods treat multiple robots as a single high-dimensional intelligent agent to obtain the optimal path, but as the number of robots increases, its computational complexity increases dramatically; at the same time, traditional methods require accurate environmental modeling and global environmental information to be implemented, so when the robot can only obtain partial observable environmental information, the algorithm success rate is low. In the existing technology, the reinforcement learning-based method plans the optimal path by establishing a Markov decision process. Therefore, the optimal path planning can be achieved based only on local environmental observations. However, this method still has problems such as slow convergence and strong randomness.

[0004] In order to overcome the above-mentioned defects of the prior art, the field is in urgent need of a multi-robot collaborative path planning technology that can effectively reduce the randomness of the algorithm, increase the model convergence speed, and thus improve the success rate of collaborative path planning. Summary of the invention

[0005] A brief summary of one or more aspects is given below to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all conceived aspects, and is neither intended to identify the key or critical elements of all aspects nor to define the scope of any or all aspects. Its only purpose is to give some concepts of one or more aspects in a simplified form as a prelude to a more detailed description that will be given later.

[0006] The present invention provides a multi-robot collaborative path planning method, a multi-robot collaborative path planning system, and a computer-readable storage medium, which can effectively reduce the randomness of the algorithm and increase the model convergence speed, thereby improving the success rate of collaborative path planning.

[0007] Specifically, the multi-robot collaborative path planning method provided in accordance with the first aspect of the present invention comprises the following steps: S1: acquiring the laser radar data of each robot to form environmental observation data; S2: determining the first action value of multiple actions in the action space of each robot through the DQN algorithm and the fully connected network according to the environmental observation data and the target point position of each robot, and determining the second action value of the multiple actions through the artificial potential field method; S3: determining the complete action value of the multiple actions through linear addition according to the first action value and the second action value; S4: determining and executing the action strategy with the highest action value according to the complete action value of the multiple actions; and S5: in response to the fact that not all of the robots have reached the target point position, repeating S1 to S4 until the robots reach the target point position.

[0008] Preferably, in an embodiment of the present invention, the laser radar data is represented as a one-dimensional feature vector to form the environmental observation data.

[0009] Preferably, in one embodiment of the present invention, S2 includes the steps of: splicing the one-dimensional feature vector of the environmental observation data and the robot target point position to determine a merged feature vector, using the DQN algorithm, and inputting the merged feature vector into the fully connected network to determine the first action value.

[0010] Preferably, in an embodiment of the present invention, the action space of each robot includes four executable action strategies: up, down, left and right.

[0011] Preferably, in one embodiment of the present invention, the step of determining the second action value of the multiple actions by the artificial potential field method in S2 includes: S21: determining the artificial potential field method gravitational vector based on the target point position and the current spatial position of each robot; S22: determining the spatial position of the obstacle based on the environmental observation data; S23: determining the artificial potential field method repulsive vector based on the current spatial position of each robot and the spatial position of the obstacle; and S24: performing vector decomposition on the artificial potential field method gravitational vector and the artificial potential field method repulsive vector to determine the second action value.

[0012] Preferably, in one embodiment of the present invention, the artificial potential field method gravity vector is calculated by the following formula:

[0013]

[0014] Among them, F att (X) is the gravitational vector of the artificial potential field method, K attis the first proportional coefficient, X is the current spatial position of each robot, X0 is the target point position, ρ(X-X0) is the distance between the current spatial position of each robot and the target point position, ρ att is the first distance threshold.

[0015] Preferably, in one embodiment of the present invention, the artificial potential field method repulsive force vector is calculated by the following formula:

[0016]

[0017] Among them, F rep,i (X,X obs,i ) is the repulsive force vector of the artificial potential field method, K rep is the second proportional coefficient, X is the current spatial position of each robot, and X obs,i is the spatial position of the i-th obstacle, ρ(X,X obs,i ) is the distance between the current spatial position of each robot and the spatial position of the i-th obstacle, ρ rep is the second distance threshold.

[0018] Preferably, in an embodiment of the present invention, S3 further includes the step of: setting a third proportional coefficient ω, and determining the complete action value q(s, a) according to the third proportional coefficient ω as:

[0019] q(s,a)=(1-ω)×q DQN (s,a)+ω×q APF (s,a),

[0020] Among them, q DQN (s,a) is the first action value, q APF (s,a) is the second action value.

[0021] In addition, the multi-robot collaborative path planning system provided according to the second aspect of the present invention includes a memory and a processor. The memory stores computer instructions. The processor is connected to the memory and is configured to execute the computer instructions stored in the memory to implement the multi-robot collaborative path planning method provided by any one of the above embodiments.

[0022] In addition, the computer-readable storage medium provided according to the third aspect of the present invention stores computer instructions. When the computer instructions are executed by the processor, the multi-robot collaborative path planning method provided by any one of the above embodiments is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above features and advantages of the present invention can be better understood after reading the detailed description of the embodiments of the present disclosure in conjunction with the following drawings. In the drawings, the components are not necessarily drawn to scale, and components with similar related properties or features may have the same or similar reference numerals.

[0024] Figure 1 A flowchart of a multi-robot collaborative path planning method provided according to some embodiments of the present invention is shown;

[0025] Figure 2 A schematic diagram of a multi-robot collaborative path planning system provided according to some embodiments of the present invention is shown;

[0026] Figure 3 A diagram showing a reinforcement learning process in a multi-robot collaborative path planning method provided according to some embodiments of the present invention; and

[0027] Figure 4 A diagram showing a calculation principle of a second action value provided according to some embodiments of the present invention is shown.

[0028] Reference numerals:

[0029] 100: Multi-robot collaborative path planning method;

[0030] S1~S5: steps;

[0031] 200: Multi-robot collaborative path planning system;

[0032] 210: Memory;

[0033] 211: Computer readable storage medium;

[0034] 220: Processor;

[0035] 301: Environmental observation data;

[0036] 302: Target point position;

[0037] 311: First Action Value;

[0038] 312: Second action value;

[0039] 313: Full action value;

[0040] 321: First layer of fully connected network;

[0041] 322: Second layer fully connected network;

[0042] 323: Layer 3 fully connected network;

[0043] 331: Artificial potential field method gravitational vector;

[0044] 332: Artificial potential field method repulsive force vector;

[0045] 340: Evaluation force module;

[0046] 350: value prediction module;

[0047] 360: Fully connected network;

[0048] 401: The current spatial position of the robot;

[0049] 402: Target point location; and

[0050] 403: The spatial position of the obstacle. DETAILED DESCRIPTION

[0051] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. Note that the aspects described below in conjunction with the accompanying drawings and specific embodiments are only exemplary and should not be construed as limiting the scope of protection of the present invention in any way.

[0052] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0053] In addition, the terms "upper", "lower", "left", "right", "top", "bottom", "horizontal" and "vertical" used in the following description should be understood as the directions shown in the paragraph and the related drawings. Such relative terms are only used for the convenience of description and do not mean that the device described therein must be manufactured or operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0054] It is understood that although the terms "first", "second", "third", etc. may be used herein to describe various components, regions, layers and / or parts, these components, regions, layers and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers and / or parts. Therefore, the first component, region, layer and / or part discussed below may be referred to as a second component, region, layer and / or part without departing from some embodiments of the present invention.

[0055] As described above, in the prior art, the reinforcement learning method plans the optimal path by establishing a Markov decision process. Thus, the optimal path can be planned based only on local environment observations. However, this method still has problems such as slow convergence and strong randomness.

[0056] In order to overcome the above defects in the prior art, the present invention provides a multi-robot collaborative path planning method, a multi-robot collaborative path planning system, and a computer-readable storage medium. The multi-robot collaborative path planning method provided by the present invention can effectively improve the success rate of collaborative path planning.

[0057] In some non-limiting embodiments, the multi-robot collaborative path planning method provided in the first aspect of the present invention can be implemented via the multi-robot collaborative path planning system provided in the second aspect of the present invention. Specifically, the multi-robot collaborative path planning system can be configured with a memory and a processor. The memory includes but is not limited to the computer-readable storage medium provided in the third aspect of the present invention, on which computer instructions are stored. The processor is connected to the memory and is configured to execute the computer instructions stored on the memory to implement the multi-robot collaborative path planning method provided in the first aspect of the present invention.

[0058] The following will first describe the working principle of the above-mentioned multi-robot collaborative path planning system in conjunction with some embodiments of the multi-robot collaborative path planning method. Those skilled in the art will understand that the embodiments of these multi-robot collaborative path planning methods are only some non-limiting implementation methods provided by the present invention, which are intended to clearly demonstrate the main concept of the present invention and provide some specific solutions that are convenient for the public to implement, rather than to limit all functions or all working modes of the multi-robot collaborative path planning system. Similarly, the multi-robot collaborative path planning system is also only a non-limiting implementation method provided by the present invention, and does not constitute a limitation on the execution subject and execution order of each step in these multi-robot collaborative path planning methods.

[0059] Please refer to Figure 1 and Figure 2 , Figure 1 A flowchart of a multi-robot collaborative path planning method provided according to some embodiments of the present invention is shown. Figure 2 A schematic diagram of a multi-robot collaborative path planning system provided according to some embodiments of the present invention is shown.

[0060] like Figure 1 As shown, the multi-robot collaborative path planning method 100 may include step S1: acquiring laser radar data of each robot to form environmental observation data.

[0061] Figure 2The multi-robot collaborative path planning system 200 shown includes a memory 210 and a processor 220. The memory 210 is configured with a computer-readable storage medium 211 provided by the present invention, and the computer-readable storage medium 211 stores computer instructions. The multi-robot collaborative path planning system 200 implements the multi-robot collaborative path planning method 100 provided by the present invention by executing the computer instructions.

[0062] The multi-robot collaborative path planning system 200 may first obtain the laser radar data of each robot. Each robot is a robot connected to and controlled by the multi-robot collaborative path planning system 200. The laser radar data may be 360-dimensional laser radar data collected by each robot at the current moment. The 360-dimensional laser radar data is represented as a one-dimensional feature vector, and the one-dimensional feature vector is used as environmental observation data.

[0063] Afterwards, if Figure 1 As shown, the multi-robot collaborative path planning system 200 can execute step S2: according to the environmental observation data and the target point position of each robot, the first action value of multiple actions in the action space of each robot is determined through the DQN (Deep Q-learning Network, a deep learning network) algorithm and the fully connected network, and the second action value of multiple actions is determined through the artificial potential field method.

[0064] The multi-robot collaborative path planning system 200 can set the motion space and target point position of each robot. Here, the motion space of each robot can include four executable motion strategies: up, down, left, and right. There can be multiple target point positions, and the target point positions of each robot can be consistent or different.

[0065] Please refer to Figure 3 , Figure 3 A diagram showing a reinforcement learning process in a multi-robot collaborative path planning method provided according to some embodiments of the present invention is shown.

[0066] exist Figure 3 In the preferred embodiment shown, the multi-robot collaborative path planning system 200 can first splice the environmental observation data 301 formed by the 360-dimensional lidar data with the 2-dimensional target point position 302 of each robot to obtain a 362-dimensional merged feature vector.

[0067] The DQN algorithm is used to input the 362-dimensional merged feature vector into the fully connected network to determine the first action value 311 of multiple actions in the action space of each robot. The result calculated based on the DQN algorithm can make the action strategy selected by the robot more robust, and at the same time enable the robot to reach the set target point faster while avoiding conflicts. Here, the number of layers of the fully connected network and the number of neurons in each layer can be dynamically adjusted according to actual conditions to achieve better path planning effects. Figure 3 In the example, the number of layers of the fully connected network can be 3, the number of neurons in the first layer of the fully connected network 321 can be 128, the number of neurons in the second layer of the fully connected network 322 can be 64, and the number of neurons in the third layer of the fully connected network 323 can be 4, thereby achieving a better path planning effect. Through the output of the three-layer fully connected network, the multi-robot collaborative path planning system 200 can determine the first action value 311. Here, the first action value can be recorded as q DQN (s,a).

[0068] like Figure 3 As shown, the multi-robot collaborative path planning system 200 can also determine the second action values ​​312 of multiple actions based on the environmental observation data 301 and the target point position 302.

[0069] First, the multi-robot collaborative path planning system 200 can respectively determine the artificial potential field method attraction vector 331 and the artificial potential field method repulsion vector 332. Specifically, the multi-robot collaborative path planning system 200 can determine the artificial potential field method attraction vector 331 according to the target point position 302 and the current spatial position of each robot. After that, the spatial position of the obstacle is determined through the environmental observation data 301, and then the artificial potential field method repulsion vector 332 is determined according to the current spatial position of each robot and the spatial position of the obstacle.

[0070] The multi-robot collaborative path planning system 200 may include an Evaluate Force Value module 340 and a Value Projection module 350. The Evaluate Force Value module 340 is used to calculate the resultant force of the artificial potential field, and the Value Projection module 350 is used to further extract features of the calculated resultant force of the artificial potential field through a neural network model.

[0071] By means of the force evaluation module 340, the value prediction module 350 and a fully connected network 360 for the action-value function, the vector decomposition of the artificial potential field method attraction vector 331 and the artificial potential field method repulsion vector 332 can be realized, thereby, the multi-robot collaborative path planning system 200 can determine the second action value 312 of multiple actions. Here, the number of neurons in the fully connected network 360 can be 4, and the second action value can be recorded as q APF (s,a).

[0072] By introducing the artificial potential field method, the multi-robot collaborative path planning system 200 can quickly calculate the repulsive force of obstacles on the robot and the gravitational force of the target point on the robot, thereby strengthening the learning action value calculation, guiding the robot to better avoid obstacles in the environment and quickly reach the destination. In addition, by using the artificial potential field method as part of the deep reinforcement learning action value calculation, the randomness of the algorithm is effectively reduced and the convergence speed of the model is improved.

[0073] Please refer to Figure 4 , Figure 4 A diagram showing a calculation principle of a second action value provided according to some embodiments of the present invention is shown. Figure 4 The current spatial position 401 of the robot, the target point position 402 and the spatial position 403 of the obstacle are shown.

[0074] A rectangular coordinate system is constructed with the current spatial position 401 of the robot as the center. Then, according to the current spatial position 401 of the robot and the target point position 402, the artificial potential field normal gravitational vector F can be determined. att According to the current spatial position 401 of the robot and the spatial position 403 of the obstacle, the repulsive force vector F of the artificial potential field method can be determined. rep . Figure 4 In the formula, F stands for F att and F rep Next, the artificial potential field normal gravitational vector F att and the repulsive force vector F of the artificial potential field method rep Converted to the corresponding action value vector q att and the action value vector q rep . Further, the action value vector q att and the action value vector q rep Decomposed into the second action value of the four action strategies of up, down, left, and right, corresponding to the second action value q in the figure U , the second action value q D , the second action value q L , the second action value q R .

[0075] Specifically, the artificial potential field method gravitational vector F att It can be calculated by the following formula:

[0076]

[0077] Among them, F att (X) is the gravitational vector of the artificial potential field method, K att is the first proportional coefficient, X is the current spatial position of each robot, X0 is the target point position, ρ(X-X0) is the distance between the current spatial position of each robot and the target point position, ρ att is the first distance threshold.

[0078] In addition, the repulsive force vector F rep It can be calculated by the following formula:

[0079]

[0080] Among them, F rep,i (X,X obs,i ) is the repulsive force vector of the artificial potential field method, K rep is the second proportional coefficient, X is the current spatial position of each robot, and X obs,i is the spatial position of the ith obstacle, ρ(X,X obs,i ) is the distance between the current spatial position of each robot and the spatial position of the i-th obstacle, ρ rep is the second distance threshold.

[0081] After that, please continue to refer to Figure 1 and Figure 3 , the multi-robot collaborative path planning system 200 can execute step S3: determine the complete action values ​​of multiple actions by linear addition according to the first action value and the second action value.

[0082] like Figure 3 As shown, the multi-robot collaborative path planning system 200 can set a third proportional coefficient ω and determine the complete action value 313 according to the third proportional coefficient ω.

[0083] The full action value can be determined according to the following formula:

[0084] q(s,a)=(1-ω)×q DQN (s,a)+ω×q APF (s,a),

[0085] Among them, q(s,a) is the complete action value 313, q DQN (s,a) is the first action value 311, q APF (s,a) is the second action value 312.

[0086] like Figure 1 As shown, the multi-robot collaborative path planning system 200 can execute step S4: determine and execute the action strategy with the highest action value based on the complete action values ​​of multiple actions.

[0087] Specifically, in a preferred embodiment, the multi-robot collaborative path planning system 200 can select and execute the action strategy with the largest action value according to the complete action values ​​q(s, a) of multiple actions. For example, the multi-robot collaborative path planning system 200 can calculate that the optimal action strategy of robot 1 is to move left, so robot 1 moves a distance m1 to the left; the optimal action strategy of robot 2 is to move right, so robot 2 moves a distance m2 to the right; the same applies to the other robots. Here, the specific path length can be determined by the map scale. Figure 1 Generally, it can be a raster map.

[0088] Afterwards, the multi-robot collaborative path planning system 200 may execute step S5: in response to each robot not all reaching the target point position, repeating steps S1 to S4 until each robot reaches the target point position.

[0089] The multi-robot collaborative path planning system 200 determines whether the current spatial position of each robot has reached the target point. If a robot has not reached the target point, it continues to obtain environmental observation data to repeat steps S1 to S4 to select an action strategy with the maximum action value for execution. The multi-robot collaborative path planning system 200 stops path planning until all robots have reached the target point.

[0090] In summary, by adopting a reinforcement learning model including the DQN algorithm, a fully connected network and an artificial potential field method, the multi-robot collaborative path planning method provided by the present invention can effectively improve the success rate of multi-robot collaborative path planning by combining the advantages of the traditional artificial potential field method in obstacle avoidance and short planning path, and the high robustness of reinforcement learning; at the same time, only the simplest fully connected network is used to calculate the path planning strategy, avoiding the iterative calculation complexity of the traditional path planning algorithm, and can quickly plan corresponding action strategies for multiple robots, thereby realizing multi-robot collaborative path planning in dense obstacle scenarios.

[0091] Although the above methods are illustrated and described as a series of actions for simplicity of explanation, it should be understood and appreciated that these methods are not limited by the order of the actions, because according to one or more embodiments, some actions may occur in a different order and / or concurrently with other actions from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art.

[0092] Those skilled in the art will appreciate that information, signals, and data may be represented using any of a variety of different technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips cited throughout the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or optical particles, or any combination thereof.

[0093] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or a combination of the two. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps are generally described above in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. The technician may implement the described functionality in different ways for each specific application, but such implementation decisions should not be interpreted as resulting in a departure from the scope of the present invention.

[0094] The various illustrative logic modules and circuits described in conjunction with the embodiments disclosed herein may be implemented or executed with a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in cooperation with a DSP core, or any other such configuration.

[0095] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software module may reside in a RAM memory, a flash memory, a ROM memory, an EPROM memory, an EEPROM memory, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read and write information from / to the storage medium. In an alternative, a storage medium may be integrated into a processor. The processor and the storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In an alternative, the processor and the storage medium may reside in a user terminal as discrete components.

[0096] In one or more exemplary embodiments, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented as a computer program product in software, each function may be stored on or transmitted by a computer-readable medium as one or more instructions or codes. Computer-readable media include both computer storage media and communication media, including any medium that facilitates the transfer of a computer program from one place to another. Storage media may be any available medium that can be accessed by a computer. As an example and not limitation, such a computer-readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of an instruction or data structure and can be accessed by a computer. Any connection is also properly referred to as a computer-readable medium. For example, if the software is transmitted from a website, a server, or other remote source using a coaxial cable, a fiber optic cable, a twisted pair, a digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwaves, the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of the medium. Disk and disc as used herein include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc, wherein disk often reproduces data magnetically, while disc reproduces data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

[0097] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-robot collaborative path planning method, characterized in that: Includes steps: S1: Obtain the lidar data of each robot to form environmental observation data; S2: according to the environmental observation data and the target point position of each robot, determine the first action value of multiple actions in the action space of each robot by using the DQN algorithm and the fully connected network, and determine the second action value of the multiple actions by using the artificial potential field method; S3: determining the complete action values ​​of the multiple actions by linear addition according to the first action value and the second action value; S4: determining and executing an action strategy with the highest action value according to the complete action values ​​of the multiple actions; and S5: In response to the fact that not all of the robots have reached the target point, repeat S1 to S4 until the robots have reached the target point.

2. The multi-robot collaborative path planning method according to claim 1, characterized in that: The lidar data is represented as a one-dimensional feature vector to form the environmental observation data.

3. The multi-robot collaborative path planning method according to claim 2, characterized in that: The S2 comprises the steps of: The one-dimensional feature vector of the environmental observation data and the robot target point position are concatenated to determine a combined feature vector, and the combined feature vector is input into the fully connected network using the DQN algorithm to determine the first action value.

4. The multi-robot collaborative path planning method according to claim 1, characterized in that: The action space of each robot includes four executable action strategies: up, down, left, and right.

5. The multi-robot collaborative path planning method according to claim 1, characterized in that: The step of determining the second action value of the plurality of actions by the artificial potential field method in S2 comprises: S21: Determine the gravitational vector of the artificial potential field method according to the position of the target point and the current spatial position of each robot; S22: Determine the spatial position of the obstacle according to the environmental observation data; S23: determining a repulsive force vector of an artificial potential field method according to the current spatial position of each robot and the spatial position of the obstacle; and S24: Perform vector decomposition on the artificial potential field method attraction vector and the artificial potential field method repulsion vector to determine the second action value.

6. The multi-robot collaborative path planning method according to claim 5, characterized in that: The artificial potential field method gravity vector is calculated by the following formula: Among them, F att (X) is the gravitational vector of the artificial potential field method, K att is the first proportional coefficient, X is the current spatial position of each robot, X0 is the target point position, ρ(X-X0) is the distance between the current spatial position of each robot and the target point position, ρ att is the first distance threshold.

7. The multi-robot collaborative path planning method according to claim 5, characterized in that: The repulsive force vector of the artificial potential field method is calculated by the following formula: Among them, F rep,i (X,X obs,i ) is the repulsive force vector of the artificial potential field method, K rep is the second proportional coefficient, X is the current spatial position of each robot, and X obs,i is the spatial position of the i-th obstacle, ρ(X,X obs,i ) is the distance between the current spatial position of each robot and the spatial position of the i-th obstacle, ρ rep is the second distance threshold.

8. The multi-robot collaborative path planning method according to claim 1, characterized in that: The S3 further comprises the steps of: The third proportional coefficient ω is set, and the complete action value q(a, a) is determined according to the third proportional coefficient ω as follows: q(s,a)=(1-ω)×q DQN (s,a)+ω×q APF (s,a), Among them, q DQN (s,a) is the first action value, q APF (s,a) is the second action value.

9. A multi-robot collaborative path planning system, characterized in that: include: a memory having computer instructions stored thereon; as well as A processor is connected to the memory and is configured to execute computer instructions stored in the memory to implement the multi-robot collaborative path planning method according to any one of claims 1 to 8.

10. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the multi-robot collaborative path planning method according to any one of claims 1 to 8 is implemented.

Images