WO2022032444A1 - 一种多智能主体避障方法、***和计算机可读存储介质 - Google Patents

一种多智能主体避障方法、***和计算机可读存储介质 Download PDF

Info

Publication number
WO2022032444A1
WO2022032444A1 PCT/CN2020/108245 CN2020108245W WO2022032444A1 WO 2022032444 A1 WO2022032444 A1 WO 2022032444A1 CN 2020108245 W CN2020108245 W CN 2020108245W WO 2022032444 A1 WO2022032444 A1 WO 2022032444A1
Authority
WO
WIPO (PCT)
Prior art keywords
intelligent
agents
task
agent
transportation
Prior art date
Application number
PCT/CN2020/108245
Other languages
English (en)
French (fr)
Inventor
程涛
于欣佳
Original Assignee
深圳技术大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳技术大学 filed Critical 深圳技术大学
Priority to PCT/CN2020/108245 priority Critical patent/WO2022032444A1/zh
Publication of WO2022032444A1 publication Critical patent/WO2022032444A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02—Control of position or course in two dimensions

Definitions

  • the present application relates to the field of swarm intelligence, and in particular, to a method, system and computer-readable storage medium for obstacle avoidance of a multi-intelligent subject.
  • Multi-intelligent agents carry goods in unmanned scenarios such as unmanned supermarkets and smart warehousing, which is a common application of intelligent agents. In these scenarios, how each intelligent agent of the multi-agent agent avoids obstacles during transportation is a problem worthy of study.
  • the sensor devices on the intelligent main body are often used to detect the obstacles, and then avoid the obstacles.
  • Embodiments of the present application provide a method, system, and computer-readable storage medium for a multi-intelligent subject to avoid obstacles, so as to solve the problem that the existing multi-intelligence subject sometimes cannot effectively avoid obstacles when carrying objects.
  • the technical solution is as follows:
  • a multi-intelligent subject obstacle avoidance method comprising:
  • the intelligent body After the intelligent body completes a handling task, it is searched whether the task list of the intelligent body that has completed the handling task is empty, and if it is not empty, the next handling task is assigned.
  • a multi-intelligent subject obstacle avoidance system comprising:
  • the map generation module is used to generate a grid map to distinguish the feasible transportation paths and shelves of multi-intelligent agents
  • an initialization module for initializing the grid map and the multi-agent
  • the task allocation module is used to traverse all the handling tasks and the multi-intelligent agents, and assign the handling tasks to one or more intelligent agents in the corresponding multi-intelligent agents according to the handling task assignment rules;
  • a path planning module used for planning the transport path of the intelligent subject assigned to the transport task
  • a path assignment module configured to assign the planned transport paths of the intelligent agents to corresponding intelligent agents for execution, so that the intelligent agents travel according to the transport paths;
  • the detection module is used for real-time monitoring of all intelligent subjects performing handling tasks, to detect whether the real-time position of each intelligent subject is normal, so as to make real-time online adjustments to intelligent subjects with abnormal positions;
  • the retrieval module is used for retrieving whether the task list of the intelligent body that has completed the transportation task is empty after the intelligent body completes a transportation task, and if it is not empty, continue to assign the next transportation task.
  • a multi-agent obstacle avoidance system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program code consisting of the one or more A processor loads and executes the operations performed by the multi-agent obstacle avoidance method.
  • a computer-readable storage medium stores a computer program loaded and executed by a processor to implement operations performed by the multi-agent obstacle avoidance method.
  • the handling tasks are allocated to one or more intelligent agents in the corresponding multi-intelligent agents according to the handling task assignment rules, and the intelligent agents assigned to the handling tasks are planned.
  • the main body plans the transportation path, and assigns the planned transportation paths of the intelligent agents to the corresponding intelligent agents for execution, so that the intelligent agents travel according to the transportation path, and monitors all intelligent agents that perform transportation tasks in real time, and detects the real-time status of each intelligent agent.
  • path planning the calculation speed is accelerated, and the selection of the optimal path is , considering the influence of the intelligent body turning, it is more practical, and the online adjustment method of the intelligent body is designed to make the intelligent body more practical and flexible.
  • FIG. 1 is a flowchart of a multi-agent obstacle avoidance method provided by an embodiment of the present application
  • FIG. 2 is a schematic structural diagram of a multi-intelligent subject obstacle avoidance system provided by an embodiment of the present application
  • FIG. 3 is a schematic functional structural diagram of a multi-agent obstacle avoidance system provided by another embodiment of the present application.
  • FIG. 1 it is a multi-intelligent subject obstacle avoidance method provided by an embodiment of the present application.
  • the method mainly includes the following steps S101 to S107 , which are described in detail as follows:
  • Step S101 generating a grid map to distinguish the feasible transport paths and shelves of the multi-intelligent agents.
  • the multi-intelligent subject includes a plurality of intelligent subjects, such as an intelligent group composed of automated guided vehicles (Automated Guided Vehicle, AGV), and each grid of the grid map represents a node or the current location of the intelligent subject.
  • AGV Automated Guided Vehicle
  • Step S102 Initialize the grid map and the multi-intelligent agent.
  • One of the purposes of initializing the grid map is to release the occupancy of each node and intelligent subject in the map by the previous handling task.
  • Step S103 traverse all the handling tasks and the multi-intelligent agents, and assign the handling tasks to one or more intelligent agents in the corresponding multi-intelligent agents according to the assignment rule of the transportation tasks.
  • the present application can traverse all the tasks to be transported and the current attributes of the multi-intelligent agents to obtain the length of the task list, the distance between each intelligent agent and the target transportation task, and the distance of the comprehensive operation, and prioritize the target transportation task. Assign it to idle intelligent agents, and then assign it to the intelligent agent with the fewest tasks in the task list, or assign the target transportation task to the intelligent agent closest to the target transportation task, or prioritize the target transportation task between a group of access tasks.
  • the travel distance is the closest intelligent agent.
  • the length of the task list can be examined, and the target handling task is given priority to idle, and then to the intelligent subject with the fewest tasks in the task list; or, the distance between each intelligent subject and the target handling task is given priority to the target handling task. It is assigned to the intelligent subject closest to the target handling task to improve the response speed of the intelligent subject to the target handling task; or, the distance of the comprehensive operation, that is, when assigning the target handling task, comprehensively consider the distance between a group of access target handling tasks. Travel distance, the target handling task is given priority to the intelligent agent with the shortest travel distance between a group of access tasks.
  • Step S104 Plan the transport path of the intelligent agent assigned to the transport task.
  • planning the transport path of the intelligent body assigned to the transport task can be implemented through the following steps S1041 to S1043:
  • Step S1041 Plan the transportation path of the intelligent agent according to the priority of the task, prioritize the transportation path of the intelligent agent that undertakes the transportation task of higher priority, and/or when the multi-agent is congested on the transportation path, according to the transportation task undertaken by the intelligent agent
  • the priority of the handling task can be determined according to the importance of the bill of lading, that is, the priority of the corresponding handling task is determined according to the importance of the pre-set bill of lading. The more important the bill of lading is, the more important the handling task is. The higher the priority is, the priority of the handling task can also be judged according to the arrival time of the bill of lading. The earlier the bill of lading arrives, the higher the priority of the corresponding handling task.
  • Step S1042 According to the planned transportation path, train the transportation model of each multi-intelligent agent.
  • training the transportation model of each multi-intelligent agent can be realized through the following steps S1 to S5:
  • Step S1 Calculate the distance from each grid in the grid map to the target grid.
  • each grid in the grid map is the current location of the intelligent agent
  • the target grid is the final node that the intelligent agent will reach, that is, the destination.
  • the distance from each grid to the target grid is represented by the Manhattan distance, where the distance from the obstacle grid to the target grid is infinite, that is, it cannot be reached, and each grid to the target grid
  • lattice distance C i is:
  • C i also represents the cost function of the ith grid
  • d(i, ig ) represents the Manhattan distance from the ith grid to the target grid ig, Indicates the weighted value, the more intelligent agents that pass through the ith grid, the greater the value.
  • Step S2 Select a grid with the smallest cost function from the target grid around the starting grid where the intelligent agent A i is currently located as a neighbor grid and store it in the route r_te.
  • the neighbor grid is the grid that the intelligent agent A i will pass through next when it goes from the current starting grid to the target grid. It should be noted that if the distance between the two grids is the same from the target grid, one of the grids is randomly selected as the neighbor grid.
  • Step S3 Repeat the above step S2 until the intelligent agent A i finally reaches the target grid.
  • Step S4 Repeat steps S1 to S3 for a total of y times to generate a path with the shortest length and the least number of turns.
  • steps S1 to S3 for a total of y times, generating y results, calculating the length of each path according to the formula length(r_te), selecting the path with the shortest length from the running paths of the y intelligent agents, and then Among the shortest paths, a conveyance path with the least number of turns is selected as the conveyance path of the intelligent subject.
  • tr represents the judgment value of whether the intelligent subject turns, if its value is 1, the intelligent subject turns, if its value is 0, the intelligent subject does not turn
  • r_te(i, 1) means the i-th route r_te
  • the abscissa of the coordinates, r_te(i, 2) represents the ordinate of the i-th coordinate of the route r_te
  • tr_num represents the total number of turns of the path of the intelligent subject, and selects a transport path with the least number of turns in the shortest path as the The transport path of the intelligent subject.
  • Step S5 Perform weighting processing on the final path of each intelligent agent to form a transport model of the multi-agent with the least collision.
  • Step S1043 According to the trained transport model, analyze the interference and collision situation between the transport paths of all intelligent agents, and adjust the transport paths of some intelligent agents.
  • the trained transportation model analyze the interference and collision between the transportation paths of all intelligent agents, and adjust the transportation paths of some intelligent agents: ensure that the transportation path of the intelligent agent with the highest priority does not move, compare The transportation path of the intelligent agent with the second priority and the node that collides with it. If it exists, different avoidance strategies are adopted according to the type of conflict.
  • the collision judgment and optimization of the path of all intelligent agents according to the priority level are carried out in turn. , until the paths of all intelligent agents are planned and there is no collision with each other.
  • the unit step length of the transportation task is used to detect whether there is a collision between the intelligent agents and where they collide.
  • the position of each intelligent agent in the first unit time is detected, and whether there is a collision or a conflict is determined.
  • the collision type is judged and solved, and then, the position of each intelligent subject in the second unit time is detected, and the cycle is repeated until all intelligent subjects reach the end point and no collision occurs.
  • the above-mentioned collision types include: when the collision nodes of two intelligent agents are at the intersection, the collision of the traffic nodes occurs; when the two intelligent agents are driving in the opposite direction, the collision occurs in the opposite direction; when the two intelligent agents travel in the same direction, when one intelligent agent When trying to overtake another agent, a catch-up collision occurs, and so on.
  • the above avoidance strategies include a non-interference collision strategy, an incomplete interference collision strategy and a complete interference collision strategy.
  • the non-interference collision strategy means that the intelligent subject with low priority waits in place, and the intelligent subject with high priority goes first. Passing through traffic nodes; the incomplete interference collision strategy means that when a low-priority intelligent agent is on the subsequent driving route of a high-priority intelligent agent, this section of the collision path is blocked, and the low-priority intelligent agent travels the most recent route.
  • the traffic node is the starting point, and the next route is re-planned according to the method described in step S5.
  • the intelligent subject with high priority keeps the original route and continues to drive.
  • the agent with low priority waits in place, and the agent with high priority drives on the original road; the collision strategy of complete interference type is about to block the collision path, and then the agent with low priority will drive to the nearest traffic node.
  • the next path planning is performed again according to the method described in step S5, and the intelligent subject with high priority continues to travel according to the original path.
  • Step S105 Allocate the planned transport paths of the intelligent agents to the corresponding intelligent agents for execution, so that the intelligent agents travel according to the transport paths.
  • Step S106 Perform real-time monitoring on all intelligent subjects performing the handling task, and detect whether the real-time positions of each intelligent subject are normal, so as to perform real-time online adjustment on the intelligent subjects with abnormal positions.
  • the intelligent subject with abnormal position includes: an intelligent subject whose running speed is too fast and an intelligent subject whose running speed is too slow.
  • the intelligent subject is ahead of the planned position. At this time, it is only necessary to reduce the speed of the intelligent subject or make it wait for a period of time.
  • the intelligent subject lags behind the planned position. At this time, the running speed of the intelligent subject is accelerated so that it can return to the planned position when the next step is long.
  • Step S107 after the intelligent body completes a transportation task, it is searched whether the task list of the intelligent body that has completed the transportation task is empty, and if it is not empty, the next transportation task is continued to be assigned.
  • the handling tasks are allocated to one or more intelligent agents in the corresponding multi-intelligence agents according to the handling task assignment rules, and the The intelligent agent plans the transportation path, and assigns the planned transportation path of the intelligent agent to the corresponding intelligent agents for execution, so that the intelligent agent travels according to the transportation path, and monitors all intelligent agents that perform the transportation task in real time.
  • path planning the calculation speed is accelerated, and the selection of the optimal path In , considering the influence of the intelligent subject's turning, it is more practical, and the online adjustment method of the intelligent subject is designed to make the intelligent subject more practical and flexible.
  • FIG. 2 is a schematic structural diagram of a multi-agent obstacle avoidance system provided by an embodiment of the present application.
  • the system may include a map generation module 201, an initialization module 202, a task allocation module 203, a path planning module 204, and a path allocation module.
  • module 205, detection module 206 and retrieval module 207 wherein:
  • the map generation module 201 is used to generate a grid map to distinguish the feasible transportation paths and shelves of the multi-intelligent agents;
  • an initialization module 202 used to initialize the grid map and the multi-intelligent agent
  • the task allocation module 203 is used for traversing all the handling tasks and the multi-intelligent agents, and assigning the handling tasks to one or more intelligent agents in the corresponding multi-intelligent agents according to the handling task assignment rules;
  • the path planning module 204 is used to plan the transport path of the intelligent subject assigned to the transport task;
  • the path assignment module 205 is used for assigning the planned transportation paths of the intelligent agents to the corresponding intelligent agents for execution, so that the intelligent agents travel according to the transportation paths;
  • the detection module 206 is used for real-time monitoring of all the intelligent subjects performing the handling task, to detect whether the real-time position of each intelligent subject is normal, so as to perform real-time online adjustment to the intelligent subjects with abnormal positions;
  • the retrieval module 207 is used to retrieve whether the task list of the intelligent agent that has completed the transportation task is empty after the intelligent agent completes a transportation task, and if it is not empty, continue to assign the next transportation task.
  • the task assignment module 203 may include a traversal unit and a control unit, wherein:
  • Traversing unit used to traverse all the tasks to be transported and the current attributes of the multi-intelligent agents, to obtain the length of the task list, the distance of each intelligent agent from the target transport task, and the distance of the comprehensive operation;
  • the priority allocation unit is used to preferentially assign the target handling task to the idle intelligent agent, and then assign it to the intelligent agent with the fewest tasks in the task list, or assign the target transportation task to the intelligent agent closest to the target transportation task first, or assign the The target handling task prioritizes a group of intelligent agents with the shortest travel distance between the access tasks.
  • the path planning module 204 may include a priority planning unit, a model training unit and an analysis and adjustment unit, wherein:
  • the priority planning unit is used to plan the transportation path of the intelligent agent according to the priority of the task, and preferentially plan the transportation path of the intelligent agent that undertakes the transportation task of higher priority and/or when the multi-agent is congested on the transportation path, according to the The higher the priority of undertaking the transportation task, the higher the priority of the right of way to plan the transportation path of the intelligent subject;
  • the model training unit is used to train the transportation model of each multi-intelligent agent according to the planned transportation path;
  • the analysis and adjustment unit is used to analyze the interference and collision between the transportation paths of all intelligent agents according to the trained transportation model, and adjust the transportation paths of some intelligent agents.
  • the priority of the handling task is judged according to the importance of the bill of lading and/or according to the arrival time of the bill of lading.
  • the model training unit may include a calculation unit, a selection unit, a generation unit and a weighting unit, wherein:
  • a calculation unit used to calculate the distance from each grid in the grid map to the target grid
  • the selection unit is used to select a grid with the smallest cost function from the target grid around the starting grid where the intelligent agent A i is currently located as a neighbor grid and store it in the route r_te;
  • the above selection unit is repeated until the intelligent agent A i finally reaches the target grid;
  • a generating unit is used to repeatedly execute steps S1 to S3 for a total of y times, and generate a path with the shortest length and the least number of turns as the final path of the intelligent subject;
  • the weighting unit is used to weight the final path of each intelligent agent to form a transport model of the multi-agent with the least collision.
  • the analysis and adjustment unit may include an avoidance policy selection unit and an optimization unit, wherein:
  • the avoidance policy selection unit is used to ensure that the transportation path of the intelligent agent with the highest priority does not move, and compares the transportation path of the intelligent agent with the second priority and whether there is a collision node. If there is, adopt different avoidance strategies according to the type of conflict ;
  • the optimization unit is used for, according to the above method, to perform collision judgment and optimization of the paths of all the intelligent agents in sequence according to the priority level, until the paths of all the intelligent agents are planned and there is no collision with each other.
  • the intelligent agents with abnormal positions include: an intelligent agent whose running speed is too fast and an intelligent agent whose running speed is too slow.
  • the multi-agent obstacle avoidance system provided by the above-mentioned embodiments, when the multi-agent obstacle avoidance is performed, only the division of the above-mentioned functional modules is used as an example for illustration.
  • the function module is completed, that is, the internal structure of the system is divided into different function modules to complete all or part of the functions described above.
  • the multi-agent obstacle avoidance system provided by the above embodiments and the multi-agent obstacle avoidance method embodiments belong to the same concept, and the specific implementation process and technical effects thereof are detailed in the method embodiments, which will not be repeated here.
  • Embodiments of the present application further provide a multi-agent obstacle avoidance system.
  • the multi-agent obstacle avoidance system is shown in FIG. 3 , which shows a schematic structural diagram of the multi-agent obstacle avoidance system involved in the embodiments of the present application. In terms of:
  • the multi-agent obstacle avoidance system may include a processor 301 of one or more processing cores, a memory 302 of one or more computer-readable storage media, a power supply 303 and an input unit 304 and other components.
  • a processor 301 of one or more processing cores may include a processor 301 of one or more processing cores, a memory 302 of one or more computer-readable storage media, a power supply 303 and an input unit 304 and other components.
  • FIG. 3 does not constitute a limitation on the multi-agent obstacle avoidance system, and may include more or less components than those shown in the figure, or combine some components, or a different arrangement of components. in:
  • the processor 301 is the control center of the multi-agent obstacle avoidance system, and uses various interfaces and lines to connect various parts of the entire multi-agent obstacle avoidance system, and by running or executing the software programs and/or modules stored in the memory 302, And call the data stored in the memory 302 to execute various functions of the multi-intelligent subject obstacle avoidance system and process data, so as to perform overall monitoring of the multi-intelligence subject obstacle avoidance system.
  • the processor 301 may include one or more processing cores; preferably, the processor 301 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, etc. , the modem processor mainly deals with wireless communication. It can be understood that, the above-mentioned modulation and demodulation processor may not be integrated into the processor 301.
  • the memory 302 can be used to store software programs and modules, and the processor 301 executes various functional applications and data processing by running the software programs and modules stored in the memory 302 .
  • the memory 302 may mainly include a stored program area and a stored data area, wherein the stored program area may store an operating system, an application program required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; Data created by the use of multi-agent obstacle avoidance systems, etc.
  • memory 302 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device. Accordingly, memory 302 may also include a memory controller to provide processor 301 access to memory 302 .
  • the multi-agent obstacle avoidance system also includes a power supply 303 for supplying power to each component.
  • the power supply 303 can be logically connected to the processor 301 through the power management system, so as to manage charging, discharging, and power consumption management through the power management system.
  • the power source 303 may also include one or more DC or AC power sources, recharging systems, power failure detection circuits, power converters or inverters, power status indicators, and any other components.
  • the multi-intelligent subject obstacle avoidance system may further include an input unit 304, which may be used to receive input numerical or character information, and generate keyboard, mouse, joystick, optical or trackball signals related to user settings and function control enter.
  • an input unit 304 which may be used to receive input numerical or character information, and generate keyboard, mouse, joystick, optical or trackball signals related to user settings and function control enter.
  • the multi-intelligent subject obstacle avoidance system may also include a display unit and the like, which will not be described herein again.
  • the processor 301 in the multi-agent obstacle avoidance system will load the executable file corresponding to the process of one or more application programs into the memory 302 according to the following instructions, and the processor 301 will to run the application program stored in the memory 302, so as to realize various functions, as follows: generate a grid map to distinguish the feasible transport paths and shelves of multi-agent agents; initialize the grid map and multi-agent agents; traverse all transport tasks and Multi-intelligent agents, according to the assignment rules of transportation tasks, assign the transportation tasks to one or more intelligent agents in the corresponding multi-intelligence agents; plan the transportation paths of the intelligent agents assigned to the transportation tasks; assign the planned transportation paths of the intelligent agents respectively Perform the execution on the corresponding intelligent subject, so that the intelligent subject travels according to the transport path; perform real-time monitoring on all intelligent subjects performing the transport task, and detect whether the real-time position of
  • the planned transportation paths of the intelligent agents are respectively assigned to the corresponding intelligent agents for execution, so that the intelligent agents can travel according to the transportation path, monitor all the intelligent agents performing the transportation tasks in real time, and detect whether the real-time positions of the intelligent agents are normal, so as to determine whether the intelligent agents are in a normal position.
  • Real-time online adjustment of the intelligent subject with abnormal position so that the transportation path of the intelligent subject is as balanced as possible, reducing the probability of collision.
  • the calculation speed is accelerated, and in the selection of the optimal path, the turning of the intelligent subject is considered. The impact brought by it is more realistic, and the method of online adjustment of the intelligent subject is designed to make the intelligent subject more practical and flexible.
  • the embodiments of the present application provide a computer-readable storage medium, in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute any of the multi-agent obstacle avoidance methods provided by the embodiments of the present application steps in .
  • the instruction can perform the following steps: generate a grid map to distinguish feasible transport paths and shelves for multi-agents; initialize the grid map and multi-agent; traverse all transport tasks and multi-agents, and assign them according to the transport task allocation rules
  • the handling task is assigned to one or more intelligent agents in the corresponding multi-intelligence agents; the intelligent agents assigned to the transportation tasks are planned to plan the transportation path; the planned intelligent agent transportation paths are respectively assigned to the corresponding intelligent agents for execution, so that The intelligent subject travels according to the transport path; real-time monitoring of all intelligent subjects performing the transport task is performed to detect whether the real-time position of each intelligent subject is normal, so as to make real-time online adjustment to the intelligent subject with abnormal position; when the intelligent subject completes a transport task After that, it is searched whether the task list of the intelligent subject that has completed the transport task is empty, and if it is not empty, the next transport task is continued to be assigned.
  • the computer-readable storage medium may include: a read-only memory (ROM, Read Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and the like.
  • any multi-agent obstacle avoidance methods provided by the embodiments of the present application can be implemented.
  • the beneficial effects that can be achieved by the intelligent subject obstacle avoidance method can be seen in the previous embodiments, and details are not repeated here.

Landscapes

  • Engineering & Computer Science (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)

Abstract

一种多智能主体避障方法、***和计算机可读存储介质，该方法包括：生成栅格地图，以区分出多智能主体可行搬运路径与货架（S101）；初始化栅格地图和多智能主体（S102）；遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体（S103）；规划分配到搬运任务的智能主体规划搬运路径（S104）；将规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使智能主体按照搬运路径行驶（S105）；对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整（S106）；当智能主体完成一个搬运任务后，检索完成搬运任务的智能主体的任务列表是否为空，若不为空，则继续分配下一搬运任务（S107）。

Description

一种多智能主体避障方法、***和计算机可读存储介质 技术领域
本申请涉及群体智能领域，特别涉及一种多智能主体避障方法、***和计算机可读存储介质。
背景技术
在群体智能领域中，智能主体(例如传感器、机器人、飞行器等)的个体能力有限，但其群体却能表现出高效的协同合作能力和高级的智能协调水平。多智能主体在无人超市、智慧仓储等无人场景下进行货物的搬运，是智能主体比较常见的应用。在这些场景下，多智能主体的每个智能主体如何在搬运时避开障碍物是值得研究的问题。现有技术在避障时，往往是依靠智能主体上的传感器件检测到障碍物，然后，避开障碍物。
然而，上述避障方法仍然存在一定局限性，例如，当多智能主体中的某个智能主体构成″障碍物″时，仅仅依靠传感器件有时也无法避开。
发明内容
本申请实施例提供了一种多智能主体避障方法、***和计算机可读存储介 质，以解决现有的多智能主体在搬运物件有时并不能有效避开障碍物的问题。该技术方案如下：
一方面，提供了一种多智能主体避障方法，该方法包括：
生成栅格地图，以区分出多智能主体可行搬运路径与货架；
初始化所述栅格地图和多智能主体；
遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体；
规划所述分配到搬运任务的智能主体规划搬运路径；
将所述规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使所述智能主体按照所述搬运路径行驶；
对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整；
当智能主体完成一个搬运任务后，检索完成搬运任务的智能主体的任务列表是否为空，若不为空，则继续分配下一搬运任务。
一方面，提供了一种多智能主体避障***，该***包括：
地图生成模块，用于生成栅格地图，以区分出多智能主体可行搬运路径与货架；
初始化模块，用于初始化所述栅格地图和多智能主体；
任务分配模块，用于遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体；
路径规划模块，用于规划所述分配到搬运任务的智能主体规划搬运路径；
路径分配模块，用于将所述规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使所述智能主体按照所述搬运路径行驶；
检测模块，用于对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整；
检索模块，用于当智能主体完成一个搬运任务后，检索完成搬运任务的智能主体的任务列表是否为空，若不为空，则继续分配下一搬运任务。
一方面，提供了一种多智能主体避障***，该***包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机程序，该计算机程序代码由该一个或多个处理器加载并执行以实现该多智能主体避障方法所执行的操作。
一方面，提供了一种计算机可读存储介质，该计算机可读存储介质存储有计算机程序由处理器加载并执行以实现该多智能主体避障方法所执行的操作。
从上述本申请提供的技术方案可知，遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体，规划分配到搬运任务的智能主体规划搬运路径，将规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使智能主体按照搬运路径行驶，对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整，使智能主体的搬运路径尽量均衡，减少了冲撞发生的概率，在路径规划中，加快了计算速度，同时在 最优路径的选择中，考虑了智能主体转弯所带来的影响，更加贴合实际，设计了智能主体在线调整的方法，使智能主体更具实用性，更有灵活性。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是本申请实施例提供的多智能主体避障方法的流程图；
图2是本申请实施例提供的多智能主体避障***的结构示意图；
图3是本申请另一实施例提供的多智能主体避障***的功能结构示意图。
具体实施方式
为使本申请的目的、技术方案和优点更加清楚，下面将结合附图对本申请实施方式作进一步地详细描述。
参见图1，是本申请实施例提供的一种多智能主体避障方法，该方法主要包括以下步骤S101至S107，详细说明如下：
步骤S101：生成栅格地图，以区分出多智能主体可行搬运路径与货架。
在本申请实施例中，多智能主体包括多个智能主体，例如自动引导车辆(Automated Guided Vehicle，AGV)组成的智能群体，栅格地图的每一个格子表示一个节点或者智能主体当前所在位置。
步骤S102：初始化栅格地图和多智能主体。
初始化栅格地图，目的之一是为了解除上一搬运任务对地图中各节点以及智能主体的占用情况。
步骤S103：遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体。
作为本申请一个实施例，可以是遍历所有待搬运任务和多智能主体的当前属性，以获取任务列表的长度、每个智能主体距离目标搬运任务的距离以及综合作业的距离，将目标搬运任务优先分配给空闲智能主体，其次分配给任务列表中任务最少的智能主体，或者，将目标搬运任务优先分配给距离目标搬运任务最近的智能主体，或者，将目标搬运任务优先一组存取任务之间的行程距离最近的智能主体。换言之，可以考察任务列表的长度，优先将目标搬运任务优先分配给空闲，其次分配给任务列表中任务最少的智能主体；或者，每个智能主体距离目标搬运任务的距离，优先将目标搬运任务优先分配给距离目标搬运任务最近的智能主体，提高智能主体对目标搬运任务的响应速度；或者，综合作业的距离，即在进行目标搬运任务分配时，综合考虑一组存取目标搬运任务之间的行程距离，优先将目标搬运任务优先一组存取任务之间的行程距离最近的智能主体。
步骤S104：规划分配到搬运任务的智能主体的搬运路径。
作为本申请一个实施例，规划分配到搬运任务的智能主体的搬运路径可通过如下步骤S1041至步骤S1043实现：
步骤S1041：根据任务的优先级进行智能主体搬运路径的规划，优先规划承担较高优先级搬运任务的智能主体的搬运路径和/或在多智能主体在搬运路径上拥堵时，按照所承担搬运任务优先级越高，给予越高的优先通行权的原则规划智能主体的搬运路径。
可以理解的是，在多智能主体中，所承担的搬运任务的优先级越高，该智能主体应该优先规划，同时，该智能主体在所规划的搬运路径上具有更高的通过某个节点的优先权，即，当两个智能主体同时经过某个节点时，为防止冲撞，所承担搬运任务优先级高的智能主体优先通过，优先级低的智能主体选择等待或者重新规划路径。上述实施例中，搬运任务的优先级高低可以根据提货单的重要程度来定，即，按照预先设置的提货单的重要程度来确定相应搬运任务的优先级，提货单的越重要，相应搬运任务的优先级越高，搬运任务的优先级还可以根据提货单的到达时间来判断，提货单来得越早，相应搬运任务的优先级越高。
步骤S1042：按照规划好的搬运路径，训练每多智能主体的搬运模型。
具体而言，按照规划好的搬运路径，训练每多智能主体的搬运模型可通过如下步骤S1至步骤S5实现：
步骤S1：计算栅格地图中每一个栅格至目标栅格的距离。
如前所述，栅格地图中的每个栅格即智能主体当前所在位置，目标栅格即智能主体所要到达的最终节点即目的地。在本申请实施例中，每一个栅格至目 标栅格的距离，距离用曼哈顿距离表示，其中，障碍物栅格到目标栅格的距离为无穷大，即不能到达，每一个栅格至目标栅格的距离C i的表达式为：
Figure PCTCN2020108245-appb-000001
上述表达式中，C i也表示第i个栅格的成本函数，d(i，i g)表示第i个栅格到目标栅格i g的曼哈顿距离，
Figure PCTCN2020108245-appb-000002
表示加权值，经过该第i个栅格的智能主体越多，其值越大。
步骤S2：在智能主体A i当前所在起始栅格四周选择距离目标栅格的成本函数最小的一个栅格作为邻居栅格并存储于路由r_te中。
邻居栅格也即智能主体A i从当前所在起始栅格去到目标栅格时，下一步要经过的栅格。需要说明的是，若两个栅格距离目标栅格的距离一样，则随机选择其中一个栅格作为邻居栅格。
步骤S3：重复执行上述步骤S2，直至智能主体A i最后到达目标栅格。
步骤S4：重复执行步骤S1至步骤S3共y次，生成长度最短和转弯次数最少的路径。
具体而言，重复执行步骤S1至步骤S3共y次，生成y个结果，根据公式leng(r_te)计算每个路径的长度，从y个智能主体的运行路径中选择出长度最短的路径，然后在最短的路径中选择一条转弯次数最少的搬运路径，作为该智能主体的的搬运路径。
进一步地，判断智能主体是否转弯的公式为：
Figure PCTCN2020108245-appb-000003
tr_num＝∑tr
上述表达式中，tr表示智能主体是否转弯的判断值，若其值为1，则智能主体转弯，若其值为0，则智能主体不转弯，r_te(i，1)表示路由r_te第i个坐标的横坐标，r_te(i，2)表示路由r_te第i个坐标的纵坐标，tr_num表示该智能主体的路径的总转弯次数，在最短的路径中选择一条转弯次数最少的搬运路径，作为该智能主体的的搬运路径。
步骤S5：对每个智能主体的最终路径进行加权处理，以形成冲撞最小的多智能主体的搬运模型。
对已经有智能主体走过的路径进行加权处理，每个智能主体已经走过的栅格其初始值增加β ij，此次β ij＝1，进一步地，对智能主体行驶过的路径加权处理可以让多个智能主体在进行路径规划的过程中，减少路径互相重合、冲突的可能性，更有利于后期的规划。
步骤S1043：按照已训练搬运模型，对所有智能主体的搬运路径之间的干扰碰撞情况进行分析并调整部分智能主体的的搬运路径。
具体地，按照已训练搬运模型，对所有智能主体的搬运路径之间的干扰碰撞情况进行分析并调整部分智能主体的的搬运路径可以是：保证优先级最高的智能主体的搬运路径不动，比较优先级第二的智能主体的搬运路径与其是否存在冲撞的节点，若存在，则针对冲突类型采取不同的避让策略，按照上述方法，依次对所有智能主体按照优先级高低进行路径的冲撞判断与优化，直至所有智 能主体的路径规划完毕且互相之间没有冲撞为止。具体地，通过搬运任务时行驶的单位步长来检测智能主体之间是否有冲撞以及在何处冲撞，第一次检测第一个单位时间各智能主体所处位置，判断是否有冲撞，有冲突进行冲撞类型的判断并解决，然后，检测第二个单位时间各智能主体的所处位置，以此循环，直到所有智能主体都到达终点，没有冲撞产生为止。上述冲撞类型包括：当两个智能主体的冲撞节点在十字路口时，即发生交通节点冲撞；当两个智能主体的相向行驶时，发生相向冲撞；两个智能主体同向行驶，当一个智能主体要超过另一个智能主体时，即发生赶超冲撞，等等。上述避让策略包括互不干扰型冲撞策略、不完全干扰型冲撞策略和完全干扰型冲撞策略，其中，互不干扰型冲撞策略即优先级低的智能主体原地等待，优先级高的智能主体先行通过交通节点；不完全干扰型冲撞策略即当优先级低的智能主体在优先级高的智能主体的后续行驶路线上时，将这一段冲撞路径封锁，优先级低的智能主体以最近的行驶过的交通节点为起点，按照步骤S5所述方法重新进行规划接下来的路径，优先级高的智能主体保持原先路线不变继续行驶，当优先级低的智能主体不在优先级高的智能主体的后续行驶路线上时，优先级低的智能主体原地等待，优先级高的智能主体按原路行驶；完全干扰型冲撞策略即将冲撞路径封锁，然后优先级低的智能主体以行驶过的最近交通节点为起点，按照步骤S5所述方法重新进行接下来的路径规划，优先级高的智能主体按原先路径继续行驶。
步骤S105：将规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使智能主体按照搬运路径行驶。
步骤S106：对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整。
在本申请实施例中，位置异常的智能主体包括：运行速度过快的智能主体和运行速度过慢的智能主体。当智能主体的运行速度过快常时，该智能主体相对于规划位置提前了，此时，只要降低该智能主体的速度或者令其等待一段时间即可，而当智能主体的运行速度过慢时，智能主体相对于规划位置落后，此时，加快该智能主体的运行速度，使它在下一步长的时候能回到计划位置。
步骤S107：当智能主体完成一个搬运任务后，检索完成搬运任务的智能主体的任务列表是否为空，若不为空，则继续分配下一搬运任务。
从上述附图1示例的技术方案可知，遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体，规划分配到搬运任务的智能主体规划搬运路径，将规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使智能主体按照搬运路径行驶，对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整，使智能主体的搬运路径尽量均衡，减少了冲撞发生的概率，在路径规划中，加快了计算速度，同时在最优路径的选择中，考虑了智能主体转弯所带来的影响，更加贴合实际，设计了智能主体在线调整的方法，使智能主体更具实用性，更有灵活性。
请参阅附图2，是本申请实施例提供的一种多智能主体避障***的结构示意图，该***可以包括地图生成模块201、初始化模块202、任务分配模块203、 路径规划模块204、路径分配模块205、检测模块206和检索模块207，其中：
地图生成模块201，用于生成栅格地图，以区分出多智能主体可行搬运路径与货架；
初始化模块202，用于初始化栅格地图和多智能主体；
任务分配模块203，用于遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体；
路径规划模块204，用于规划分配到搬运任务的智能主体的搬运路径；
路径分配模块205，用于将规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使智能主体按照搬运路径行驶；
检测模块206，用于对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整；
检索模块207，用于当智能主体完成一个搬运任务后，检索完成搬运任务的智能主体的任务列表是否为空，若不为空，则继续分配下一搬运任务。
在一种可能实现方式中，任务分配模块203可以包括遍历单元和控制单元，其中：
遍历单元，用于遍历所有待搬运任务和多智能主体的当前属性，以获取任务列表的长度、每个智能主体距离目标搬运任务的距离以及综合作业的距离；
优先分配单元，用于将目标搬运任务优先分配给空闲智能主体，其次分配给任务列表中任务最少的智能主体，或者，将目标搬运任务优先分配给距离目 标搬运任务最近的智能主体，或者，将目标搬运任务优先一组存取任务之间的行程距离最近的智能主体。
在一种可能实现方式中，路径规划模块204可以包括优先规划单元、模型训练单元和分析调整单元，其中：
优先规划单元，用于根据任务的优先级进行智能主体搬运路径的规划，优先规划承担较高优先级搬运任务的智能主体的搬运路径和/或在多智能主体在搬运路径上拥堵时，按照所承担搬运任务优先级越高，给予越高的优先通行权的原则规划智能主体的搬运路径；
模型训练单元，用于按照规划好的搬运路径，训练每多智能主体的搬运模型；
分析调整单元，用于按照已训练搬运模型，对所有智能主体的搬运路径之间的干扰碰撞情况进行分析并调整部分智能主体的的搬运路径。
在一种可能实现方式中，搬运任务的优先级的高低根据提货单的重要程度和/或根据提货单的到达时间来判断。
在一种可能实现方式中，模型训练单元可以包括计算单元、选择单元、生成单元和加权单元，其中：
计算单元，用于计算所述栅格地图中每一个栅格至目标栅格的距离；
选择单元，用于在智能主体A i当前所在起始栅格四周选择距离目标栅格的成本函数最小的一个栅格作为邻居栅格并存储于路由r_te中；
上述选择单元重复执行，直至所述智能主体A i最后到达所述目标栅格；
生成单元，用于重复执行步骤S1至步骤S3共y次，生成长度最短和转弯次数最少的路径作为所述智能主体的最终路径；
加权单元，用于对每个智能主体的最终路径进行加权处理，以形成冲撞最小的多智能主体的搬运模型。
在一种可能实现方式中，分析调整单元可以包括避让政策选取单元和优化单元，其中：
避让政策选取单元，用于保证优先级最高的智能主体的搬运路径不动，比较优先级第二的智能主体的搬运路径与其是否存在冲撞的节点，若存在，则针对冲突类型采取不同的避让策略；
优化单元，用于按照上述方法，依次对所有智能主体按照优先级高低进行路径的冲撞判断与优化，直至所有智能主体的路径规划完毕且互相之间没有冲撞为止。
在一种可能实现方式中，位置异常的智能主体包括：运行速度过快的智能主体和运行速度过慢的智能主体。
需要说明的是，上述实施例提供的多智能主体避障***在多智能主体避障时，仅以上述各功能模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能模块完成，即将***的内部结构划分成不同的功能模块，以完成以上描述的全部或者部分功能。另外，上述实施例提供的多智能主体避障***与多智能主体避障方法实施例属于同一构思，其具体实现过程以及技术效果详见方法实施例，此处不再赘述。
本申请实施例还提供一种多智能主体避障***，该多智能主体避障***如图3所示，其示出了本申请实施例所涉及的多智能主体避障***的结构示意图，具体来讲：
该多智能主体避障***可以包括一个或者一个以上处理核心的处理器301、一个或一个以上计算机可读存储介质的存储器302、电源303和输入单元304等部件。本领域技术人员可以理解，图3中示出的多智能主体避障***结构并不构成对多智能主体避障***的限定，可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置。其中：
处理器301是该多智能主体避障***的控制中心，利用各种接口和线路连接整个多智能主体避障***的各个部分，通过运行或执行存储在存储器302内的软件程序和/或模块，以及调用存储在存储器302内的数据，执行多智能主体避障***的各种功能和处理数据，从而对多智能主体避障***进行整体监控。可选的，处理器301可包括一个或多个处理核心；优选的，处理器301可集成应用处理器和调制解调处理器，其中，应用处理器主要处理操作***、用户界面和应用程序等，调制解调处理器主要处理无线通信。可以理解的是，上述调制解调处理器也可以不集成到处理器301中。
存储器302可用于存储软件程序以及模块，处理器301通过运行存储在存储器302的软件程序以及模块，从而执行各种功能应用以及数据处理。存储器302可主要包括存储程序区和存储数据区，其中，存储程序区可存储操作***、至少一个功能所需的应用程序(比如声音播放功能、图像播放功能等)等；存 储数据区可存储根据多智能主体避障***的使用所创建的数据等。此外，存储器302可以包括高速随机存取存储器，还可以包括非易失性存储器，例如至少一个磁盘存储器件、闪存器件、或其他易失性固态存储器件。相应地，存储器302还可以包括存储器控制器，以提供处理器301对存储器302的访问。
多智能主体避障***还包括给各个部件供电的电源303，可选地，电源303可以通过电源管理***与处理器301逻辑相连，从而通过电源管理***实现管理充电、放电、以及功耗管理等功能。电源303还可以包括一个或一个以上的直流或交流电源、再充电***、电源故障检测电路、电源转换器或者逆变器、电源状态指示器等任意组件。
该多智能主体避障***还可包括输入单元304，该输入单元304可用于接收输入的数字或字符信息，以及产生与用户设置以及功能控制有关的键盘、鼠标、操作杆、光学或者轨迹球信号输入。
尽管未示出，多智能主体避障***还可以包括显示单元等，在此不再赘述。具体在本实施例中，多智能主体避障***中的处理器301会按照如下的指令，将一个或一个以上的应用程序的进程对应的可执行文件加载到存储器302中，并由处理器301来运行存储在存储器302中的应用程序，从而实现各种功能，如下：生成栅格地图，以区分出多智能主体可行搬运路径与货架；初始化栅格地图和多智能主体；遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体；规划分配到搬运任务的智能主体规划搬运路径；将规划好的智能主体搬运路径分别分配 给对应的智能主体进行执行，以使智能主体按照所述搬运路径行驶；对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整；当智能主体完成一个搬运任务后，检索完成搬运任务的智能主体的任务列表是否为空，若不为空，则继续分配下一搬运任务。
以上个操作的具体实施例可参见前面的实施例，在此不再赘述。
由以上可知，遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体，规划分配到搬运任务的智能主体规划搬运路径，将规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使智能主体按照搬运路径行驶，对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整，使智能主体的搬运路径尽量均衡，减少了冲撞发生的概率，在路径规划中，加快了计算速度，同时在最优路径的选择中，考虑了智能主体转弯所带来的影响，更加贴合实际，设计了智能主体在线调整的方法，使智能主体更具实用性，更有灵活性。
本领域普通技术人员可以理解，上述实施例的各种方法中的全部或部分步骤可以通过指令来完成，或通过指令控制相关的硬件来完成，该指令可以存储于一计算机可读存储介质中，并由处理器进行加载和执行。
为此，本申请实施例提供一种计算机可读存储介质，其中存储有多条指令，该指令能够被处理器进行加载，以执行本申请实施例所提供的任一种多智能主 体避障方法中的步骤。例如，该指令可以执行如下步骤：生成栅格地图，以区分出多智能主体可行搬运路径与货架；初始化栅格地图和多智能主体；遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体；规划分配到搬运任务的智能主体规划搬运路径；将规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使智能主体按照所述搬运路径行驶；对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整；当智能主体完成一个搬运任务后，检索完成搬运任务的智能主体的任务列表是否为空，若不为空，则继续分配下一搬运任务。
以上各个操作的具体实施方式可参见前面的实施例，在此不再赘述。
其中，该计算机可读存储介质可以包括：只读存储器(ROM，Read Only Memory)、随机存取记忆体(RAM，Random Access Memory)、磁盘或光盘等。
由于该计算机可读存储介质中所存储的指令，可以执行本申请实施例所提供的任一种多智能主体避障方法中的步骤，因此，可以实现本申请实施例所提供的任一种多智能主体避障方法所能实现的有益效果，详见前面的实施例，在此不再赘述。
以上对本申请实施例所提供的一种多智能主体避障方法、设备和计算机可读存储介质进行了详细介绍，本文中应用了具体个例对本申请的原理及实施方式进行了阐述，以上实施例的说明只是用于帮助理解本申请的方法及其核心思 想；同时，对于本领域的技术人员，依据本申请的思想，在具体实施方式及应用范围上均会有改变之处，综上，本说明书内容不应理解为对本申请的限制。

Claims (10)

  1. 一种多智能主体避障方法，其特征在于，所述方法包括：
    生成栅格地图，以区分出多智能主体可行搬运路径与货架；
    初始化所述栅格地图和多智能主体；
    遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体；
    规划所述分配到搬运任务的智能主体的搬运路径；
    将所述规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使所述智能主体按照所述搬运路径行驶；
    对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整；
    当智能主体完成一个搬运任务后，检索完成搬运任务的智能主体的任务列表是否为空，若不为空，则继续分配下一搬运任务。
  2. 根据权利要求1所述的多智能主体避障方法，其特征在于，所述遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体，包括：
    遍历所述所有待搬运任务和多智能主体的当前属性，以获取任务列表的长度、每个智能主体距离目标搬运任务的距离以及综合作业的距离；
    将所述目标搬运任务优先分配给空闲智能主体，其次分配给任务列表中任 务最少的智能主体，或者，将所述目标搬运任务优先分配给距离所述目标搬运任务最近的智能主体，或者，将所述目标搬运任务优先一组存取任务之间的行程距离最近的智能主体。
  3. 根据权利要求1所述多智能主体避障方法，其特征在于，所述规划所述分配到搬运任务的智能主体的搬运路径，包括：
    根据任务的优先级进行智能主体搬运路径的规划，优先规划承担较高优先级搬运任务的智能主体的搬运路径和/或在多智能主体在所述搬运路径上拥堵时，按照所承担搬运任务优先级越高，给予越高的优先通行权的原则规划所述智能主体的搬运路径；
    按照所述规划好的搬运路径，训练每多智能主体的搬运模型；
    按照所述已训练搬运模型，对所有智能主体的搬运路径之间的干扰碰撞情况进行分析并调整部分智能主体的的搬运路径。
  4. 根据权利要求3所述多智能主体避障方法，其特征在于，所述搬运任务的优先级的高低根据提货单的重要程度和/或根据提货单的到达时间来判断。
  5. 根据权利要求3所述多智能主体避障方法，其特征在于，所述按照所述规划好的搬运路径，训练每多智能主体的搬运模型，包括：
    S1、计算所述栅格地图中每一个栅格至目标栅格的距离；
    S2、在智能主体A i当前所在起始栅格四周选择距离目标栅格的成本函数最小的一个栅格作为邻居栅格并存储于路由r_te中；
    S3、重复执行上述步骤S2，直至所述智能主体A i最后到达所述目标栅格；
    S4、重复执行步骤S1至步骤S3共y次，生成长度最短和转弯次数最少的路径作为所述智能主体的最终路径；
    S5，对每个所述智能主体的最终路径进行加权处理，以形成冲撞最小的多智能主体的搬运模型。
  6. 根据权利要求3所述多智能主体避障方法，其特征在于，所述按照所述已训练搬运模型，对所有智能主体的搬运路径之间的干扰碰撞情况进行分析并调整部分智能主体的的搬运路径，包括：
    保证优先级最高的智能主体的搬运路径不动，比较优先级第二的智能主体的搬运路径与其是否存在冲撞的节点，若存在，则针对冲突类型采取不同的避让策略；
    按照上述方法，依次对所有智能主体按照优先级高低进行路径的冲撞判断与优化，直至所有智能主体的路径规划完毕且互相之间没有冲撞为止。
  7. 根据权利要求1所述多智能主体避障方法，其特征在于，所述位置异常的智能主体包括：运行速度过快的智能主体和运行速度过慢的智能主体。
  8. 一种多智能主体避障***，其特征在于，所述***包括：
    地图生成模块，用于生成栅格地图，以区分出多智能主体可行搬运路径与货架；
    初始化模块，用于初始化所述栅格地图和多智能主体；
    任务分配模块，用于遍历所有搬运任务和多智能主体，按照搬运任务分配规则将搬运任务分配给对应的多智能主体中的一个或多个智能主体；
    路径规划模块，用于规划所述分配到搬运任务的智能主体的搬运路径；
    路径分配模块，用于将所述规划好的智能主体搬运路径分别分配给对应的智能主体进行执行，以使所述智能主体按照所述搬运路径行驶；
    检测模块，用于对所有执行搬运任务的智能主体进行实时监控，检测各智能主体的实时位置是否正常，以对位置异常的智能主体进行实时在线调整；
    检索模块，用于当智能主体完成一个搬运任务后，检索完成搬运任务的智能主体的任务列表是否为空，若不为空，则继续分配下一搬运任务。
  9. 一种多智能主体避障***，所述***包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机程序，所述计算机程序代码由该一个或多个处理器加载并执行以实现如权利要求1至7任意一项所述方法的步骤。
  10. 一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，其特征在于，所述计算机程序被处理器执行时实现如权利要求1至7任意一项所述方法的步骤。
PCT/CN2020/108245 2020-08-10 2020-08-10 一种多智能主体避障方法、***和计算机可读存储介质 WO2022032444A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/108245 WO2022032444A1 (zh) 2020-08-10 2020-08-10 一种多智能主体避障方法、***和计算机可读存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/108245 WO2022032444A1 (zh) 2020-08-10 2020-08-10 一种多智能主体避障方法、***和计算机可读存储介质

Publications (1)

Publication Number Publication Date
WO2022032444A1 true WO2022032444A1 (zh) 2022-02-17

Family

ID=80247500

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/108245 WO2022032444A1 (zh) 2020-08-10 2020-08-10 一种多智能主体避障方法、***和计算机可读存储介质

Country Status (1)

Country Link
WO (1) WO2022032444A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115291534A (zh) * 2022-10-09 2022-11-04 华东交通大学 智能体家具协同运动控制方法、***和计算机终端
CN117495223A (zh) * 2023-12-29 2024-02-02 江西丹巴赫机器人股份有限公司 多台两向穿梭车同区域作业的仓储物流优化方法

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150139762A1 (en) * 2013-11-18 2015-05-21 Park Plus Inc. Hall effect sensor grid array guidance system
CN108897316A (zh) * 2018-06-14 2018-11-27 北京航空航天大学 一种基于信息素导航的集群仓储机器人***控制方法
CN109765915A (zh) * 2019-03-13 2019-05-17 中南大学 一种基于计算机算法的仓储搬运机器人任务调度***
CN109839935A (zh) * 2019-02-28 2019-06-04 华东师范大学 多agv的路径规划方法及设备
CN109947120A (zh) * 2019-04-29 2019-06-28 西安电子科技大学 仓储***中的路径规划方法
CN110009259A (zh) * 2019-04-18 2019-07-12 天津工业大学 一种应用于双向路径下柔性制造车间的多agv调度方法
CN110264120A (zh) * 2019-05-06 2019-09-20 盐城品迅智能科技服务有限公司 一种基于多agv的智能仓储路线规划***和方法
CN110989570A (zh) * 2019-10-15 2020-04-10 浙江工业大学 一种多agv防碰撞协同路径规划方法
CN111487962A (zh) * 2020-03-30 2020-08-04 北京化工大学 一种应用于仓储环境下多机器人路径快速规划方法

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150139762A1 (en) * 2013-11-18 2015-05-21 Park Plus Inc. Hall effect sensor grid array guidance system
CN108897316A (zh) * 2018-06-14 2018-11-27 北京航空航天大学 一种基于信息素导航的集群仓储机器人***控制方法
CN109839935A (zh) * 2019-02-28 2019-06-04 华东师范大学 多agv的路径规划方法及设备
CN109765915A (zh) * 2019-03-13 2019-05-17 中南大学 一种基于计算机算法的仓储搬运机器人任务调度***
CN110009259A (zh) * 2019-04-18 2019-07-12 天津工业大学 一种应用于双向路径下柔性制造车间的多agv调度方法
CN109947120A (zh) * 2019-04-29 2019-06-28 西安电子科技大学 仓储***中的路径规划方法
CN110264120A (zh) * 2019-05-06 2019-09-20 盐城品迅智能科技服务有限公司 一种基于多agv的智能仓储路线规划***和方法
CN110989570A (zh) * 2019-10-15 2020-04-10 浙江工业大学 一种多agv防碰撞协同路径规划方法
CN111487962A (zh) * 2020-03-30 2020-08-04 北京化工大学 一种应用于仓储环境下多机器人路径快速规划方法

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115291534A (zh) * 2022-10-09 2022-11-04 华东交通大学 智能体家具协同运动控制方法、***和计算机终端
CN115291534B (zh) * 2022-10-09 2023-01-24 华东交通大学 智能体家具协同运动控制方法、***和计算机终端
CN117495223A (zh) * 2023-12-29 2024-02-02 江西丹巴赫机器人股份有限公司 多台两向穿梭车同区域作业的仓储物流优化方法

Similar Documents

Publication Publication Date Title
CN110989570B (zh) 一种多agv防碰撞协同路径规划方法
CN114519448A (zh) 用于在操作环境中优化路线规划的***和方法
WO2017118001A1 (zh) 用于机器人场地回流的方法和装置
CN109991977A (zh) 机器人的路径规划方法及装置
WO2022032444A1 (zh) 一种多智能主体避障方法、***和计算机可读存储介质
CN113075927A (zh) 基于预约表的仓储潜伏式多agv路径规划方法
Guney et al. Dynamic prioritized motion coordination of multi-AGV systems
Lienert et al. No More Deadlocks-Applying The Time Window Routing Method To Shuttle Systems.
CN114489062A (zh) 面向车间物流的多自动导引车分布式动态路径规划方法
TW202321652A (zh) 自動導引車調度方法、電子裝置及電腦可讀儲存媒體
Li et al. Simulation analysis of a deep reinforcement learning approach for task selection by autonomous material handling vehicles
Shi et al. Task allocation and path planning of many robots with motion uncertainty in a warehouse environment
Zhou et al. Anisotropic Q-learning and waiting estimation based real-time routing for automated guided vehicles at container terminals
Solichudin et al. Conflict-free dynamic route multi-agv using dijkstra Floyd-warshall hybrid algorithm with time windows
Basile et al. An auction-based approach to control automated warehouses using smart vehicles
Klaas et al. Simulation aided, knowledge based routing for AGVs in a distribution warehouse
Liu et al. Simultaneous planning and scheduling for multi-autonomous vehicles
WO2024051507A1 (zh) 多机器人路径规划方法、装置及计算设备
Ali et al. Prioritized SIPP for multi-agent path finding with kinematic constraints
Xia et al. A multi-AGV optimal scheduling algorithm based on particle swarm optimization
CN115638804B (zh) 一种无死锁的无人车辆在线路径规划方法
CN112987713A (zh) 自动驾驶设备的控制方法、装置及存储介质
Chen et al. Deadlock-solving traffic control methods for automated guided vehicle systems
Zhang et al. Multi-AGVs pathfinding based on improved jump point search in logistic center
Xing et al. Collision avoidance and give way of heterogeneous and variable-sized multiple mobile robots based on glued nodes

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20948943

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20948943

Country of ref document: EP

Kind code of ref document: A1