Path planning method

Abstract

The invention discloses a path planning method, and the method comprises the steps: determining a first moving object which meets a path conflict condition based on the initial motion paths of a plurality of moving objects; wherein the first moving object is at least a part of objects in the plurality of moving objects; based on the fault information of each first moving object, determining a collision probability of collision between the first moving objects when at least one first moving object has a fault on a conflict path corresponding to the path conflict condition; determining target motion paths of the plurality of moving objects based on the collision probability and the initial motion paths of the plurality of moving objects; wherein the collision probability obtained based on the target motion path does not meet the collision condition.

Description

Technical Field

[0001] The present application relates to, but is not limited to, the field of path planning technology, and in particular to a path planning method. Background Art

[0002] Multi-robot path planning is an important and complex task in modern intelligent systems. It requires multiple robots to move from a starting position to a target position in a known complex environment while avoiding collisions with other robots. With the rise of applications such as autonomous vehicles, logistics robots, and collaborative drones, multi-robot path planning has become a research hotspot in recent years.

[0003] Related technologies typically use heuristic search algorithms or swarm intelligence algorithms to construct path planning frameworks, optimizing path node search strategies and dynamic collision avoidance mechanisms to achieve multi-robot coordinated scheduling. However, in practical applications, due to factors such as the uncertainty of the robots' dynamic behavior, the accuracy of multi-robot paths planned using this approach is low. Therefore, there is an urgent need for a new path planning method. Summary of the Invention

[0004] The embodiments of the present application hope to provide a path planning method.

[0005] The technical solution of this application is achieved as follows:

[0006] In a first aspect, an embodiment of the present application provides a path planning method, the method comprising:

[0007] Determining a first moving object that meets a path conflict condition based on initial motion paths of the plurality of moving objects; wherein the first moving object is at least a portion of the plurality of moving objects;

[0008] determining, based on the fault information of each first moving object, a collision probability of collision between the first moving objects when at least one first moving object fails on a conflicting path corresponding to the path conflict condition;

[0009] Target motion paths of the multiple moving objects are determined based on the collision probability and initial motion paths of the multiple moving objects; wherein the collision probability obtained based on the target motion path does not meet the collision condition.

[0010] In a second aspect, an embodiment of the present application provides a path planning device, the device comprising:

[0011] A first determining module is configured to determine, based on initial motion paths of the plurality of moving objects, a first moving object that satisfies a path conflict condition; wherein the first moving object is at least a portion of the plurality of moving objects;

[0012] a second determining module, configured to determine, based on the fault information of each first moving object, a collision probability of collision between the first moving objects when at least one first moving object fails on a conflicting path corresponding to the path conflict condition;

[0013] The third determining module is configured to determine target motion paths of the multiple moving objects based on the collision probability and the initial motion paths of the multiple moving objects; wherein the collision probability obtained based on the target motion path does not satisfy the collision condition.

[0014] In a third aspect, an embodiment of the present application provides an electronic device, the electronic device comprising: a processor and a memory;

[0015] The memory stores a computer program executable on the processor;

[0016] The processor executes the computer program stored in the memory to implement the steps of the above-mentioned path planning method.

[0017] In a fourth aspect, an embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by at least one processor, the steps of the above-mentioned path planning method are implemented.

[0018] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program or instructions, which, when executed by a processor, implements the steps of the above-mentioned path planning method.

[0019] An embodiment of the present application provides a path planning method, which determines, based on the initial motion paths of multiple mobile objects, a first mobile object that meets a path conflict condition; wherein the first mobile object is at least part of the multiple mobile objects; based on fault information of each first mobile object, determines a collision probability of collision between the first mobile objects when at least one first mobile object fails on a conflict path corresponding to the path conflict condition; based on the collision probability and the initial motion paths of the multiple mobile objects, determines a target motion path of the multiple mobile objects; wherein the collision probability obtained based on the target motion path does not meet the collision condition. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic flow chart of an optional path planning method provided in an embodiment of the present application;

[0021] Figure 2 A schematic diagram of scenarios of different conflict types corresponding to an optional path conflict condition provided in an embodiment of the present application;

[0022] Figure 3 A schematic diagram of an optional path planning scenario provided for an embodiment of the present application;

[0023] Figure 4 A schematic flow chart of an optional path planning method provided in an embodiment of the present application;

[0024] Figure 5 A schematic flow chart of an optional path planning method provided in an embodiment of the present application;

[0025] Figure 6 A schematic flow chart of an optional path planning method provided in an embodiment of the present application;

[0026] Figure 7 A schematic flow chart of an optional path planning method provided in an embodiment of the present application;

[0027] Figure 8 A schematic diagram of an optional scenario after adjusting the planned path provided in an embodiment of the present application;

[0028] Figure 9 A schematic flow chart of an optional path planning method provided in an embodiment of the present application;

[0029] Figure 10 A schematic flow chart of an optional path planning method provided in an embodiment of the present application;

[0030] Figure 11 A schematic flow chart of an optional path planning method provided in an embodiment of the present application;

[0031] Figure 12 A flowchart of an optional LSTM model training process provided in an embodiment of the present application;

[0032] Figure 13 A schematic structural diagram of an optional path planning device provided in an embodiment of the present application;

[0033] Figure 14 A schematic structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0035] The terms "first," "second," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish between different objects, not to describe a particular order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0036] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0037] The embodiment of the present application provides a path planning method, which is applied to electronic equipment, referring to Figure 1 As shown, the method includes the following steps:

[0038] Step 101: Determine a first moving object that meets a path conflict condition based on initial motion paths of multiple moving objects.

[0039] The first moving object is at least a part of the multiple moving objects.

[0040] In an embodiment of the present application, the mobile object may be a device or object that can move autonomously in an automated transportation scenario. The mobile object may be an automated guided vehicle (AGV), an autonomous mobile robot (AMR), an intelligent warehouse robot, a logistics robot, and other mobile robots used for various purposes; of course, the mobile object may also be other mobile devices or objects such as forklifts, self-driving vehicles, and drones.

[0041] In an embodiment of the present application, the initial motion path may be the initial driving route planned by the mobile object before it starts moving, based on its own mission objectives, the current environmental scene, etc. The initial motion path may be obtained by planning the path of each mobile object in the current environmental scene using an existing path planning algorithm. It should be noted that the initial motion paths between multiple mobile objects satisfy single path planning and collaborative collision avoidance. Among them, the existing path planning algorithm may be a heuristic search algorithm, such as the Dijkstra algorithm, the A* search algorithm, and the conflict-based search (CBS) algorithm; the path planning algorithm may also be a swarm intelligence algorithm, such as the ant colony algorithm, and this application does not impose any specific restrictions on this.

[0042] In an embodiment of the present application, the path conflict condition is used to determine whether there are intersections and / or overlapping paths between the initial motion paths of the moving objects; in some embodiments, the path conflict condition includes one or more of the following:

[0043] The first condition includes: there is an intersection between the motion paths of multiple moving objects;

[0044] The second condition includes: the motion paths of the multiple moving objects have overlapping paths, and the moving directions of the multiple moving objects on the overlapping paths are the same;

[0045] The third condition includes: there is an overlapping path among the motion paths of the multiple moving objects, and the moving directions of the multiple moving objects on the overlapping paths are opposite.

[0046] In the embodiment of the present application, the first moving objects that meet the path conflict condition may be one group or multiple groups, and the first moving objects included in each group are analyzed according to the subsequent steps.

[0047] In one possible scenario, refer to Figure 2 As shown, there are intersections in the motion paths of multiple moving objects, such as Figure 2 As shown in the left figure, the motion paths of moving objects 1 and 2 have an intersection C. The motion paths of multiple moving objects have overlapping paths, and the moving directions of multiple moving objects on the overlapping paths are the same, as shown in Figure 2. Figure 2 As shown in the middle figure, the motion paths of moving objects 1 and 2 have overlapping paths CD, and the moving directions of moving objects 1 and 2 on the overlapping paths are the same. The motion paths of multiple moving objects have overlapping paths, and the moving directions of multiple moving objects on the overlapping paths are opposite, as shown in the figure. Figure 2 As shown in the right figure in , there is an overlapping path CD between the motion paths of the moving object 1 and the moving object 2, and the moving directions of the moving object 1 and the moving object 2 on the overlapping path are opposite.

[0048] In the embodiments of the present application, the first moving objects may be some or all of the objects selected from a plurality of moving objects that meet the path conflict condition. It should be noted that there may be path conflicts between the initial motion paths of the plurality of first moving objects. Therefore, there may be situations where the initial motion paths of the first moving objects need to be adjusted or replanned to avoid problems such as collisions with other first moving objects when at least one first moving object malfunctions.

[0049] In practical applications, the electronic device may be a hardware device with data storage, processing and analysis functions.

[0050] In an embodiment of the present application, when an electronic device performs path planning for multiple mobile objects in a preset scenario, it first obtains an initial motion path planned for each of the multiple mobile objects. Then, based on a preset path conflict condition, the initial motion paths of the multiple mobile objects are checked and judged. By comparing the distribution of the initial motion paths of the various mobile objects in time and / or space, the mobile object corresponding to the initial motion path with an intersection and / or overlapping path is determined, and it is determined as the first mobile object, thereby obtaining the first mobile object that meets the path conflict condition. In this way, by analyzing the initial motion paths of multiple mobile objects, the first mobile object that meets the path conflict condition is accurately located, blind planning is avoided, and targeted conflict detection can quickly lock in the key problem objects of path planning, reduce planning redundancy, and improve the efficiency and accuracy of path planning.

[0051] In a feasible scenario, refer to Figure 3 As shown, based on Figure 3 The environment scene constructed by the black graphics in Figure 3 As shown in the left figure in , in this environmental scenario, there are three moving objects G1, G2, and G3, among which the priority between the moving objects G1, G2, and G3 is G1>G2>G3, and the current positions of the moving objects G1, G2, and G3 are the starting positions of each moving object (also called the starting movement points), and the ending positions (also called the ending movement points) corresponding to the moving objects G1, G2, and G3 are D1, D2, and D3 respectively.

[0052] First, the electronic device uses an existing path planning algorithm such as the ant colony algorithm to enter the path planning stage. It performs path planning based on the starting and ending positions of the three moving objects, and obtains three motion paths G1-F1, G2-F2, and G3-F3 corresponding to the three moving objects G1, G2, and G3, respectively. Figure 3As shown in the middle figure in

[15] , the existing intersections and overlapping paths are detected. Since the starting positions of the moving objects G1 and G3 in the movement paths G1-F1 and G3-F3 are 21 and 15 meters (meters, m) away from the intersection M1 respectively, and the starting positions of the moving objects G2 and G3 in the movement paths G2-F2 and G3-F3 are 9 and 7 meters away from the first intersection M2 respectively, it can be seen that no collision will occur after starting at the same time, as shown in

[15] : Figure 3 As shown in the right figure in , the three initial motion paths G1-F1, G2-F2, and G3-F3 corresponding to the three moving objects G1, G2, and G3 are obtained respectively.

[0053] Furthermore, entering the collaborative collision avoidance stage, since there is an intersection M1 between the initial motion paths G1-F1 and G3-F3, a group of first moving objects G1 and G2 that meet the path conflict conditions are obtained for the initial motion paths G1-F1 and G3-F3; and since there is an overlapping path M2-M3 between the initial motion paths G2-F2 and G3-F3, another group of first moving objects G1 and G2 that meet the path conflict conditions are obtained for the initial motion paths G2-F2 and G3-F3. Subsequently, each group of first moving objects that meet the path conflict conditions is analyzed separately.

[0054] Step 102 : Based on the fault information of each first moving object, determine a collision probability of collision between the first moving objects when at least one first moving object fails on a conflicting path corresponding to a path conflict condition.

[0055] In this embodiment of the present application, the fault information may be the failure rate of a mobile object at each time point during its movement or motion over a future period. That is, the fault information includes a set of failure rates for the mobile objects, with the failure rates for the same mobile object at different time points being the same or different. Using the failure rates of the mobile objects is crucial for analyzing the probability of collisions between them in the event of a path conflict.

[0056] In the embodiment of the present application, the failure of the mobile object may be any situation that may affect its normal movement along the planned path, including but not limited to failure of the robot's power system, sensor failure, communication interruption, etc.

[0057] In the embodiment of the present application, the conflicting paths corresponding to the path conflict conditions may be intersections or overlapping paths. It should be noted that the moving directions of the moving objects on the overlapping paths may be in the same direction or in opposite directions.

[0058] In this embodiment of the present application, the collision probability measures the likelihood of a collision between first moving objects if at least one first moving object fails on a conflicting path corresponding to a path conflict condition. The collision probability is a probability value between 0 and 1, where 0 indicates that a collision is unlikely and 1 indicates that a collision is certain to occur.

[0059] Understandably, related technologies typically employ heuristic search algorithms or swarm intelligence algorithms to construct path planning frameworks, achieving multi-robot collaborative scheduling by optimizing path node search strategies and dynamic collision avoidance mechanisms. However, as the number of robots increases, the probability of robot failure also increases. Traditional path planning models often only perform collision detection for instantaneous conflicts, lacking dynamic prediction of the robot's motion state and fault-tolerant design. This results in significant position prediction errors during the path planning process, making it difficult to output a global optimal solution in complex dynamic scenarios. This results in low accuracy in the multi-robot planned paths, which in turn affects the overall operational efficiency and safety of the multi-robot system.

[0060] To address the above issues, embodiments of the present application obtain fault information for each first moving object at various time points while moving along the initial motion paths, targeting first moving objects with conflicting paths. Based on the fault information for each first moving object, the system considers the likelihood of a collision between the first moving objects, i.e., the collision probability, in a specific scenario, when at least one first moving object fails on the conflicting path. Thus, based on the fault information of each first moving object, the collision probability is quantified when a fault occurs on the conflicting path, enabling dynamic tracking and risk prediction of the operating status of multiple moving objects. Even when there are a large number of moving objects, such as robots, and the probability of failure increases, potential collision risks can be estimated in advance, providing a more reliable path planning solution for the system, enhancing the robustness of the multi-moving object system in the face of equipment failures, and reducing the possibility of collision accidents caused by failures. Furthermore, by integrating fault information with path conflict analysis, a globally optimal path that balances safety and efficiency can be output, thereby reducing collision accidents caused by path planning deviations, improving path planning accuracy and the safety of multiple moving objects, and effectively enhancing the overall efficiency of multi-moving object collaborative operations.

[0061] For example, continue with Figure 3 Taking the corresponding scenario as an example, the electronic device can obtain the corresponding failure rates of mobile objects G1, G2, and G3 at various time points in the future period, where U1(t) is the failure rate of mobile object G1, U2(t) is the failure rate of mobile object G2, and U3(t) is the failure rate of mobile object G3. The failure rates of mobile objects G1, G2, and G3 at various time points in the next 60 seconds are shown in Table 1.

[0062] Time point 1s 2s 3s … 60s <![CDATA[U1(t) / U2(t) / U3(t)]]> 0.012 / 0.01 / 0.002 0.06 / 0.006 / 0.001 0.05 / 0.002 / 0.005 … 0.11 / 0.018 / 0.1

[0063] Table 1

[0064] Step 103: Determine target motion paths of the multiple moving objects based on the collision probability and the initial motion paths of the multiple moving objects.

[0065] The collision probability obtained based on the target motion path does not meet the collision condition.

[0066] In an embodiment of the present application, the collision condition may be that the collision probability is greater than or equal to a collision probability threshold, and the collision probability threshold may be 0.2. Of course, it may also be other values, and the present application does not impose any specific restrictions on this.

[0067] In an embodiment of the present application, the target motion path refers to a path obtained after adjustment and planning based on the collision probability and the initial motion paths of multiple moving objects, which enables multiple moving objects to avoid mutual collision during the movement and smoothly reach the target position from their respective starting points. In this way, taking into account the failure rate of the moving objects, the safe and efficient operation of multiple moving objects in complex scenarios is met.

[0068] In the embodiment of the present application, the target motion path may be the initial motion path of the moving object, or the target motion path may be a path obtained by adjusting and planning the initial motion path of the moving object.

[0069] In the embodiment of the present application, the fact that the collision probability obtained based on the target motion path does not satisfy the collision condition can be understood as follows: for the target motion path of the mobile object, based on the fault information of each mobile object, when it is determined that at least one mobile object has a fault on the conflict path corresponding to the path conflict condition, the collision probability of collision between the mobile objects is less than the collision probability condition. Of course, the fact that the collision probability obtained based on the target motion path does not satisfy the collision condition can also be understood as follows: for the target motion path of the first mobile object that meets the path conflict condition, based on the fault information of each first mobile object, when it is determined that at least one mobile object has a fault on the conflict path corresponding to the path conflict condition, the collision probability of collision between the first mobile objects is less than the collision probability condition; it should be noted that the initial motion paths of the remaining mobile objects except the first mobile object that meets the path conflict condition among the multiple mobile objects can be used as their target motion paths.

[0070] In an embodiment of the present application, based on the situation where multiple first moving objects move to a conflict path, when at least one first moving object fails, the collision probability of collision between the first moving objects, and the initial motion path of each moving object, the collision probability and the initial motion path of each moving object are analyzed to determine whether the motion path of each moving object should be replanned to obtain a target motion path. It should be noted that the target motion path is generated to meet the needs of safe and efficient operation of the multi-moving object system. It is a target motion path obtained by improving and perfecting the initial motion path of the multi-moving object. For the target motion path, when at least one moving object fails on the conflict path corresponding to the path conflict condition, the collision probability of collision between the multiple moving objects does not meet the collision condition. In this way, the target motion path is generated by combining the collision probability and the initial motion path, so that the path planning has dynamic adjustment capabilities. In this way, even if the motion state of the moving object itself changes (such as stopping movement due to a fault), the path can be optimized in time to ensure that the collision probability calculated based on the target motion path always does not meet the collision condition, so that the multi-moving object system can flexibly respond to complex and changeable scenarios such as autonomous driving, logistics transportation, and drone collaboration, maintain efficient and stable operation, and improve the overall performance of the system.

[0071] In some embodiments, step 102 is a process of determining the collision probability between the first moving objects when at least one first moving object fails on the conflict path corresponding to the path conflict condition based on the fault information of each first moving object. Figure 4 Provide explanation.

[0072] Step 401: Determine reference conflict points corresponding to each first moving object.

[0073] The reference conflict point is determined based on the path conflict condition.

[0074] In the embodiment of the present application, the reference conflict point can be an intersection point, or one of the endpoints in the overlapping path, wherein the endpoint is determined by obtaining the movement duration of each first moving object when it moves to the nearest endpoint in the overlapping path, and taking the endpoint closest to the starting point of the first moving object with the shortest movement duration in the overlapping path as the endpoint. Figure 2If the path conflict condition is the first condition, the reference conflict point is the intersection point, such as point C. If the path conflict condition is the second condition, the reference conflict point is an endpoint in the overlapping path closest to the starting point of each first moving object, such as point C. If the path conflict condition is the third condition, the movement duration of each first moving object when it moves to the closest endpoint in the overlapping path is obtained, and the endpoint in the overlapping path closest to the starting point of the first moving object with the shortest movement duration is set as that endpoint, such as point C. This application does not impose any specific restrictions on this.

[0075] In the embodiment of the present application, the electronic device determines a first moving object that meets a path conflict condition based on the initial motion paths of multiple moving objects. Then, based on the conflicting paths corresponding to the path conflict conditions, the electronic device determines a reference conflict point for each first moving object. Determining the reference conflict point for each first moving object accurately locates the key locations of the path conflict, allowing subsequent analysis to focus on areas where collisions are likely to occur, avoiding a broad analysis of the entire path and improving analysis efficiency and accuracy.

[0076] Step 402: Obtain a reference motion duration of each first moving object from a starting motion point of a corresponding initial motion path to a reference conflict point.

[0077] In an embodiment of the present application, after the electronic device determines the reference conflict point of each first moving object, it obtains the preset moving speed of each first moving object, and determines the first moving distance of each first moving object based on the starting moving point and the reference conflict point of the initial motion path of each first moving object; further, based on the first moving distance and the corresponding moving speed of each first moving object, the reference movement time required for each first moving object to move from the corresponding starting moving point to the reference conflict point is obtained. In this way, the reference movement time of each first moving object from the starting moving point to the reference conflict point is obtained, which provides a quantitative basis for evaluating the time distribution of the moving object near the conflict point. Subsequently, combined with the fault information, the possibility of the moving object failing and causing a collision at different time points can be more accurately analyzed.

[0078] Step 403 : Determine, based on the reference running time and fault information of each first mobile object, a sub-collision probability of a collision between any two first mobile objects when at least one first mobile object fails on the conflicting path.

[0079] In an embodiment of the present application, the electronic device obtains a reference motion duration of each first moving object from the starting motion point of the corresponding initial motion path to the reference conflict point, and obtains the predicted target motion duration of each first moving object, and the fault information corresponding to each time point within the target motion duration. Based on the reference motion duration and the fault information of each first moving object, the electronic device determines the sub-collision probability of a collision between any two first moving objects when at least one first moving object collides on the conflict path. Here, the target motion duration can be a preset duration, and the target duration can be greater than the reference motion duration.

[0080] By considering the different speeds and initial position determination times of different first moving objects, as well as the predicted fault information corresponding to each first moving object's reference movement time, the probability of collision between them can be more accurately assessed, avoiding overly general estimates. In a multi-moving object system, the interactions between multiple objects are complex. By calculating the sub-collision probability between any two first moving objects, we can meticulously analyze the potential collision relationships between each object in the system, contributing to a deeper understanding of the overall system's operating mechanisms and providing strong support for further system optimization.

[0081] Step 404: Determine the collision probability between the first moving objects based on the multiple sub-collision probabilities.

[0082] In an embodiment of the present application, the electronic device obtains a sub-collision probability of a collision between any two first moving objects, thereby obtaining multiple sub-collision probabilities. Based on the multiple sub-collision probabilities, the electronic device uses the inclusion-exclusion theorem to determine the collision probability of a collision between some or all of the first moving objects. In this way, by calculating the sub-collision probability of a collision between any two first moving objects and then determining the collision probability between the first moving objects based on the multiple sub-collision probabilities, this hierarchical calculation method can comprehensively and meticulously consider the complex interactions in a multi-moving object system, thereby more accurately assessing the overall collision risk.

[0083] Here, based on multiple sub-collision probabilities, the collision probability between the first moving objects is determined using the inclusion-exclusion theorem, which can be expressed by the following formula (1).

[0084]

[0085] Wherein, N is the number of first moving objects; and P is the collision probability of collision among the N first moving objects.

[0086] Among them, the sub-collision probability of collision between any two first moving objects among N first moving objects can be respectively Among them, P iis the sub-collision probability of collision between any two first moving objects of the i-th order, and the value range of i is i is an integer; P j The sub-collision probability of a collision between any two first moving objects of the jth order is j is an integer; P l The sub-collision probability of a collision between any two first moving objects of the lth order is l. The value range of l is l is an integer; the following will be deduced by analogy and no further examples will be given.

[0087] In some embodiments, step 403 combines the process of determining the sub-collision probability of collision between any two first mobile objects when at least one first mobile object fails on the conflicting path based on the reference running time and fault information of each first mobile object. Figure 5 Provide explanation.

[0088] Step 501: If a second mobile object fails on a conflicting path, determine a failure stop point of the second mobile object.

[0089] The second moving object is a moving object corresponding to a first reference movement duration in any two reference movement durations among the multiple reference movement durations, and the fault stagnation point is a point on the conflict path.

[0090] In the embodiment of the present application, the fault stagnation point may be a point on the conflict path. For example, if the conflict path is an intersection, the fault stagnation point may be an intersection, such as Figure 2 Point C in the left figure; if the conflicting path is an overlapping path, the failure stagnation point can be any point on the overlapping path, such as Figure 2 Any point k on the CD segment in the middle figure and the right figure.

[0091] In the embodiment of the present application, the second moving object is one of the moving objects between any two first moving objects, and the reference motion duration of the second moving object is the shorter reference motion duration of the reference motion durations corresponding to the two first moving objects.

[0092] In the embodiment of the present application, the first reference exercise duration is the shorter of the two reference exercise durations.

[0093] In an embodiment of the present application, during the movement of multiple first moving objects on corresponding initial motion paths, a reference motion duration of each first moving object from the starting motion point of the corresponding initial motion path to the reference conflict point is obtained, thereby obtaining multiple reference motion durations; then, a shorter reference motion duration is determined from any two reference motion durations among the multiple reference motion durations, that is, the first moving object corresponding to the first reference motion duration, and it is used as the second moving object; assuming that the second moving object fails during its movement on the conflict path, the failure stagnation point of the second moving object is obtained.

[0094] Step 502: Obtain the actual movement duration of the second moving object moving to the fault stagnation point.

[0095] In an embodiment of the present application, after the electronic device determines the fault stagnation point of the second moving object, it obtains the preset movement speed of the second moving object and determines the second movement distance of the second moving object based on the starting point of the second moving object's initial movement path and the fault stagnation point. Furthermore, based on the second movement distance and the corresponding movement speed of the second moving object, the actual movement time required for the second moving object to move from the corresponding starting point to the fault stagnation point is obtained. In this way, if the second moving object fails on the conflicting path, determining its fault stagnation point and obtaining the actual movement time allows for detailed analysis of the object's precise position and movement state at the time of the failure.

[0096] Step 503: Based on the conflict type of the conflict path, the reference movement duration, actual movement duration and fault information of the second mobile object, and the reference movement duration and fault information of the third mobile object, determine the sub-collision probability of a collision between the second mobile object and the third mobile object when the second mobile object fails on the conflict path.

[0097] The third moving object is the moving object corresponding to the second reference movement duration of any two reference movement durations.

[0098] In the embodiment of the present application, the third moving object is one of the moving objects between any two first moving objects, and the reference motion duration of the third moving object is the longer reference motion duration of the reference motion durations corresponding to the two first moving objects.

[0099] In the embodiment of the present application, the conflict types of the conflicting paths include intersection conflicts and overlapping path conflicts.

[0100] In the embodiment of the present application, the second reference exercise duration is the longer exercise duration of the two reference exercise durations.

[0101] In an embodiment of the present application, when obtaining the reference movement duration, actual movement duration, and predicted target movement duration of the second moving object, the fault information corresponding to each time point within the target duration is obtained; when obtaining the reference movement duration of the third moving object, and predicted target movement duration of the third moving object, the fault information corresponding to each time point within the target duration is obtained. Furthermore, based on the reference movement duration, actual movement duration, and fault information of the second moving object, as well as the reference movement duration and fault information of the third moving object, a collision probability calculation model corresponding to the conflict type of the conflict path is used to calculate the sub-collision probability of a collision between the second moving object and the third moving object when the second moving object fails on the conflict path. Here, the target duration can be a pre-set duration, and the target duration can be greater than or equal to the actual movement duration.

[0102] In this way, different conflict path types may have different impacts on the collision risk of mobile objects. By considering the conflict type of the conflict path, it is possible to comprehensively and meticulously consider the impact of various factors on the collision probability between the second and third moving objects based on the specific scene characteristics, combined with the reference movement duration and fault information of the second and third moving objects, and the actual movement duration of the second moving object, and accurately analyze the collision risk between these two specific objects.

[0103] In some embodiments, step 503 combines the process of determining the sub-collision probability of a collision between the second moving object and the third moving object when the second moving object fails on the conflict path based on the conflict type of the conflict path, the reference movement duration, the actual movement duration and the fault information of the second moving object, and the reference movement duration and the fault information of the third moving object. Figure 6 Provide explanation.

[0104] Step 601: Determine a first target movement duration of the second moving object corresponding to the conflict type based on a reference movement duration or an actual movement duration of the second moving object.

[0105] In an embodiment of the present application, the determination of the first target motion duration of the second moving object is associated with the conflict type. If the conflict type is an intersection conflict type, the reference motion duration of the second moving object and the actual motion duration of the second moving object are the same. At this time, the first target motion duration of the second moving object can be the reference motion duration, or the actual motion duration; if the conflict type is an overlapping path conflict type, the reference motion duration of the second moving object and the actual motion duration of the second moving object may be different. At this time, the first target motion duration of the second moving object can be the actual motion duration.

[0106] Step 602: Determine a second target motion duration of the third moving object corresponding to the conflict type based on the reference motion duration and actual motion duration of the second moving object and one or more of the reference motion duration of the third moving object.

[0107] In this embodiment of the present application, the determination of the second target motion duration of the third moving object is associated with the conflict type. If the conflict type is an intersection conflict type, the reference motion duration of the third moving object and the actual motion duration of the third moving object are the same. In this case, the second target motion duration of the third moving object can be the reference motion duration or the actual motion duration. If the conflict type is an overlapping path conflict type, the second target motion duration of the third moving object can be determined based on the reference motion duration and actual motion duration of the second moving object, as well as the reference motion duration of the third moving object.

[0108] In some embodiments, overlapping path conflict types include two scenarios: first, there is an overlapping path and the two moving objects on the overlapping path move in the same direction; second, there is an overlapping path and the two moving objects on the overlapping path move in opposite directions. Here, based on the reference motion duration and actual motion duration of the second moving object, as well as the reference motion duration of the third moving object, the second target motion duration of the third moving object can be determined in combination with the motion directions of the two moving objects on the overlapping path in the overlapping path conflict type. This can be achieved in the following two ways:

[0109] The first one is that if the overlapping path conflict type is that there is an overlapping path and the two moving objects have the same moving direction on the overlapping path, the sum of the actual movement duration of the second moving object and the reference movement duration of the third moving object is obtained, and the difference obtained by subtracting the reference movement duration of the second moving object from the sum is used as the second target movement duration of the third moving object.

[0110] Here, if the overlapping path conflict type is that there is an overlapping path and the moving directions of the two moving objects on the overlapping path are the same, combined with Figure 2 As shown in the middle figure in , the second target motion duration of the third moving object can be expressed by the following formula (2).

[0111]

[0112] Wherein, T2 is the second target motion duration of the third moving object n, is the actual movement time of the second mobile object m to reach the fault stagnation point k, is the reference running time for the third mobile object n to reach the reference conflict point C, is the reference running time for the second mobile object m to reach the reference conflict point C.

[0113] The second type is that if the overlapping path conflict type is that there is an overlapping path and the two moving objects move in opposite directions on the overlapping path, obtain the sum of the reference movement duration of the second moving object and the reference movement duration of the third moving object, and subtract the actual movement duration of the second moving object from the sum to obtain the difference as the second target movement duration of the third moving object.

[0114] Here, if the overlapping path conflict type is that there is an overlapping path and the moving directions of the two moving objects on the overlapping path are opposite, combined Figure 2 As shown in the right figure in , the second target motion duration of the third moving object can be expressed by the following formula (3).

[0115]

[0116] Wherein, T2 is the second target motion duration of the third moving object n, is the actual movement time of the second mobile object m to reach the fault stagnation point k, is the reference running time for the third mobile object n to reach the reference conflict point D, is the reference running time for the second mobile object m to reach the reference conflict point C.

[0117] Step 603: Determine a first probability that the second moving object operates normally before reaching the fault stagnation point based on the fault information of the second moving object and the first target movement duration; and determine a second probability that the third moving object operates normally when reaching the fault stagnation point based on the fault information of the third moving object and the second target movement duration.

[0118] In an embodiment of the present application, the electronic device obtains predicted fault information corresponding to each time point within the first target movement duration when the second moving object moves for the first target movement duration; further, based on the fault information of the second moving object and the first target movement duration, a first probability of normal operation of the second moving object before reaching the fault stagnation point is determined. Here, the first probability can be achieved as follows:

[0119] The non-fault information (also called non-fault rate) corresponding to each time point of the second moving object within the first target movement duration is obtained, and the non-fault information corresponding to each time point is multiplied to obtain a first probability of normal operation of the second moving object before reaching the fault stagnation point.

[0120] Here, the first probability can be achieved by the following formula (4).

[0121]

[0122] Among them, P collision(m) is the first probability of normal operation of the second mobile object m before it reaches the fault stagnation point Q. The fault stagnation point Q can be the intersection C in the intersection conflict type or any point k in the overlapping path CD in the overlapping path conflict type; U m (t) is the failure rate of the second mobile object m at time point t, is the first target motion duration when the second moving object m reaches the fault stagnation point Q, and the value range of t is 1-U m (t) is the non-failure rate of the second mobile object m at time point t.

[0123] In an embodiment of the present application, the electronic device obtains predicted fault information corresponding to each time point within the second target movement duration when the third moving object moves for the second target movement duration; further, based on the fault information of the third moving object and the second target movement duration, a second probability of normal operation of the third moving object before reaching the fault stagnation point is determined. Here, the second probability can be achieved as follows:

[0124] The non-fault information (also called non-fault rate) corresponding to each time point of the third moving object within the second target movement duration is obtained, and the non-fault information corresponding to each time point is multiplied to obtain a second probability of normal operation of the third moving object when it reaches the fault stagnation point.

[0125] Here, the second probability can be realized by the following formula (5).

[0126]

[0127] Among them, P collision (n) is the second probability of normal operation of the third mobile object n when it reaches the fault stagnation point Q. The fault stagnation point Q can be the intersection C in the intersection conflict type or any point k in the overlapping path CD in the overlapping path conflict type; U n (t) is the failure rate of the third moving object n at time point t, T2 is the second target motion duration when the third moving object n reaches the failure stagnation point Q, and the value range of t is 1≤t <T2;1-U n (t) is U n (t) is the non-failure rate of the third mobile object n at time point t.

[0128] Step 604: Determine a sub-collision probability based on the collision type, the first probability, the second probability, and a third probability that a failure occurs when the second moving object reaches the failure stop point.

[0129] In the embodiment of the present application, after obtaining a first probability that the second moving object will operate normally before reaching the fault stagnation point, and a second probability that the third moving object will operate normally when reaching the fault stagnation point, the electronic device obtains a third probability that the second moving object will fail when reaching the fault stagnation point. Based on the first, second, and third probabilities, and utilizing a collision probability calculation model corresponding to the collision type of the collision path, the sub-collision probability between the second and third moving objects is calculated. This fully considers the impact of the fault state of the moving object on the collision risk, enabling more accurate quantification of the probability of a collision between the two moving objects in different situations and avoiding errors caused by simple estimation.

[0130] In some embodiments, step 604 combines the process of determining the sub-collision probability based on the collision type, the first probability, the second probability, and the third probability of the second moving object failing when the second moving object reaches the failure stop point. Figure 7 Provide explanation.

[0131] Step 701: Perform a first operation on the first probability, the second probability, and the third probability to obtain an intermediate sub-collision probability of a collision between the second moving object and the third moving object at any fault stagnation point.

[0132] Here, the intermediate sub-collision probability of the second mobile object and the third mobile object colliding at any fault stagnation point can be realized by the following formula (6).

[0133]

[0134] Among them, P collision (m,n) q is the intermediate sub-collision probability of the second mobile object m and the third mobile object n colliding at any fault stagnation point Q, is the failure rate of the second mobile object m when it reaches the failure stagnation point Q, is the first target motion duration when the second mobile object m reaches the fault stagnation point Q.

[0135] Step 702: Determine the sub-collision probability that the second moving object and the third moving object collide at all the fault stagnation points according to the conflict type and the intermediate sub-collision probability.

[0136] It is understandable that, based on the conflict type and the intermediate sub-collision probability, determining the sub-collision probability of the second moving object and the third moving object colliding at all the fault stagnation points can be achieved in the following manner:

[0137] If the conflict type is intersection conflict and the fault stagnation point is one, the intermediate sub-collision probability is determined as the sub-collision probability;

[0138] If the conflict type is an overlapping path conflict and there are multiple failure stuck points, a second operation is performed on the multiple intermediate sub-collision probabilities based on the intermediate sub-collision probabilities corresponding to the second mobile object and the third mobile object at each failure stuck point to obtain a sub-collision probability.

[0139] In one possible scenario, if the conflict type is an intersection conflict, such as Figure 2 In the left figure, there is only one stuck point Q, namely intersection C. In this case, the intermediate subcollision probability can be used as the subcollision probability, thereby obtaining the subcollision probability between the second and third mobile objects. For intersection conflicts, since there is only one stuck point, the intermediate subcollision probability can be directly used as the subcollision probability. This eliminates the need for complex comprehensive calculations and quickly and accurately reflects the collision probability in this scenario.

[0140] In another possible scenario, if the conflict type is overlapping path conflict, such as Figure 2 In the middle and right figures, there are multiple fault stuck points k, that is, multiple fault stuck points k on the overlapping path CD. In this case, the intermediate sub-collision probabilities corresponding to each fault stuck point can be accumulated and summed to obtain the sub-collision probability of a possible collision between the second mobile object and the third mobile object on the overlapping path.

[0141] Here, the sub-collision probability of a possible collision between the second mobile object and the third mobile object on the overlapping path can be expressed by the following formula (7).

[0142]

[0143] Among them, P collision (m,n) is the sub-collision probability of the second mobile object m and the third mobile object n colliding on the overlapping path, P collision (m,n) k is the intermediate sub-collision probability of the second mobile object m and the third mobile object n colliding at any fault stagnation point k, is the first target motion duration when the second mobile object m reaches the fault stagnation point k, is the movement time of the second moving object m to reach the nearest endpoint in the overlapping path, is the movement time of the second moving object m to reach the farthest endpoint in the overlapping path.

[0144] In this way, when the conflict type is an overlapping path conflict and there are multiple failure stagnation points, since multiple failure stagnation points may have collision risks in overlapping path conflicts, by calculating multiple intermediate sub-collision probabilities, the collision possibilities of different positions on the entire overlapping path can be fully considered, avoiding the risk of missing certain key positions, thereby more accurately calculating the sub-collision probabilities and improving the accuracy of collision probability assessment.

[0145] In some embodiments, step 103 is used to illustrate a process of determining target motion paths of the multiple moving objects based on the collision probability and the initial motion paths of the multiple moving objects.

[0146] If the collision probability does not meet the collision condition, the initial motion path of each moving object is determined as the target motion path;

[0147] If the collision probability meets the collision condition, the path of the moving object to be adjusted is re-planned to obtain the target motion path of the moving object to be adjusted; wherein the target motion path of the moving object to be adjusted does not meet the collision condition with the initial motion path of the unadjusted moving object, the moving object to be adjusted is determined based on the first moving object, and the unadjusted moving object is the remaining moving objects among the multiple moving objects except the moving object to be adjusted.

[0148] In the embodiment of the present application, if the collision probability is less than the collision probability threshold, it is determined that the collision probability does not meet the collision condition. This indicates that the probability of collision between the multiple first mobile objects due to a mobile object failure is relatively small. In this case, the initial motion path of each mobile object (including the first mobile object) can be used as its target motion path. This avoids unnecessary replanning based on the original safe path, saves computing resources and time costs, enables the mobile objects to operate efficiently according to the original plan, and improves the operating efficiency of the system.

[0149] In a feasible scenario, continue to refer to Figure 3 As shown, the initial motion path intersection M1 of the moving objects G1 and G3 is analyzed and meets the intersection conflict type. Figure 2 For the first group of mobile objects G1 and G3 that meet the path conflict condition, when mobile object G3 fails at intersection M1, the collision probability of a collision between mobile objects G1 and G3 is calculated by combining the above formulas (4), (5), and (6). For example, based on actual data, the above formulas (4), (5), and (6) are combined to obtain formula (8), which is used to calculate the collision probability of a collision between mobile objects G1 and G3.

[0150]

[0151] In this way, since the collision probability between the mobile objects G1 and G3 is less than the collision probability threshold, there is no need to replan the initial motion paths of the mobile objects G1 and G3, which saves computing resources and time costs, enables the mobile objects to operate efficiently according to the original plan, and improves the operating efficiency of the system.

[0152] In an embodiment of the present application, if the collision probability is greater than or equal to the collision probability threshold, it is determined that the collision probability meets the collision condition. Based on the planning information of multiple first mobile objects, such as priority or task urgency, the mobile object to be adjusted can be determined, so that the path planning of the mobile object to be adjusted can be re-performed to obtain the target motion path of the mobile object to be adjusted; the remaining mobile objects except the mobile object to be adjusted in the multiple mobile objects are determined as unadjusted mobile objects, and the initial motion path of the unadjusted mobile object can be used as its target motion path, thereby obtaining the target motion path of all mobile objects. It should be noted that the collision probability obtained based on the target motion path of all mobile objects does not meet the collision condition. In this way, the collision probability meets the collision condition, and the path planning is re-performed for the mobile object to be adjusted, which can accurately solve the path problem of the mobile object with collision risk, instead of re-planning all mobile objects, thereby reducing the workload and complexity of planning. At the same time, it is ensured that the collision probability obtained based on the target motion path of all mobile objects does not meet the collision condition, thereby ensuring the rationality and safety of the path planning of the entire system.

[0153] In a feasible scenario, continue to refer to Figure 3 As shown, the initial motion paths of the moving objects G2 and G3 are analyzed. There is an overlapping path M2-M3, and the moving directions of the moving objects G2 and G3 on the overlapping path M2-M3 are the same, which meets the overlapping path conflict type. Figure 2 The middle figure in FIG. For the second group of mobile objects G2 and G3 that meet the path conflict condition, when mobile object G3 fails on the overlapping path M2-M3, the collision probability of a collision between mobile objects G2 and G3 is calculated by combining the above formulas (2), (4), (5), (6), and (7). For example, based on actual data, the above formulas (2), (4), (5), (6), and (7) are combined to obtain formula (9), and the collision probability of a collision between mobile objects G2 and G3 is calculated using formula (9).

[0154]

[0155] in, is the first target motion duration when the mobile object G3 collides with the fault stagnation point k on the overlapping path.

[0156] It should be noted that, since the collision probability of the mobile objects G2 and G3 is greater than a collision probability threshold, such as 0.2, the initial motion path of the mobile object G2 or G3 needs to be replanned.

[0157] In some embodiments, determining the moving object to be adjusted may be achieved by: determining the priority of the first moving object; and determining the first moving object that meets the priority condition as the moving object to be adjusted.

[0158] In the embodiment of the present application, the first moving object that meets the priority condition may be the moving object with the lowest priority among the multiple first moving objects.

[0159] In the embodiment of the present application, the priority of each first moving object is obtained, and the first moving object with the lowest priority among the multiple first moving objects is determined, and the first moving object with the lowest priority is determined as the moving object to be adjusted.

[0160] In a feasible scenario, continue to refer to Figure 3 As shown, for the second group of mobile objects G2 and G3 that meet the path conflict condition and whose collision probability is greater than or equal to the collision probability threshold, according to the priority between the second group of mobile objects G2 and G3 that meet the path conflict condition, it is determined to re-plan the local path of the mobile object G2 with a low priority, repeat the above process and obtain the final path result, that is, the target motion path, as shown in Figure 8 As shown, Figure 8 The dotted line in is the target motion path obtained by replanning the moving object G2, and the initial motion path of the moving object G3 can be directly used as its target motion path.

[0161] This way, when rerouting, the lowest-priority mobile object is selected as the one to be adjusted. This avoids unnecessary adjustments to the paths of mission-critical or high-priority mobile objects, reducing computing resources and time. Furthermore, high-priority mobile objects often carry more important tasks, and frequent adjustments to their paths can cause task delays or failures. Adjusting the paths of lower-priority objects, on the other hand, has less impact on the overall system, thereby improving path planning efficiency and reducing resource costs.

[0162] In some embodiments, before step 102 determines the collision probability of the first moving objects colliding with each other when at least one first moving object fails on the conflict path corresponding to the path conflict condition based on the fault information of each first moving object, a further step may be performed. Figure 9 Steps in .

[0163] Step 901: Obtain object information corresponding to a plurality of moving objects;

[0164] Step 902: Input the object information of each moving object into the trained fault prediction model to obtain the fault information of each moving object within the target time period.

[0165] In an embodiment of the present application, the object information may include one or more of the battery power, temperature, current, moving speed and vibration amplitude of the mobile object. Of course, the object information of the mobile object may also include the equipment maintenance record of the mobile object, so as to obtain the failure rate when a failure occurs in the mobile object.

[0166] In the embodiment of the present application, the target period may be a pre-designed certain duration.

[0167] In the embodiments of the present application, the fault prediction model can be trained based on a Long Short-Term Memory (LSTM) model. Of course, the embodiments of the present application can also utilize other fault prediction models in the prior art to predict the failure rate of a moving object in the future based on the object information of the moving object. This application does not impose any specific restrictions on this, as long as the predicted failure rate of the moving object at each time point after the moving object begins to move can be obtained.

[0168] In an embodiment of the present application, object information corresponding to a plurality of moving objects is obtained, and the object information of each moving object is input into a trained fault prediction model. After analysis and processing by the fault prediction model, the failure rate corresponding to each time point in a future time period, i.e., a target time period, is predicted for each moving object when it begins to move, i.e., the fault information. It should be noted that after the moving object begins to move, the moving object's movement duration may be greater than the duration corresponding to the target time period. In this case, after the moving object moves for the duration corresponding to the target time period, the current object information of the moving object is re-acquired, and the current object information of the moving object is re-predicted using the fault prediction model to obtain the failure rate corresponding to each time point in the next target time period of the moving object's movement, until the failure rate corresponding to each time point within the maximum movement duration is obtained when the moving object moves for the maximum movement duration.

[0169] In some embodiments, the training process for the fault prediction model in step 902 includes combining Figure 10 Provide explanation.

[0170] Step 1001: Obtain first sample data, and use the first sample data to train a fault prediction model to be trained to obtain a trained intermediate fault prediction model.

[0171] The first sample data includes object information and fault labels collected by multiple mobile objects at consecutive sampling points.

[0172] In an embodiment of the present application, the first sample data may include object information collected from multiple mobile objects at consecutive sampling points, such as object information collected from each mobile object at 5-second intervals over the past week. Of course, the first sample data may also include a fault label, which may represent the probability of a mobile object malfunctioning. The fault label includes a probability of no fault occurring and a probability of a fault occurring, where the probability of no fault occurring can be represented by 0 and the probability of a fault occurring can be represented by 1. For example, referring to Table 2, Table 2 shows partial object information collected from a mobile object at 5-second intervals.

[0173]

[0174]

[0175] Table 2

[0176] In the embodiment of the present application, after obtaining the first sample data, normalization processing may be performed on the first sample data.

[0177] In the embodiment of the present application, first sample data is obtained and used to train a fault prediction model to be trained, thereby obtaining a trained intermediate fault prediction model. Thus, the initial model is trained using first sample data containing object information and fault labels for multiple moving objects at consecutive sampling points. This data reflects the actual operating status and fault conditions of the moving objects, providing rich training data for the model.

[0178] Step 1002: Input the first sample data into the trained intermediate fault prediction model to obtain the predicted failure rate of the next sampling point corresponding to each object information of the mobile object.

[0179] In an embodiment of the present application, when the electronic device completes training of the fault prediction model and obtains an intermediate fault prediction model, the first sample data is input into the trained intermediate fault prediction model to obtain the predicted failure rate of each mobile object at the next sampling point corresponding to each object information. For example, if the first sample data is the object information corresponding to the sampling point 00:00:05 in Table 1, the object information corresponding to the sampling point 00:00:05 is input into the trained intermediate fault prediction model to obtain the predicted failure rate of each mobile object at the next sampling point 00:00:10 corresponding to the object information corresponding to the sampling point 00:00:05. In this way, the first sample data is input into the intermediate model to obtain the predicted failure rate, and then the first sample data is adjusted based on this. This feedback mechanism enables the model to pay attention to the deviation of its own prediction and highlight the features related to the fault in the adjusted data. When trained again, the model can capture these features more accurately, thereby improving the accuracy of fault prediction.

[0180] Step 1003: Based on the predicted failure rate, adjust the first sample data to obtain adjusted first sample data.

[0181] The adjusted first sample data includes object information and predicted failure rates collected by multiple moving objects at consecutive sampling points.

[0182] In an embodiment of the present application, the electronic device can modify the fault labels corresponding to the object information collected by multiple mobile objects in the first sample data at each sampling point to the predicted failure rate corresponding to the next sampling point, thereby adjusting the first sample data.

[0183] Step 1004: Segment the adjusted first sample data according to the time window to obtain multiple groups of second sample data. Use the multiple groups of second sample data to train the trained intermediate fault prediction model again to obtain a trained fault prediction model.

[0184] In an embodiment of the present application, the time window can be a preset time period, and the adjusted first sample data of each moving object can be divided according to the time window to obtain multiple groups of adjusted first sample data. For example, the size of the time window can be 30s or 60s, or other sizes, and this application does not impose any specific restrictions on this. In an achievable scenario, the adjusted first sample data is obtained, normalized, and divided using a time window of a preset time size, such as a 30s time step, to obtain grouped adjusted first sample data. For example, the grouped adjusted first sample data is shown in Table 3.

[0185]

[0186] Table 3

[0187] In an embodiment of the present application, the adjusted first sample data is obtained, and the adjusted first sample data is segmented according to the time window to obtain multiple groups of adjusted first sample data. The trained intermediate fault prediction model is recursively trained using the multiple groups of adjusted first sample data to obtain a trained fault prediction model. In this way, using the finally trained fault prediction model, the fault rate corresponding to each time point (also known as a sampling point) in the future time window (corresponding to the above-mentioned target period) when the mobile object starts to move can be predicted based on the current object information of the mobile object. In this way, segmenting the adjusted first sample data through the time window allows the model to focus on the characteristics of the data in different time periods. The operating status of multiple mobile objects may show different rules in different time periods. The segmented data helps the model capture the characteristics of these dynamic changes and better learn the relationship between these changes and the occurrence of faults, thereby more accurately predicting the failure rate at each time point in the future. At the same time, the intermediate fault prediction model is recursively trained using multiple sets of adjusted first sample data, so that the model continuously optimizes parameters in each training. Recursive training enables the model to learn the complex patterns and rules in the data more deeply, gradually improve the ability to recognize fault characteristics, and thus improve the accuracy of fault prediction.

[0188] Here, in an achievable scenario, a path planning method provided in an embodiment of the present application is described.

[0189] Currently, warehousing systems with track-based management (such as AGV / AMR systems) are developing rapidly, especially in fields with a high degree of automation, such as e-commerce, manufacturing, and logistics distribution. Although many companies have applied mature algorithms and technologies in AGV scheduling systems, such as path planning and task scheduling, with the increasing complexity of application scenarios and the diversification of needs, there are still many problems that need to be solved urgently, such as: multi-AGV collaborative scheduling, multi-target path planning, etc. The quality of the algorithm greatly affects production efficiency.

[0190] In response to the above problems, heuristic search algorithms such as the Dijkstra algorithm and the A* algorithm are used in related technologies, or swarm intelligence algorithms such as the ant colony algorithm are used for path planning. However, no matter which algorithm is used, the specific application is roughly divided into a single AGV path planning stage and a collaborative collision avoidance stage. Generally speaking, in the collaborative collision avoidance stage, collisions can be identified and adjusted for multiple AGV paths, but when the number of AGVs increases, the probability of AGV failure will also increase, and the collision avoidance stage will only consider conflict collisions at the same time point, while the algorithm does not take into account the frequency of failures in actual production, which may lead to inaccurate AGV position estimation at the same time point. Therefore, even in the collaborative collision avoidance stage, there is no way to identify specific collisions in operation.

[0191] In order to solve the above problems, Figure 11 As shown, the embodiment of the present application provides a path planning method, the process is as follows:

[0192] Step 1101: Use the trained LSTM model to obtain the future failure rate of each AGV.

[0193] Here, for the training of LSTM model, you can combine Figure 12 accomplish.

[0194] A training sample set is obtained 1201; the training sample set includes sensor data and fault event records for each AGV. The AGV sensor data is sensor data collected at each sampling point, including the AGV's temperature, vibration amplitude, current, movement speed, and battery charge; the fault event records correspond to the aforementioned fault labels. The training sample set is processed according to a time window to obtain a shape for each training sample 1202; the training sample shape includes the number of training samples, the time window size, and the number of features. The training sample set is normalized 1203. An LSTM model is trained 1204; the LSTM model is trained using the normalized training sample set to obtain a trained LSTM model. A sigmoid activation function is used in the LSTM model. The LSTM model is validated 1205; after the LSTM model is trained, the validation sample set is used to validate the trained LSTM model.

[0195] Step 1102: Calculate the optimal path for each AGV using an existing path planning algorithm.

[0196] Here, the optimal path corresponds to the initial motion path described above. It should be noted that the optimal path for each AGV is calculated using existing path planning algorithms, thereby achieving preliminary path planning for each AGV. Furthermore, after the initial path planning phase is completed, the AGVs exchange information with each other, and each AGV will receive the optimal path obtained by the other AGVs during the path planning phase, and the planning will enter the collaborative collision avoidance phase.

[0197] Step 1103: Calculate the collision rate.

[0198] Here, we first calculate whether AGVs overlap at the same time point to determine whether the paths conflict. For the remaining intersections or overlapping paths, we add the AGV failure rate to make a judgment.

[0199] Here, when analyzing the collision probability of multiple paths' intersections or overlapping paths, the conflict types can be divided into the following categories:

[0200] Type 1: There is a common intersection between the optimal paths of multiple AGVs. For example, refer to Figure 2As shown in the left figure, there is a common intersection C between the two AGVs.

[0201] Type 2: There are repeated paths (also called overlapping paths, repeated segments) between the optimal paths of multiple AGVs, and multiple AGVs move or drive in the same direction. For example, refer to Figure 2 As shown in the middle figure, there is a repeated path CD between the two AGVs and they move in the same direction.

[0202] Type 2: There are repeated paths (also called overlapping paths, repeated segments) between the optimal paths of multiple AGVs, and multiple AGVs move or drive in opposite directions. For example, refer to Figure 2 As shown in the right figure, there is a repeated path CD between the two AGVs and they move in opposite directions.

[0203] Below, we analyze the above three situations.

[0204] For Type 1, there is an optimal path for N AGVs to intersect at point C, and each AGV is defined as GAV i , i∈(1,2,…,N), and the AGV corresponding to the time point t i The failure rate is U i (t), the running time of each AGV from the initial point (corresponding to the above starting point) to the intersection C (corresponding to the above actual movement time) is according to For N AGVs i Sort in ascending order. For example, the sorted is the minimum value.

[0205] Furthermore, define any two AGVs such as AGV m and AGV n ,in, AGV m and AGV n The probability that no collision occurs at the intersection C is P nocollision (m,n), the probability of collision is P collision (m,n), then AGV m and AGV n The probability of a collision occurring at the intersection C (corresponding to the above-mentioned sub-collision probability) can be expressed by formula (10).

[0206]

[0207] in, For AGV m The time to move from the initial point to the intersection point C, For AGV n The time from the initial point to the intersection point C, Um (t) is AGV m The predicted failure rate at time t, U n (t) is AGV n The predicted failure rate at time t, For AGV m At the time point The predicted failure rate when .

[0208] It should be noted that since the optimal paths of N AGVs have the same intersection C, when at least one AGV fails at the intersection C, the probability of collision between the optimal paths of any two AGVs can be obtained by the above formula (10), thus obtaining probability, such as Further, based on According to the inclusion-exclusion theorem, the probability of collision between the optimal paths of N AGVs (corresponding to the above collision probability) can be expressed by the above formula (1), thereby obtaining the total probability of collision between N AGVs when at least one AGV fails.

[0209] For Type 2, there are N AGVs whose optimal paths have repeated segments CD and travel in the same direction. Each AGV is defined as AGV i , i∈(1,2,…,N), and the AGV corresponding to the time point t i The failure rate is U i (t), the running time of each AGV from the initial point to point C is according to For N AGVs i Sort in ascending order. For example, after sorting is the minimum value.

[0210] Based on the analysis of Type 1, first consider the probability of two AGVs colliding:

[0211] Two AGVs such as AGV m and AGV n There is only one case where a collision occurs, namely AGV m If a fault occurs in the CD section, the AGV stops moving and n Can reach the failure stagnation point, So when AGV m At the sth second, a fault stagnation occurs. AGV n Need to Therefore, AGV m and AGV nThe probability of a collision occurring in the repeated segment CD (corresponding to the above-mentioned sub-collision probability) can be expressed by formula (11).

[0212]

[0213] in, For AGV m The time to move from the initial point to point C in the overlapping segment, For AGV m The time it takes for the AGV to move from the initial point to point D in the overlapping segment, s is m The actual operating time when the fault occurred, For AGV n The time to move from the initial point to point C in the overlapping segment, U m (t) is AGV m The predicted failure rate at time t, U n (t) is AGV n The predicted failure rate at time t, U m (s) is AGV m The predicted failure rate at time s.

[0214] Thus, through the above formula (11), we can get probability. Further, based on According to the inclusion-exclusion theorem, the probability of collision between the optimal paths of N AGVs can be expressed by the above formula (1), thereby obtaining the total probability of collision between N AGVs when at least one AGV fails.

[0215] For Type 3, there are N AGVs whose optimal paths have repeated segments CD and they are traveling in opposite directions. Each AGV is defined as AGV i , i∈(1,2,…,N), the average speed of AGV is Vm / min. The AGV corresponding to time point t i The failure rate is U i (t), the running time of each AGV from the initial point to the nearest end point of the repeated segment is according to For N AGVs i Sort in ascending order. For example, after sorting is the minimum value.

[0216] In this scenario, if there is no fault and no collision, there is a situation where the AGV has not reached one end of the repeat segment, while the other AGV has already completed the repeat segment. m and AGV n There is only one collision situation, which defines the distance AGV mThe closer end is C and the other end is D, which is AGV m In the repeat section CD fails and AGV n Can reach the failure stagnation point, When AGV m At the s minute, a malfunction occurred. AGV n Need to Therefore, AGV m and AGV n The probability of a collision occurring in the repeated segment CD (corresponding to the above-mentioned sub-collision probability) can be expressed by formula (12).

[0217]

[0218] Thus, through the above formula (12), we can get probability. Further, based on According to the inclusion-exclusion theorem, the probability of collision between the optimal paths of N AGVs can be expressed by the above formula (1), thereby obtaining the total probability of collision between N AGVs when at least one AGV fails.

[0219] Step 1104: Whether a collision occurs.

[0220] Here, a predefined conflict acceptance rate r (corresponding to the above-mentioned collision probability threshold) is obtained. When the calculated collision rate between AGVs is greater than or equal to the conflict acceptance rate, it indicates that a collision has occurred between the AGVs, and the process returns to step 1103. When the calculated collision rate between AGVs is less than the conflict acceptance rate, it indicates that no collision has occurred between the AGVs, and the process executes step 1105.

[0221] Step 1105: Output the planned path result.

[0222] Here, when the collision probability at this location is calculated to be greater than or equal to the conflict acceptance rate, the priority of the AGV in conflict is obtained, and judgment is made based on the priority of the AGV. The AGV with high priority operates according to the original planned path, and the AGV with low priority re-plans a local optimal path. The entire path planning process is completed until the paths planned by all AGVs can safely reach the target point, and the planned path result is output.

[0223] The embodiment of the present application provides a path planning device, referring to Figure 13 As shown, the path planning device 13 includes:

[0224] A first determining module 1301 is configured to determine a first moving object that satisfies a path conflict condition based on initial motion paths of the plurality of moving objects; wherein the first moving object is at least part of the plurality of moving objects;

[0225] A second determining module 1302 is configured to determine, based on the fault information of each first moving object, a collision probability of collision between the first moving objects when at least one first moving object fails on a conflicting path corresponding to the path conflict condition;

[0226] The third determining module 1303 is configured to determine target motion paths of the multiple moving objects based on the collision probability and the initial motion paths of the multiple moving objects; wherein the collision probability obtained based on the target motion path does not satisfy the collision condition.

[0227] Based on the above embodiments, the present application provides an electronic device, referring to Figure 14 As shown, the electronic device 14 includes: a processor 1401 and a memory 1402, wherein,

[0228] The memory 1402 stores computer programs that can be executed on the processor 1401;

[0229] The processor 1401 executes the computer program stored in the memory 1402 to implement the following steps:

[0230] Determining a first moving object that meets a path conflict condition based on initial motion paths of the plurality of moving objects; wherein the first moving object is at least a portion of the plurality of moving objects;

[0231] determining, based on the fault information of each first moving object, a collision probability of collision between the first moving objects when at least one first moving object fails on a conflicting path corresponding to the path conflict condition;

[0232] Target motion paths of the multiple moving objects are determined based on the collision probability and initial motion paths of the multiple moving objects; wherein the collision probability obtained based on the target motion path does not meet the collision condition.

[0233] The method provided in the embodiment of the present application can be directly embodied as a combination of software modules executed by the processor 1401. The software module can be located in a storage medium, and the storage medium is located in the memory 1402. The processor 1401 reads the executable instructions included in the software module in the memory 1402 and completes the method provided in the embodiment of the present application in combination with the necessary hardware.

[0234] As an example, the processor 1401 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0235] It should be noted that the specific implementation process of the steps executed by the processor in this embodiment can refer to the steps in the method provided in the above embodiment, and will not be repeated here.

[0236] An embodiment of the present application provides a storage medium storing a computer program. When the computer program is executed by at least one processor, the steps of the method provided in the above embodiment are implemented, which will not be described in detail here.

[0237] An embodiment of the present application provides a computer program product, including a computer program or instructions. When the computer program or instructions are executed by a processor, the steps in the method provided in the above embodiment are implemented, which will not be repeated here.

[0238] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. Furthermore, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage and optical storage, etc.) containing computer-usable program code.

[0239] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0240] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0241] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0242] The above are merely preferred embodiments of the present application and are not intended to limit the scope of protection of the present application.

Claims

1. A path planning method, comprising: Determining a first moving object that meets a path conflict condition based on initial motion paths of a plurality of moving objects; wherein the first moving object is at least part of the plurality of moving objects; determining, based on the fault information of each of the first mobile objects, a collision probability of collision between the first mobile objects when at least one of the first mobile objects fails on the conflicting path corresponding to the path conflict condition; Based on the collision probability and the initial motion paths of the multiple moving objects, target motion paths of the multiple moving objects are determined; wherein the collision probability obtained based on the target motion path does not meet a collision condition.

2. The method according to claim 1, wherein determining, based on the fault information of each of the first mobile objects, a collision probability of a collision between the first mobile objects when at least one of the first mobile objects fails on the conflicting path corresponding to the path conflict condition comprises: determining a reference conflict point corresponding to each of the first mobile objects, wherein the reference conflict point is determined based on the path conflict condition; Obtaining a reference motion duration of each first moving object moving from a starting motion point of a corresponding initial motion path to the reference conflict point; determining, based on the reference running time and fault information of each of the first mobile objects, a sub-collision probability of a collision between any two first mobile objects when at least one of the first mobile objects fails on the conflicting path; Based on the multiple sub-collision probabilities, a collision probability between the first moving objects is determined.

3. The method according to claim 2, wherein determining, based on the reference running time and fault information of each first mobile object, the sub-collision probability of a collision between any two first mobile objects when at least one first mobile object fails on the conflicting path comprises: If a second moving object fails on the conflicting path, determining a failure stagnation point of the second moving object; wherein the second moving object is a moving object corresponding to a first reference motion duration of any two reference motion durations among a plurality of reference motion durations, and the failure stagnation point is a point on the conflicting path; Obtaining an actual movement time of the second moving object moving to the fault stagnation point; Based on the conflict type of the conflict path, the reference movement duration, actual movement duration and fault information of the second mobile object, and the reference movement duration and fault information of the third mobile object, the sub-collision probability of a collision between the second mobile object and the third mobile object when the second mobile object fails on the conflict path is determined; wherein, the third mobile object is the mobile object corresponding to the second reference movement duration of any two reference movement durations.

4. The method according to claim 3, wherein determining the sub-collision probability of a collision between the second mobile object and the third mobile object when the second mobile object fails on the conflict path based on the conflict type of the conflict path, the reference movement duration, the actual movement duration, and the fault information of the second mobile object, and the reference movement duration and the fault information of the third mobile object comprises: determining a first target movement duration of the second moving object corresponding to the conflict type based on a reference movement duration or an actual movement duration of the second moving object; determining a second target movement duration of the third moving object corresponding to the conflict type based on one or more of the reference movement duration and the actual movement duration of the second moving object and the reference movement duration of the third moving object; determining a first probability that the second moving object operates normally before reaching the fault stagnation point based on the fault information of the second moving object and the first target movement duration, and determining a second probability that the third moving object operates normally when reaching the fault stagnation point based on the fault information of the third moving object and the second target movement duration; The sub-collision probability is determined based on the conflict type, the first probability, the second probability, and a third probability that the second mobile object fails when reaching the failure stop point.

5. The method according to claim 4, wherein determining the sub-collision probability based on the collision type, the first probability, the second probability, and a third probability that the second mobile object fails when reaching the failure stop point comprises: performing a first operation on the first probability, the second probability, and the third probability to obtain an intermediate sub-collision probability of a collision between the second moving object and the third moving object at any fault stagnation point; The sub-collision probability that the second mobile object and the third mobile object collide at all the fault stagnation points is determined according to the conflict type and the intermediate sub-collision probability.

6. The method according to claim 5, wherein determining the sub-collision probability that the second moving object and the third moving object collide at all the failure stuck points based on the conflict type and the intermediate sub-collision probability comprises: If the conflict type is an intersection conflict and the fault stagnation point is one, the intermediate sub-collision probability is determined as the sub-collision probability; If the conflict type is an overlapping path conflict and there are multiple failure stuck points, a second operation is performed on the multiple intermediate sub-collision probabilities corresponding to each of the failure stuck points of the second mobile object and the third mobile object to obtain the sub-collision probability.

7. The method according to any one of claims 1 to 6, wherein determining target motion paths of the plurality of moving objects based on the collision probability and the initial motion paths of the plurality of moving objects comprises: If the collision probability does not satisfy the collision condition, determining the initial motion path of each of the moving objects as the target motion path; If the collision probability satisfies the collision condition, the path planning is performed again on the moving object to be adjusted to obtain the target motion path of the moving object to be adjusted; wherein, the target motion path of the moving object to be adjusted and the initial motion path of the unadjusted moving object do not satisfy the collision condition, the moving object to be adjusted is determined based on the first moving object, and the unadjusted moving object is the remaining moving objects among the multiple moving objects except the moving object to be adjusted.

8. The method according to claim 7, comprising: determining a priority of the first moving object; The first moving object that meets the priority condition is determined as the moving object to be adjusted.

9. The method according to any one of claims 1 to 6, further comprising: Obtaining object information corresponding to each of the plurality of moving objects; The object information of each of the moving objects is input into a trained fault prediction model to obtain the fault information of each of the moving objects within a target time period.

10. The method according to claim 9, wherein the training process of the fault prediction model comprises: Obtaining first sample data, and training a fault prediction model to be trained using the first sample data to obtain a trained intermediate fault prediction model, wherein the first sample data includes object information and fault labels collected by multiple moving objects at consecutive sampling points; Inputting the first sample data into the trained intermediate fault prediction model to obtain a predicted failure rate of the next sampling point corresponding to each object information of the mobile object; Based on the predicted failure rate, the first sample data is adjusted to obtain adjusted first sample data; wherein the adjusted first sample data includes object information collected by the multiple moving objects at consecutive sampling points and the predicted failure rate; The adjusted first sample data is segmented according to the time window to obtain multiple groups of second sample data, and the trained intermediate fault prediction model is trained again with the multiple groups of second sample data to obtain the trained fault prediction model.

Images