Agricultural machine multi-machine path planning method oriented to time-varying communication constraint

Abstract

The invention discloses an agricultural machine multi-machine path planning method oriented to time-varying communication constraints, and the method specifically comprises the following steps: S1, constructing a time cost evaluation function when an agricultural machine task migrates through the analysis of the migration time of the agricultural machine; s2, according to the agricultural machinery task migration time cost evaluation function constructed in the step S1, constructing a node evaluation function during path planning; s3, constructing a grid map for edge node communication features in the farmland map; s4, performing path planning on the single agricultural machine through the node evaluation function during path planning construction in the step S2; s5, performing multi-machine path planning on multiple agricultural machines in combination with the step S4; through an improved path node search strategy, operation routes of other agricultural machines are considered in the time dimension, so that the route conflict risk in collaborative operation is avoided. In an unmanned agricultural machinery operation environment, agricultural machinery cooperation efficiency and safety are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural machinery path planning, and particularly relates to a multi-machine path planning method for agricultural machinery facing time-varying communication constraints. Background Art

[0002] In multi-agricultural machinery collaborative operations, path planning is one of the important technologies to achieve efficient collaborative operations. From the whole process, the multi-agricultural machinery collaborative operation consists of three processes: operation task allocation, agricultural machinery scheduling, and collaborative operation control. Reasonable path planning for each agricultural machinery greatly reduces the scheduling time, and thus improves the overall collaborative operation efficiency. Facing the gradually complex farm operation environment, the complexity of the path planning algorithm is also increasing, which will lead to abnormal communication between traditional cloud computing and agricultural machinery when facing a huge amount of computing. Deploying the relevant calculations of multi-machine collaborative services such as path planning under edge computing can effectively alleviate this problem. However, with the increase in the operation scale and the number of agricultural machinery, the frequency and span of the movement of agricultural machinery gradually increase, which makes the service migration of agricultural machinery among multiple edge nodes become frequent, and then leads to abnormal problems such as increased communication delay, jamming, and even interruption between the edge computing nodes and the agricultural machinery. Since key services such as driverless and environmental perception are involved in the edge computing nodes, if the time-varying communication constraint in the edge computing is not considered in the path planning, it will not only lead to communication waiting behavior in agricultural machinery operations, but may even lead to safety production accidents such as out-of-control of agricultural machinery.

[0003] Although there have been methods to study path planning methods with goals such as the shortest moving distance, the fewest turning times, or the minimum obstacle avoidance cost, these methods are all based on an ideal communication environment. In an edge computing environment with time-varying communication, even if the shortest route is planned, the agricultural machinery may have communication delays or even interruptions with the edge nodes it is located in when tracking the route, and then there will be a behavior of the agricultural machinery stopping and waiting, and ultimately its operation time is not the most efficient. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-machine path planning method for agricultural machinery facing time-varying communication constraints in view of the deficiencies of the prior art.

[0005] The technical solution for the present invention to solve the above problems is: a multi-machine path planning method for agricultural machinery facing time-varying communication constraints, specifically including the following steps:

[0006] Step S1: Analyze the migration time of agricultural machinery to construct an evaluation function for the time spent during the task migration of agricultural machinery;

[0007] Step S2: Construct a node evaluation function for path planning according to the evaluation function for the time spent during the task migration of agricultural machinery constructed in Step S1;

[0008] Step S3: Construct a grid map for the communication characteristics of edge nodes in the farmland map;

[0009] Step S4: Perform path planning for the single agricultural machine through the node evaluation function in path planning constructed in Step S2;

[0010] Step S5: Combine Step S4 to perform multi-machine path planning for multiple agricultural machines.

[0011] Further, the time cost evaluation function constructed in Step S1 for the migration of agricultural machine tasks is:

[0012]

[0013] represents the amount of data carried during the migration of task i in agricultural machine am1, R load (T c ) is the CPU load rate of this node, is the size of the task waiting to be processed in this node, B mecx is the edge node mec x theoretical maximum broadband, P available is the available computing resource of this node, U mecx (t) is the network utilization rate, the bandwidth transmission data time T i of task trans , C taski is the computing requirement of the task, the computing power of the edge node mec x is P mecx , its computing load rate is R load , T net.queuei is the queuing waiting time of the task in the network.

[0014] Further, the node evaluation function in path planning constructed in Step S2:

[0015]

[0016] F(P[i]) is the final evaluation result of point P[i], G(P[i]) is the distance cost from the starting point to this point, H(P[i]) is the distance cost from this point to the end point, D(P[i]) is the evaluation of the network delay in Step S1, mainly evaluating the network communication delay by the migration time of the task in the node, where w is the weight of the communication delay cost in evaluating the path, and its main role is to dynamically balance between efficiency and security, P[i].mec represents the serial number of the edge node corresponding to this point, Belong mec represents the serial number of the current edge node during evaluation. When the two are equal, it means that this point is at the serial number Belong mecWithin the range of the edge node, there is no need to consider the communication delay cost caused by task migration. When the two are not equal, it means that this point belongs to other edge nodes.

[0017] Furthermore, the specific steps of step S4 are as follows:

[0018] Step S4.1: Mark the starting point and set the evaluation function value of the starting point as the shortest distance from the starting point to the target point;

[0019] Step S4.2: Search for the node with the minimum evaluation function value among the global open nodes in the grid map, select the node with the minimum evaluation function, and remove it from the global open nodes;

[0020] Step S4.3: Check whether the target is reached;

[0021] Step S4.4: Calculate the relationship between the parent node and the current node, calculate the time from the current node's parent node to the target point, and update the obstacle influence area of each node according to the trajectory of the dynamic obstacle;

[0022] Step S4.5: Sample and update the obstacle trajectory;

[0023] Step S4.6: Judge whether the obstacle affects the path;

[0024] Step S4.7: Select successor nodes and update the evaluation function: Select the successor nodes around the current node and calculate their evaluation function values;

[0025] Step S4.8: Continue the search until a path is found. If the evaluation function value of the successor node is less than that of its parent node, update the parent node and continue to execute step S4.7; If no path is found yet, continue to select a new minimum evaluation node from the global open nodes;

[0026] Step S4.9: End of path search: If the global open nodes are empty, it means that no path can be found, return "Path cannot be found", otherwise, finally return the shortest path from the starting point to the target point.

[0027] Furthermore, the specific steps of step S5 are as follows:

[0028] Step S5.1: Set the number of agricultural machines required for the current path planning as AMNum;

[0029] Step S5.2: Perform path planning sorting according to the independent operation ability of the agricultural machines, and use the array AM[AMNum] to store the agricultural machine information;

[0030] Step S5.3: Set the index i = 1;

[0031] Step S5.4: P[start]=AM[i].Start, P[end]=AM[i].End. Obtain PATH[i] through Step S4, update the map MAP = MAP + PATH[i], and i = i + 1. When i ≠ AMNum, continue to execute S5.4; otherwise, execute S5.5.

[0032] Step S5.5: Return the multi - agricultural - machine path planning path set PATH.

[0033] The present invention has beneficial effects:

[0034] The present invention provides a multi - agricultural - machine path planning method for time - varying communication constraints, which incorporates an evaluation model of communication delay in the edge - node environment into the evaluation function and improves the path - node selection strategy in combination with time factors. Among them, through the accurate quantification of time - varying communication delay from two aspects: the time - varying load state of edge nodes and the task migration of agricultural machines, the perception ability of the algorithm evaluation function for the end - edge communication state is increased, and finally the impact of communication delay on the driving of agricultural - machine routes is reduced; through the improved path - node search strategy, the operation routes of other agricultural machines are considered in the time dimension to avoid the risk of route conflicts in collaborative operations. In the environment of unmanned agricultural - machine operations, the collaborative efficiency and safety of agricultural machines are greatly improved. Description of the Drawings

[0035] Figure 1 It is the path - planning grid map of the present invention.

[0036] Figure 2 It is the single - machine path - planning flowchart of Step S4 of the present invention. Detailed Embodiments

[0037] A multi - agricultural - machine path planning method for time - varying communication constraints specifically includes the following steps:

[0038] Step S1: By analyzing the migration time of agricultural machines, construct an evaluation function for the time cost during the task migration of agricultural machines:

[0039]

[0040] When the agricultural machine am x enters other edge nodes mec x at time t, the migration time of its task Task amx ={task1, task2, …, task h , …, task N}. represents the amount of data carried during the migration of task i in the agricultural machine am1, and R load (T c) is the CPU load rate of this node, is the size of the tasks waiting to be processed in this node, B mecx is the edge node mec x Theoretical maximum bandwidth, P available is the computing resources available for this node, U mecx (t) is the network utilization rate, task i The bandwidth transmission data time T of trans , C taski (Unit: FLOPs) is the computing requirement of the task, edge node mec x The computing power of is P mecx (Unit FLOPs / s, taking CPU as an example), its computing load rate is R load , T net.queuei is the queuing waiting time of the task in the network. Analyzing the agricultural machinery migration time can effectively calculate the time spent by the agricultural machinery task during migration, and then effectively evaluate the communication delay between the agricultural machinery and the node.

[0041] Step S2: Construct a node evaluation function for path planning:

[0042]

[0043] Among them, F(P[i]) is the final evaluation result of point P[i], G(P[i]) is the distance cost from the starting point to this point, H(P[i]) is the distance cost from this point to the end point, D(P[i]) is the evaluation of the network delay in step S1, mainly evaluating the network communication delay based on the migration time of the task in the node. Among them, ω is the weight of the communication delay cost when evaluating the path, and its main role is to dynamically balance between efficiency and security. P[i].mec represents the serial number of the edge node corresponding to this point, Belong mec represents the serial number of the current edge node during evaluation. When the two are equal, it means that this point is within the range of the edge node with the serial number Belong mec and there is no need to consider the communication delay cost caused by task migration. When the two are not equal, it means that this point belongs to other edge nodes, and the communication delay cost caused by task migration needs to be considered during evaluation. Using the designed node evaluation function makes the decision-making of the node pay more attention to the impact of communication delay.

[0044] Step S3: Construct a grid map as shown in Figure 1 for the communication characteristics of the edge nodes in the farmland map. The grid map can more specifically reflect the communication coverage of the edge nodes and provide effective data assistance for path planning.

[0045] Step S4: Perform path planning for the single agricultural machine through the node evaluation function in path planning construction in Step S2. When performing path planning for the single agricultural machine, this step makes the single agricultural machine pay more attention to the impact of communication delay on the subsequent travel of agricultural machines. The finally calculated route is also the operation route with the most ideal communication quality.

[0046] The specific steps are as follows:

[0047] Step S4.1: Mark the starting point P[start] as openlist, and set the evaluation function value F(P[start]) of the starting point to the shortest distance from the starting point to the target point (usually 0);

[0048] Step S4.2: Search for the node with the minimum evaluation function value in openlist. Select the node P[i] with the minimum evaluation function F(P[i]) from openlist, and remove it from openlist. Mark the node P[i] as closelist, indicating that it is a processed node;

[0049] Step S4.3: Check if the target is reached: If the current node P[i] is equal to the target point P[end], the path planning is completed, and return "Path found";

[0050] Step S4.4: Calculate the relationship between the parent node and the current node, calculate the time Tc from the parent node P[i-1] of the node P[i] to the target point, and update the obstacle influence area of each node according to the trajectory of the dynamic obstacle;

[0051] Step S4.5: Sample the obstacle trajectory and update: Sample the trajectories of n known dynamic obstacles, and obtain their positions Dn[Tc] at time Tc; Generate the navigation area of each obstacle, and generate a prohibited area that collides with the obstacle, and these areas are defined as Obsn[Tc];

[0052] Step S4.6: Determine whether the obstacle affects the path: Check whether the obstacle Obsn[Tc] belongs to openlist. If it does not belong to openlist, mark it as closelist to avoid repeated calculation. If the obstacle conflicts with the current path, continue to find a feasible path.

[0053] Step S4.7: Select successor nodes and update the evaluation function: Select the successor nodes P[j] around the current node P[i], and calculate their evaluation function values F(P[j]). If the successor node P[j] belongs to an obstacle or closelist, skip this node. If the successor node has not been processed, add it to openlist, and update its parent node information according to the current path.

[0054] Step S4.8: Continue the search until a path is found. If the evaluation function value of the successor node is less than that of its parent node, update the parent node and continue to execute Step S4.7. If no path is found yet, continue to select a new minimum evaluation node from the open list.

[0055] Step S4.9: End of path search: If the open list is empty, it means that no path can be found, and return "Path cannot be found". Otherwise, finally return the shortest path from the starting point to the target point.

[0056] Step S5: Based on Step S4, perform the following multi - machine path planning for agricultural machines. In the subsequent planning of agricultural machines, further consider the route conflict problem with other agricultural machines, and finally plan a safe operation route that meets the communication requirements for each agricultural machine in the agricultural machine group.

[0057] The specific steps are as follows:

[0058] Step S5.1: Set the number of agricultural machines for which the path needs to be planned currently as AMNum;

[0059] Step S5.2: Sort the path planning according to the independent operation ability of the agricultural machines, and use the array AM[AMNum] to store the agricultural machine information;

[0060] Step S5.3: Set the index i = 1;

[0061] Step S5.4: P[start]=AM[i].Start, P[end]=AM[i].End, obtain PATH[i] through Step S4, update the map MAP = MAP + PATH[i], i = i + 1. When i ≠ AMNum, continue to execute S5.4, otherwise execute S5.5;

[0062] Step S5.5: Return the path set PATH of the multi - machine path planning for agricultural machines.

[0063] The above - mentioned is only the preferred embodiment of the present invention, and it is not a limitation to the present invention in other forms. Any person skilled in the relevant art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change, and modification made to the above - mentioned embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for multi-machine path planning for agricultural machinery with time-varying communication constraints, characterized by: The specific steps include: Step S1: By analyzing the agricultural machinery migration time, a time cost evaluation function for agricultural machinery task migration is constructed; Step S2: constructing a node evaluation function for path planning based on the time cost evaluation function for agricultural machinery task migration constructed in step S1; Step S3: constructing a grid map based on the communication characteristics of edge nodes in the farmland map; Step S4: performing path planning for a single agricultural machine by using the node evaluation function constructed during path planning in step S2; Step S5: Combined with step S4, multiple machine paths are planned for multiple agricultural machines.

2. The method for multi-machine path planning for agricultural machinery with time-varying communication constraints as claimed in claim 1, characterized in that: The time cost evaluation function for agricultural machinery task migration constructed in step S1 is: Indicates the task in agricultural machinery am1 i The amount of data carried during migration, R load (T c ) is the CPU load rate of the node, is the amount of tasks waiting to be processed in the node, B mecx For edge nodes mec x Theoretical maximum bandwidth, P available is the computing resources available to the node, U mecx (t) is the network utilization, task i Bandwidth data transmission time T trans , C taski To meet the computing requirements of the task, the edge node mec x The computing power is P mecx , the calculated load rate is R load , T net.queuei It is the waiting time of the task in the network.

3. The method for multi-machine path planning for agricultural machinery with time-varying communication constraints as claimed in claim 2, characterized in that: The node evaluation function during path planning constructed in step S2 is: F(P[i]) is the final evaluation result of point P[i], G(P[i]) is the distance cost from the starting point to the point, H(P[i]) is the distance cost from the point to the end point, D(P[i]) is the evaluation of the network delay in step S1, which mainly evaluates the network communication delay based on the migration time of the task in the node, where w is the weight of the communication delay cost when evaluating the path, and its main function is to dynamically balance efficiency and security. P[i].mec represents the edge node number corresponding to the point, Belong mec Indicates the current edge node number during evaluation. When the two are equal, it means that the point is in the sequence number Belong mec Within the range of edge nodes, there is no need to consider the communication delay cost caused by task migration. When the two are not equal, it means that the point belongs to other edge nodes.

4. A method for multi-machine path planning for agricultural machinery with time-varying communication constraints as described in any one of claims 1 to 3, characterized in that: The specific steps of step S4 are as follows: Step S4.1: Mark the starting point and set the evaluation function value of the starting point to be the shortest distance from the starting point to the target point; Step S4.2: Search for the node with the minimum evaluation function value among the global open nodes of the grid graph, select the node with the minimum evaluation function, and remove it from the global open nodes; Step S4.3: Check whether the target is reached; Step S4.4: Calculate the relationship between the parent node and the current node, calculate the time from the parent node of the current node to the target point, and update the obstacle influence area of ​​each node according to the trajectory of the dynamic obstacle; Step S4.5: Sample obstacle trajectory and update; Step S4.6: Determine whether the obstacle affects the path; Step S4.7: Select successor nodes and update evaluation function: Select successor nodes around the current node and calculate their evaluation function values; Step S4.8: Continue searching until a path is found. If the evaluation function value of the successor node is less than that of its parent node, update the parent node and continue to execute step S4.

7. If the path is still not found, continue to select a new minimum evaluation node from the global open nodes. Step S4.9: End of path search: If the global open node is empty, it means that the path cannot be found, and "path cannot be found" is returned; otherwise, the shortest path from the starting point to the target point is finally returned.

5. The method for multi-machine path planning for agricultural machinery with time-varying communication constraints as claimed in claim 1, characterized in that: The specific steps of step S5 are as follows: Step S5.1: Set the number of agricultural machines currently required for path planning to AMNum; Step S5.2: perform path planning and sorting according to the independent operation capability of the agricultural machinery, and use the array AM[AMNum] to store the agricultural machinery information; Step S5.3: set index i=1; Step S5.4: P[start] = AM[i].Start, P[end] = AM[i].End, get PATH[i] through step S4, update the map MAP = MAP + PATH[i], i = i + 1. When i≠AMNum, continue to execute S5.4, otherwise execute S5.5; Step S5.5: Return the agricultural machinery multi-machine path planning path set PATH.

Images