A robot interaction collaboration method, system and storage medium

Abstract

The present application relates to the technical field of intelligent interaction, and more particularly to a robot interaction coordination method, system and storage medium. The method comprises the following steps: constructing a topological graph of a logistics environment, calculating the cut set appearance frequency and physical capacity parameters of nodes in the topological graph, and dividing the cut set appearance frequency by the physical capacity parameters to obtain a node congestion degree coefficient of the nodes; screening a global key node set according to the node congestion degree coefficient; calculating a task cut set after the robot receives a task, calculating the intersection of the task cut set and the global key node set to form a task bottleneck node set; calculating the path redundancy of each node in the task bottleneck node set; and the robot broadcasting a node occupation declaration comprising a node identifier, an estimated arrival time, an estimated departure time and a path redundancy. The present application uses path redundancy optimization decision-making to reduce unnecessary path re-planning, and realizes off-peak traffic through time elasticity allowance and speed adjustment to improve task continuity and system throughput.

Description

Technical Field

[0001] This invention relates to the field of intelligent interaction technology, and in particular to a robot interaction and collaboration method, system and storage medium. Background Technology

[0002] Through efficient collaboration and interaction among robots, complex logistics tasks such as warehousing, transportation, and sorting are completed. Core technologies include environmental perception and modeling, multi-robot task allocation and coordination, path planning and dynamic obstacle avoidance, communication and information sharing, human-robot interaction and control, and learning and optimization. Robots achieve environmental perception and dynamic modeling by integrating LiDAR, cameras, and ultrasonic sensors, combined with SLAM algorithms, to construct global and local maps to support navigation in complex scenarios. Task allocation employs centralized, distributed, or market-based auction algorithms, combined with cooperative game theory and distributed optimization to achieve robot collaboration.

[0003] However, in high-density logistics environments, robots are prone to congestion and conflicts when sharing channels or nodes. Traditional methods typically rely on static path planning or simple priority rules, which leads to inefficiency when robots wait or detour at nodes, limiting overall throughput. Secondly, traditional methods lack assessment of path redundancy, making it impossible for robots to determine the importance and detourability of nodes when performing tasks. When encountering conflicts, they often resort to blind replanning or stagnation. Summary of the Invention

[0004] Therefore, the present invention needs to provide a robot interaction and collaboration method, system and storage medium to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above objectives, a robot interaction and collaboration method includes the following steps:

[0006] Step S1: Construct a topology graph of the logistics environment, calculate the cut set occurrence frequency and physical capacity parameter of the nodes in the topology graph, divide the cut set occurrence frequency by the physical capacity parameter to obtain the node congestion coefficient; select the global key node set based on the node congestion coefficient.

[0007] Step S2: After receiving the task, the robot calculates the task cut set, and calculates the intersection of the task cut set and the global key node set to form the task bottleneck node set; calculate the path redundancy for each node in the task bottleneck node set.

[0008] Step S3: The robot broadcasts a node occupancy declaration including node identifier, estimated arrival time, estimated departure time, and path redundancy.

[0009] Step S4: The robot receives node occupancy declarations from other robots. When a time window conflict is detected, the robot calculates the path redundancy difference between the two robots. If the path redundancy difference is greater than the preset redundancy threshold, the robot performs path replanning. Otherwise, it adjusts the time elasticity margin and achieves off-peak passage by adjusting the speed.

[0010] The present invention also provides a robot interaction and collaboration system for executing the above-described robot interaction and collaboration method, the robot interaction and collaboration system comprising:

[0011] The bottleneck modeling module is used to construct a topology graph of the logistics environment, calculate the cut set occurrence frequency and physical capacity parameter of the nodes, divide the cut set occurrence frequency by the physical capacity parameter to obtain the node congestion coefficient, and select the global key node set based on the node congestion coefficient.

[0012] The bottleneck analysis module is used to calculate the task cut set after the robot receives the task, and to calculate the intersection of the task cut set and the global key node set to form the task bottleneck node set; and to calculate the path redundancy of each node in the task bottleneck node set.

[0013] The occupancy broadcast module is used to broadcast a node occupancy declaration, including the node identifier, estimated arrival time, estimated departure time, and path redundancy.

[0014] The conflict coordination module is used by the robot to receive node occupancy declarations from other robots. When a time window conflict is detected, it calculates the difference in path redundancy between the two robots. If the difference in path redundancy is greater than the preset redundancy threshold, the robot itself performs path replanning. Otherwise, it adjusts the time elasticity margin and achieves staggered passage by adjusting the speed.

[0015] The present invention also provides a computer storage medium storing a computer program, which, when executed, implements the robot interaction and collaboration method described in any of the above embodiments.

[0016] This invention achieves proactive identification and refined management of traffic bottlenecks by tightly coupling the structural characteristics of the logistics environment with robot task scheduling, thereby improving the overall system efficiency and predictability. First, by sampling and calculating the frequency of cut sets and combining it with physical capacity parameters to obtain the node congestion coefficient, the relative importance of each node under network connectivity and physical constraints can be objectively quantified, facilitating priority protection of key nodes and avoiding concentrated conflicts during the planning phase. Second, task-level bottleneck identification and path redundancy assessment allow each robot to obtain a measure of its path vulnerability upon receiving a task, enabling decisions on whether to replan or adjust arrival times based on redundancy differences. This redundancy-based decision-making rule balances efficiency and robustness, reducing large-scale replanning caused by immediate preemption. Furthermore, a distributed broadcast mechanism based on node occupancy declarations and a time-elastic speed adjustment and peak-shifting strategy allow robots to resolve short-term conflicts by fine-tuning their speed without frequently changing the global path, reducing communication and computational overhead and improving path stability. When replanning is indeed necessary, temporary re-searching to remove conflicting nodes and dynamic updates to the key node set ensure that the system maintains global bottleneck awareness after replanning. The method considers scalability and realistic constraints in details such as sampling and gridded point selection, physical capacity quantification, and path search length upper limit design, making it suitable for scenarios with large node scale and heterogeneous passage conditions. The introduction of time elasticity margin and safety interval ensures transportation efficiency while taking into account safety and controllability. Overall, this method, through structured bottleneck modeling, task-aware redundancy assessment, distributed occupancy coordination, and priority-driven replanning and speed adjustment, not only improves throughput and on-time task completion rate but also significantly enhances the system's adaptability to local congestion and unexpected situations. Thus, it provides an efficient, stable, and easily engineering-implemented solution for multi-robot collaboration in dense logistics environments. Attached Figure Description

[0017] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0018] Figure 1 This is a flowchart illustrating the steps of the robot interaction and collaboration method of the present invention;

[0019] Figure 2 This is a schematic diagram of cutset computation according to an embodiment of the present invention;

[0020] Figure 3 This is a detailed step-by-step diagram illustrating path redundancy calculation according to an embodiment of the present invention;

[0021] Figure 4 This is a flowchart of a conflict resolution decision tree according to an embodiment of the present invention;

[0022] Figure 5 This is a schematic diagram of the robot interaction and collaboration system of the present invention. Detailed Implementation

[0023] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0024] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0025] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0026] To achieve the above objectives, please refer to Figures 1 to 5 This invention provides a robot interaction and collaboration method applied to logistics robots, the method comprising the following steps:

[0027] Step S1: Construct a topology graph of the logistics environment, calculate the cut set occurrence frequency and physical capacity parameter of the nodes in the topology graph, divide the cut set occurrence frequency by the physical capacity parameter to obtain the node congestion coefficient; select the global key node set based on the node congestion coefficient.

[0028] Step S2: After receiving the task, the robot calculates the task cut set, and calculates the intersection of the task cut set and the global key node set to form the task bottleneck node set; calculate the path redundancy for each node in the task bottleneck node set.

[0029] Step S3: The robot broadcast includes a node occupancy declaration with node identifier, estimated arrival time, estimated departure time, and path redundancy.

[0030] Step S4: The robot receives node occupancy declarations from other robots. When a time window conflict is detected, the robot calculates the path redundancy difference between the two robots. If the path redundancy difference is greater than the preset redundancy threshold, the robot performs path replanning. Otherwise, it adjusts the time elasticity margin and achieves off-peak passage by adjusting the speed.

[0031] Furthermore, the calculation of the frequency of cut sets in step S1 includes:

[0032] Select multiple sampling points in the topology graph to form a sampling point set;

[0033] In one embodiment, several representative nodes are identified as sampling points in the established logistics environment topology map to form a sampling point set for subsequent cutset statistics and analysis. These sampling points can typically be selected based on topological characteristics, spatial distribution, or previous task data to cover different areas of the map (but the specific sampling point selection rules can be predetermined according to system needs).

[0034] For example, if the topology map contains 10 warehouse passage intersection nodes, all 10 intersection nodes can be directly used as the sampling point set to ensure that the statistics cover all passage hubs.

[0035] Select any two distinct sampling points from the set of sampling points as a pair of sampling points, and calculate the minimum cut set from one of the sampling points to the other. The minimum cut set is the smallest set of nodes that makes the pair of sampling points unconnected after removal.

[0036] In one embodiment, any two different sampling points in the sampling point set constitute a sampling point pair. Using one as the source and the other as the sink, the minimum cut set is calculated on the topology graph—that is, the minimum set of nodes that, when removed, makes the source and sink disconnected. To compute the node cut set, the node cut problem can be transformed into an equivalent maximum flow minimum cut problem, and a mature maximum flow algorithm (such as Dinic or Edmonds-Karp) can be called to obtain a minimum cut.

[0037] For example, if the sampling point pair is A and B, and the minimum cut of the topology graph is obtained by the algorithm as the set of nodes {N3, N7}, then the minimum cut set of (A, B) is considered to be this set of nodes.

[0038] Count the cumulative number of times each node appears in all minimum cut sets, and record it as the cumulative number of times the node appears;

[0039] In one embodiment, the implementation of counting the cumulative occurrences of each node in all minimal cut sets includes: for each pair in the sample point pair set, incrementing the counter of each node contained in the cut set according to the minimal cut set obtained in the previous step, and after traversing all sample point pairs, obtaining the cumulative occurrences of each node in all minimal cut sets.

[0040] For example, if there are 5 sampling point pairs, and node N3 appears in the minimum cut of the 1st, 3rd, and 5th pairs, then the cumulative number of occurrences of N3 is 3.

[0041] Divide the cumulative number of occurrences of each node by the total number of sampling point pairs to obtain the cut set occurrence frequency of that node.

[0042] In one embodiment, after calculating the minimum cut for all sampling point pairs and accumulating the occurrence count, the cut set occurrence frequency of each node is obtained by dividing the cumulative occurrence count of each node by the total number of sampling point pairs (i.e., the number of pairs obtained by counting unordered or ordered pairs in the sampling point set), which is used to quantify the importance of the node in the case of network cuts.

[0043] For example, if the total number of sampling point pairs is 45 and the cumulative occurrence of the corresponding node N3 is 9, then the frequency of the cut set of N3 is 9 / 45 = 0.2 (i.e. 20%).

[0044] It should be noted that the calculation of minimum cut is affected by the definition of node capacity and edge weight in the graph. If the “cut cost” or importance of the node needs to be taken into account in the implementation, the node capacity can be introduced into the maximum flow solution by splitting the node (converting the node into incoming and outgoing edges and adding the node capacity in the middle) or setting large weights on the edges. However, these implementation details should be consistent with the physical capacity parameters and weight definitions in the system.

[0045] For the case of a minimum cut with multiple solutions, if we want to reflect the probabilistic information of "occurring in any minimum cut", we can enumerate all equivalent minimum cuts in the implementation and perform more detailed counting based on the enumeration results (of course, the enumeration cost is high). Otherwise, we can use the single minimum cut returned by the deterministic algorithm and perform statistics.

[0046] To improve computational efficiency, the minimum cut of different sampling point pairs can be calculated in parallel, or the sampling point pairs can be calculated in batches and the node occurrence count can be updated incrementally to meet the needs of real-time or near-real-time bottleneck analysis.

[0047] See Figure 2 This demonstrates the concept of minimal cut sets and their application in topological networks. Figure 2 It consists of three parts: the left side shows the complete topology network, showing that the source node A and the sink node B are connected by multiple paths, and the intermediate nodes N1 to N7 provide different connectivity paths; the middle part is marked with the minimum cut set {N3, N7}, and these two key cut set nodes are marked with red dashed circles. These nodes represent the bottleneck positions in the network; the right side shows the network state after the cut set is removed. After the nodes N3 and N7 are removed, the connection between A and B is completely broken, proving that this is the minimum set of nodes that makes the two points unable to connect. Figure 2The bottom provides a legend: green circles represent source / sink nodes, gray circles represent regular nodes, yellow circles represent cut-set nodes, and red dashed circles represent removed nodes. This diagram visually illustrates that in a logistics robot system, cut-set nodes represent critical bottleneck locations. Removing these nodes will cause path planning failure, thus these nodes are highly important in the system.

[0048] Furthermore, the selection of the sampling point set includes:

[0049] Count the total number of nodes in the topology diagram of the logistics environment;

[0050] In one embodiment, counting the total number of nodes in the logistics environment topology graph can be achieved using a centralized or distributed graph data access method. The unique identifiers of all nodes are read from the established topology graph data structure, and these identifiers are traversed and counted to obtain the total number of nodes. This topology graph is typically stored in the form of an adjacency list, sparse matrix, or graph database; therefore, node counting can be accomplished by directly scanning the graph structure.

[0051] For example, if the system reads that the topology graph contains a total of 87 nodes from N1 to N87, then the total number of recorded nodes is 87.

[0052] When the total number of nodes is less than 100, all nodes in the topology graph are selected as sampling points to form a sampling point set.

[0053] In one embodiment, when the total number of nodes is less than 100, the system directly uses all nodes in the topology graph as sampling points to form a complete set of sampling points. During execution, the system copies the node list as a set of sampling points without any additional filtering or sampling, ensuring the most sufficient data coverage when calculating the frequency of cut sets.

[0054] For example, if the total number of nodes is 87, the system will add all 87 nodes to the sampling point set for use in the construction of subsequent sampling point pairs.

[0055] When the total number of nodes is greater than or equal to 100, the topology graph is divided into several grid cells according to the spatial coordinates of the nodes. At least one node is randomly selected in each grid cell as a sampling point of that grid cell, and the total number of sampling points is counted.

[0056] In one embodiment, the grid division step size is set according to the actual size of the logistics environment and the distribution range of node coordinates. Then, the two-dimensional space is divided into several rectangular grid areas according to the step size, and each node is assigned to the corresponding grid cell according to its coordinates. At least one node is randomly selected in each grid cell as a representative sampling point to ensure the uniformity of the sampling points in space.

[0057] For example, if a warehouse area is divided into 10×10 grids (100 grids in total) and the total number of nodes is 350, each grid may contain 1-7 nodes. The system randomly selects one node in each grid to obtain at least 100 initial sampling points.

[0058] The target number of sampling points is obtained by calculating the square root of the total number of nodes and rounding it up.

[0059] In one embodiment, the square root of the total number of nodes is calculated and rounded up to obtain the target number of sampling points. This strategy balances sampling scale with computational cost, ensuring that cutset analysis remains computable in large-scale topologies. During calculation, the system passes the total number of nodes as a parameter to the mathematical operation module, performs the square root operation, and rounds up.

[0060] For example, after removing 5 empty grids from the above 100 grids, a total of 95 sampling points are selected, so the current total number of sampling points is 95.

[0061] For example, if the total number of nodes is 350, its square root is approximately 18.7, which, when rounded up, gives a target sampling point count of 19.

[0062] Check if the total number of sampling points has reached the target number of sampling points. If not, perform supplementary selection to form a set of sampling points.

[0063] In one embodiment, it is checked whether the current total number of sampling points has reached the target number of sampling points. If not, supplementary selection is performed to form the final set of sampling points. Supplementary selection can be based on node distribution density, random sampling, or a strategy of prioritizing nodes in key areas. For example, if the current number of sampling points is 95, but the target number of sampling points is 19, it means that the initial sampling is much larger than the target number, and no supplementation is needed; however, if the area density is low, resulting in only 12 nodes in the initial sampling, the system continues to randomly sample nodes from the unselected node pool until the lower limit requirement of 19 sampling points is met.

[0064] For example, after obtaining 12 sampling points in the initial sampling, the system continues to randomly select 7 unselected nodes from all grids, bringing the number of sampling points to 19.

[0065] Furthermore, the calculation of the physical capacity parameters in step S1 includes:

[0066] Obtain the corresponding physical channel width and set speed limit value for each node in the topology graph;

[0067] In one embodiment, when constructing the logistics environment topology map, the physical space parameters associated with each node are recorded simultaneously, including the minimum passage width for robot passage, the maximum speed limit allowed within the node area, and node type (such as inflection point, intersection, narrow passage entrance, etc.). When calculating the physical capacity parameters, the system reads the above two key parameters one by one from the node attribute table.

[0068] For example, if a node is an intersection leading to the main road, its road width may be 2.1 meters, and the system sets the safe speed limit for this node to 1.2 m / s, then the system directly reads these two values ​​for subsequent calculations.

[0069] Obtain the robot's feature width and nominal speed;

[0070] In one embodiment, a robot equipment parameter table is maintained, which includes relevant information such as robot model, chassis width, maximum safe speed, and full-load speed. When calculating the physical capacity of a node, the system reads the characteristic width corresponding to the robot type performing the task and uses the robot's normal operating speed under unconstrained conditions as the nominal speed.

[0071] For example, if the robot chassis width is 0.55 meters and the nominal speed is 1.5 m / s, then the above two parameters will be used for subsequent calculations of parallel throughput and speed reduction factor.

[0072] The parallel throughput capability is obtained by dividing the physical channel width by the feature width and then rounding down.

[0073] In one embodiment, dividing the physical channel width by the characteristic width and rounding down to obtain the parallel passage capability is used to estimate the maximum parallel passage capability of the node in space. Dividing the node's physical channel width by the robot's characteristic width yields the maximum possible side-by-side travel value; since robots need to maintain a minimum distance in reality, the system performs a rounding down operation on the result to obtain a reliable parallel passage capability.

[0074] For example, if the channel width is 2.1 meters and the robot width is 0.55 meters, the calculated result is 3.81, which is rounded down to 3, meaning that a maximum of three robots can pass through the node in parallel.

[0075] Divide the speed limit by the nominal speed to obtain the speed reduction factor, which is used to represent the proportion of the robot's speed at that node that is reduced relative to the nominal speed.

[0076] In one embodiment, the speed limit value is divided by the nominal speed to obtain a speed reduction factor, which reflects the degree to which the robot's actual travel speed at that node decreases relative to its normal operating speed. The ratio is calculated by dividing the node's speed limit value by the robot's nominal speed; this value is typically between 0 and 1, representing the relative efficiency after the speed reduction.

[0077] For example, if the node speed limit is 1.2 m / s and the robot's nominal speed is 1.5 m / s, then the speed reduction factor is 0.8, which means that the robot's operating efficiency at that node is about 80% of the normal speed.

[0078] The product of parallel throughput capacity and speed reduction factor is used as the physical capacity parameter.

[0079] In one embodiment, the parallel throughput capacity multiplied by the speed reduction factor is used as the physical capacity parameter of the node, which comprehensively reflects the node's throughput capacity in both spatial and speed dimensions. The parallel throughput capacity P obtained in the preceding steps is multiplied by the speed reduction factor V to obtain the physical capacity parameter C = P × V. This parameter measures the node's capacity to handle high-traffic tasks and is used as the denominator in the cut set occurrence frequency division to reflect congestion sensitivity.

[0080] For example, if a node has a parallel throughput capacity of 3 and a speed reduction factor of 0.8, then the physical capacity parameter is 2.4; this result means that the node is equivalent to accommodating 2.4 passage units at 80% efficiency at the same time.

[0081] It should be noted that this physical capacity parameter is a continuous value and there is no requirement to round down. As a logical capacity indicator, the larger the value, the less likely the node is to become a bottleneck.

[0082] Furthermore, step S2, calculating the path redundancy for each node in the task bottleneck node set, includes:

[0083] Select one node from the set of task bottleneck nodes as the current evaluation node;

[0084] In one embodiment, when calculating path redundancy for each node in the task bottleneck node set, the set is traversed sequentially. During the traversal, a node is selected from the task bottleneck node set each time as the current evaluation node, and this node is marked as the core object of the redundancy evaluation in this round. This operation is typically implemented through index traversal, queue, or set iteration, enabling the system to calculate the bottleneck nodes one by one.

[0085] For example, assuming the bottleneck node set includes nodes N3, N7, and N12, the system will first select N3 as the current evaluation node, and after completing the path redundancy calculation for N3, it will then select N7 for subsequent evaluation.

[0086] Obtain the starting node and target node from the task received by the robot;

[0087] In one embodiment, after selecting the current evaluation node, the start node and target node of the task are read from the task description currently received by the robot. These two nodes are usually provided by the scheduling system when the task is issued, recording the starting position and final destination of the task.

[0088] For example, for a certain handling robot task, the starting node can be the warehouse shelving area entrance node N1, and the target node can be the outbound buffer node N20. This pair of nodes can be extracted through structured fields in the task object, task protocol data packets, or serializable task parameters to provide necessary boundary information for subsequent path search.

[0089] Starting from the initial node and ending at the target node, search for all shortest paths passing through the currently evaluated node in the topology graph, and count the number of paths as the number of direct paths.

[0090] In one embodiment, a path planning algorithm (such as Dijkstra's algorithm or BFS-based equal-weight shortest path search) is performed on the topology graph, with the task start node as the search starting point and the target node as the search ending point, and the search results are constrained to pass through the current evaluation node. Specifically, all shortest paths from the start node to the current evaluation node can be searched, and then all shortest paths from the current evaluation node to the target node can be searched. The two paths can then be combined to obtain a complete set of shortest paths.

[0091] Temporarily remove the current evaluation node and all edges connected to it from the topology graph, and re-search for feasible paths from the starting node to the target node. Count the number of paths that do not pass through the current evaluation node as the number of detour paths.

[0092] In one embodiment, the currently evaluated node and all edges connected to it are temporarily removed from the topology graph, rendering it "invalid" in the topology. The system then re-executes the path search from the starting node to the target node, but this time the search is not allowed to pass through the removed node.

[0093] Add the number of direct paths to a constant 1 to obtain the base number of direct paths; divide the number of detour paths by the base number of direct paths to obtain the path redundancy of the current evaluation node.

[0094] In one embodiment, after obtaining the number of direct paths and the number of detour paths, the path redundancy of the current evaluation node is calculated according to a preset formula. Specifically, the number of direct paths is first added to a constant 1 to obtain the base number of direct paths; then the number of detour paths is divided by this base number to obtain the path redundancy of the current node.

[0095] See Figure 3 When calculating the path redundancy of evaluation node N7, the system first performs a direct path search. In this stage, with task start point N1 as the search starting point and task end point N20 as the search ending point, the shortest path algorithm constrains the search results to pass through the current evaluation node N7. Since the shortest path from N1 to N7 in the topology graph is fixed as N1→N5→N7, and the shortest connecting path from N7 to N20 requires at least two additional edges, the length of the resulting complete path exceeds the global shortest distance. Therefore, there is no path in the illustrated structure that simultaneously satisfies both the conditions of "passing through N7" and "maintaining the shortest path length". Accordingly, the system records the number of direct shortest paths passing through N7 as 0, indicating that N7 is not a necessary point in the global shortest path.

[0096] Subsequently, the system enters the detour path search phase. At this point, node N7 and its connecting edges are temporarily removed from the graph, a local topology without N7 is constructed, and the shortest path search from N1 to N20 is re-executed. As shown in the graph, after N7 is removed, the topology maintains connectivity; the path N1→N5→N12→N13→N20 still achieves the shortest distance connection between the origin and destination. Therefore, in this example, the number of detour paths is 1. Combined with the result that the number of direct paths is 0, it can be determined that N7 is not a critical node; removing it still preserves the equivalent shortest path, making it less structurally sensitive in the overall network.

[0097] Through the two-stage search described above, the system can accurately identify and evaluate the path substitution capabilities of nodes under different combinations of task start and end points. This process not only quantifies the necessity of nodes in the shortest path but also provides a reliable data foundation for subsequent robot collaborative obstacle avoidance strategies. For example, in the conflict resolution stage, the path redundancy can be used to determine whether it is necessary to replan the path or implement deceleration to give way, thereby improving the overall passage efficiency and collaborative behavior stability of multiple robots in complex environments.

[0098] Of particular importance, the feasible search paths include:

[0099] Calculate the shortest path length from the starting node to the target node and record it as the baseline path length;

[0100] The maximum length of the calculated path search is equal to the base path length multiplied by a preset path extension factor;

[0101] Starting from the starting node, traverse all nodes connected to the starting node. During the traversal, accumulate the length of the current search path. When the accumulated length exceeds the upper limit of the path search length, terminate the search of that branch. When the search reaches the target node, record the search path as a feasible path.

[0102] Count the total number of feasible paths across all records.

[0103] Furthermore, step S4 includes the following steps:

[0104] Step S41: The robot receives node occupancy declarations broadcast by other robots, and extracts the node identifier, the other party's estimated arrival time, the other party's estimated departure time, and the other party's path redundancy from the received node occupancy declarations;

[0105] In one embodiment, robots establish a lightweight message subscription / publishing mechanism through a local area network or dedicated wireless communication channel. Each robot periodically or when its state changes publishes a node occupancy declaration to a public topic. The occupancy declaration message format uses structured fields (e.g., node identifier, robot ID, estimated arrival time, estimated departure time, path redundancy, timestamp, and signature / verification code). After receiving the message, the receiver first verifies the message integrity and timestamp validity, and then writes the extracted node identifier, the other party's estimated arrival time, the other party's estimated departure time, and the other party's path redundancy into the local conflict detection cache.

[0106] For example, when robot A plans to pass through node N, it broadcasts the message "Node N, expected arrival 10:12:05, expected departure 10:12:20, redundancy 0.3". After receiving the message, robot B inserts it into its local occupancy declaration list for later comparison.

[0107] It should be noted that packet loss or delay in communication may cause occupancy claims to become outdated. When implementing this system, the validity period of the claim should be retained and expired claims should be removed by combining timestamps and heartbeat mechanisms to avoid incorrect judgments.

[0108] Step S42: When a node identifier belongs to the task bottleneck node set of its own robot, obtain its own estimated arrival time, its own estimated departure time, and its own path redundancy from its own robot's occupation declaration of that node;

[0109] In one embodiment, when the received node identifier belongs to the set of bottleneck nodes of the robot's current task, the robot retrieves and prepares its estimated arrival time, estimated departure time and path redundancy for the corresponding node from the locally maintained occupancy declaration. Specifically, this is achieved by generating and continuously updating a local occupancy declaration table (containing the robot's estimated arrival / departure time and real-time speed estimate for each bottleneck node) during the task planning phase. When a declaration from another party is received and a node match is identified, the system simultaneously reads the node record in its own table and puts the data from both parties into a temporary buffer in the conflict determination process.

[0110] For example, if robot B's task bottleneck node set includes node N, after receiving robot A's declaration about node N, robot B reads its expected arrival time of 10:12:10 and expected departure time of 10:12:25 from its own declaration table and compares them with its own redundancy of 0.6.

[0111] Step S43: When the other party's estimated departure time is greater than one's estimated arrival time and the other party's estimated arrival time is less than one's estimated departure time, a time window conflict is identified;

[0112] In one embodiment, if the other party's expected departure time is greater than one's expected arrival time and the other party's expected arrival time is less than one's expected departure time, then the time windows are considered to overlap and thus a conflict is determined. In practice, the comparison is made with millisecond precision or a set minimum time resolution, and a safety interval parameter is introduced for boundary cases (for example, if the difference between the two is only less than the safety interval, it is also considered a conflict to ensure safety).

[0113] For example, if robot A is expected to leave at 10:12:20, robot B is expected to arrive at 10:12:15, and robot B is expected to leave at 10:12:30, then robot B's expected arrival (10:12:15) is less than robot A's expected departure (10:12:20), while robot A's expected departure (10:12:20) is greater than robot B's expected arrival (10:12:10), then a conflict is determined to exist.

[0114] Step S44: When there is a time window conflict, calculate the difference in path redundancy between the two parties; if the difference in path redundancy is greater than the preset redundancy threshold, remove the conflicting node in the topology map and the robot itself performs path replanning; otherwise, adjust the time elasticity margin and achieve staggered passage by adjusting the speed.

[0115] In one embodiment, the path redundancy is retrieved from the cache. Path redundancy with the other party Calculate the redundancy difference (Or use absolute difference combined with priority rules); if If the value is less than or equal to the preset redundancy threshold, then the time elastic margin adjustment and speed regulation peak-shifting strategy will be implemented; if If the value exceeds a preset threshold, the conflicting node and its connected edges are temporarily removed from the local topology graph representation, and the path replanning module is invoked to perform a new path search. During replanning, it should first be verified that the new path exists and satisfies the constraints (time, length, priority of key nodes). If a usable new path is found, the set of global key nodes on that path is replaced and recalculated. Then, the updated occupancy declaration is rebroadcast and subsequent steps are continued. If replanning fails, it reverts to the speed adjustment / waiting strategy and reports the failure to replan to the scheduling or upper-level coordination module.

[0116] For example, robot B collides with robot A, and the calculation is as follows: =0.4, and the threshold is set to 0.2. Then, robot B performs replanning: temporarily removes conflicting node N in the topology graph and searches for alternative paths. If a detour path is found, it updates its own path and broadcasts the new occupancy; otherwise, it calculates its own time elasticity margin and postpones the expected arrival time by reducing its speed to achieve off-peak passage.

[0117] Of particular importance, the path replanning specifically includes:

[0118] When the path redundancy difference is greater than the preset redundancy threshold, the conflicting node and all edges connected to it are temporarily removed from the topology graph.

[0119] After removing the node, the path from the task start node to the task target node is searched again in the topology graph to obtain the new driving path;

[0120] Check if a new travel path exists. If a new travel path exists, replace the robot's original travel path with the new travel path.

[0121] Based on the new driving path, recalculate the nodes that belong to the global critical node set among the nodes passed through, and re-execute steps S2 to S4.

[0122] Furthermore, the adjustment of the time flexibility margin in step S44 specifically refers to:

[0123] Extract the path segment from the robot's current position to the conflict node. The path segment contains several sub-path segments.

[0124] In one embodiment, after receiving the conflict node identifier and confirming the existence of a time window conflict, the robot first locates the path segment index of its current position from the currently planned execution path, and continues to search forward along the path until the conflict node is found; it then forms an ordered path segment by arranging all the continuous edges along the path from the current position to the conflict node. This path segment is usually composed of multiple sub-path segments, and each sub-path segment consists of two adjacent path nodes and their corresponding road attributes.

[0125] For example, if the robot is currently located at node A and the current path is A→B→C→D→E→F, and the conflict node is E, then the path segments are four sub-path segments: [A→B], [B→C], [C→D], and [D→E].

[0126] For each sub-path segment, determine the minimum speed and nominal speed based on the path attributes, and calculate the time difference between dividing the sub-path segment length by the minimum speed and dividing it by the nominal speed, which is used as the time extension of the sub-path segment.

[0127] In one embodiment, the path attributes corresponding to each sub-path segment are read (e.g., whether it is a narrow passage, whether it is a multi-lane road, whether it is located in an intersection area, whether there is a speed limit requirement, etc.). Based on the attributes, the minimum allowable speed of the sub-path segment and the nominal speed used in task planning are determined. Then, the length of the sub-path segment is calculated and divided by the minimum speed and the nominal speed respectively to obtain two time intervals. The difference between the two is used as the time extension of the sub-path segment, that is, the available time that can be delayed by reducing the speed while maintaining safety.

[0128] For example, if a sub-path segment is 2.0 meters long, has a nominal speed of 1.0 m / s and a minimum speed of 0.6 m / s, then the time extension of this segment is (2.0 / 0.6 - 2.0 / 1.0) ≈ 1.33 seconds.

[0129] The total time elasticity margin is obtained by summing the time extension of all sub-path segments.

[0130] In one embodiment, the time delays of each sub-path segment within the path segment are summed sequentially to obtain a scalar value representing the maximum time range that the robot can use to delay reaching the conflict node while maintaining a safe and controllable state. This value serves as an important criterion for determining whether the robot can achieve off-peak passage through speed adjustment.

[0131] For example, if the time extensions for the four sub-path segments are 1.2 seconds, 0.8 seconds, 0 seconds, and 1.1 seconds, respectively, then the total time elasticity margin is 3.1 seconds.

[0132] The overlap of the time windows is calculated by subtracting the larger of the expected arrival times from the smaller of the expected departure times of both parties.

[0133] In one embodiment, the expected arrival and departure times of the robot and the other robot at the conflict node are obtained, and the intersection length of their time intervals at that node is calculated. Specifically, the smaller of the expected departure times of both robots is taken, and the larger of the expected arrival times of both robots is taken. The difference between the two is used to obtain the time window overlap. If the value is greater than zero, it indicates that the time windows overlap, and the magnitude of the value indicates the duration of the overlap.

[0134] For example, if your estimated time to occupy the space is from 10:20:10 to 10:20:30, and the other party's estimated time to occupy the space is from 10:20:05 to 10:20:25, then the overlap is min(30,25)-max(10,5)=25-10=15 seconds.

[0135] If the total time margin is greater than the time window overlap, a speed reduction strategy will be adopted.

[0136] In one embodiment, the calculated total time margin is compared with the time window overlap. If the total time margin is greater than the time window overlap, the robot has the ability to delay its arrival at the conflict node by reducing its speed, thereby enabling a speed adjustment strategy to achieve off-peak passage. The specific speed adjustment process usually involves gradually reducing the robot's speed on the path segment from the nominal speed proportionally, so that the cumulative delay time reaches or slightly exceeds the overlap. If the total time margin is less than or equal to the time window overlap, it indicates that the speed reduction is insufficient to completely avoid the conflict, and a replanning strategy needs to be invoked to reroute the route.

[0137] For example, if the total time margin is 3 seconds and the overlap is 2 seconds, the robot can moderately reduce its speed in the next path segment to postpone the expected arrival time by more than 2 seconds to avoid the conflict.

[0138] Furthermore, the speed reduction strategy includes:

[0139] The amount of time that needs to be delayed is calculated by subtracting one's own estimated arrival time from the other party's estimated departure time and adding a 2-second safety interval.

[0140] In one embodiment, when the robot detects a bottleneck node with an overlapping time window with another robot, its scheduling module reads the other robot's estimated departure time at that node from the conflict data, and simultaneously reads the robot's estimated arrival time at that node. The system then calculates the required extension of travel time without changing its path using the formula: "Required delay time = Other robot's estimated departure time - Robot's estimated arrival time + 2 seconds".

[0141] For example, if the other robot is expected to leave a node at 10:15:20, and your own robot is expected to arrive at the node at 10:15:15, the calculated delay time is 5 seconds. Adding the preset 2-second safety buffer, the final delay is 7 seconds.

[0142] Determine a new average speed based on the total path length of the robot from its current position to the conflict node and the amount of time required to delay it;

[0143] In one embodiment, the system first calculates the total path length from the robot's current position to the conflict node along the currently planned path, and records this length as the remaining path distance. After obtaining the delay time, the system adds the delay time to the robot's originally estimated arrival time to obtain a new target arrival time. Subsequently, the system uses the calculation method "new average speed = remaining path distance / (new target arrival time - current time)" to determine the new average speed that the robot needs to maintain.

[0144] For example, if the robot is currently 4 meters away from the conflict node and the current time is 10:15:10, and the robot is required to arrive at 10:15:20 according to the amount of time delay, then the robot should travel 4 meters in 10 seconds, so the new average speed is about 0.4 meters per second.

[0145] Send a speed adjustment command to the robot's motion controller, travel at the new average speed, and rebroadcast after updating the expected arrival and departure times in the node occupancy declaration.

[0146] In one embodiment, the newly calculated average speed is used to generate a control command, which is then sent to the underlying motion control module, causing the robot to travel along the current path at the new average speed. Next, the system re-estimates the robot's arrival and departure times from conflict nodes based on the new speed and writes these times into a new node occupancy declaration, while retaining the previous path redundancy information. Subsequently, the robot broadcasts the new node occupancy declaration, enabling other robots to receive its updated time window information in real time and avoiding misjudgments.

[0147] For example, when a robot adjusts its speed from 1 meter per second to 0.4 meters per second, it will immediately report the new estimated arrival time to the system, such as updating it from the original estimated 10:15:15 to 10:15:20, and broadcast the updated time along with the occupancy declaration to other robots.

[0148] See Figure 4 This demonstrates the complete decision-making process for resolving conflicts between robots. The process begins with the detection of a conflict within a time window, first calculating the difference in path redundancy between the two robots. The first judgment node is evaluated. If the redundancy exceeds a preset threshold (e.g., 0.2): If yes, it indicates that the redundancy of the current device is significantly higher than that of the other device, and a path replanning strategy is executed, including three sub-steps: removing conflicting nodes, searching for new paths, and updating occupancy declarations; if no, it indicates that the redundancy of both devices is similar, and the second stage is to calculate the time elasticity margin. The second judgment node evaluates whether the time elasticity margin is greater than the time window overlap: if yes, a speed reduction strategy is adopted to adjust the speed to achieve off-peak passage; if no, it reverts to waiting for the other device to complete or reporting an anomaly to the scheduling center. Figure 4 The bottom provides three typical decision-making examples and key parameter explanations, including specific values ​​and decision results in different scenarios, to help understand the judgment logic of this collaborative decision-making mechanism in practical applications. This decision tree ensures that multi-robot systems can quickly and intelligently select the optimal solution when conflicts occur.

[0149] See Figure 5The present invention also provides a robot interaction and collaboration system 100 for executing the above-described robot interaction and collaboration method, the robot interaction and collaboration system comprising:

[0150] Bottleneck modeling module 101 is used to construct a topology graph of the logistics environment, calculate the cut set occurrence frequency and physical capacity parameter of the nodes, divide the cut set occurrence frequency by the physical capacity parameter to obtain the node congestion coefficient, and select the global key node set based on the node congestion coefficient.

[0151] The bottleneck analysis module 102 is used to calculate the task cut set after the robot receives the task, calculate the intersection of the task cut set and the global key node set to form the task bottleneck node set; and calculate the path redundancy for each node in the task bottleneck node set.

[0152] The occupancy broadcast module 103 is used to broadcast a node occupancy declaration, including node identifier, estimated arrival time, estimated departure time, and path redundancy.

[0153] The conflict coordination module 104 is used for the robot to receive node occupancy declarations from other robots. When a time window conflict is detected, it calculates the difference in path redundancy between the two robots. If the difference in path redundancy is greater than the preset redundancy threshold, the robot itself performs path replanning. Otherwise, it adjusts the time elasticity margin and achieves staggered passage by adjusting the speed.

[0154] The present invention also provides a computer storage medium storing a computer program, which, when executed, implements the robot interaction and collaboration method described in any of the above embodiments.

[0155] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0156] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A robot interaction and collaboration method, characterized in that, When applied to logistics robots, the following steps are included: Step S1: Construct a topology graph of the logistics environment, calculate the cut set occurrence frequency and physical capacity parameter of the nodes in the topology graph, and divide the cut set occurrence frequency by the physical capacity parameter to obtain the node congestion coefficient; select the global key node set based on the node congestion coefficient; the calculation of the cut set occurrence frequency includes: Select multiple sampling points in the topology graph to form a sampling point set; Select any two distinct sampling points from the set of sampling points as a pair of sampling points, and calculate the minimum cut set from one of the sampling points to the other. The minimum cut set is the smallest set of nodes that makes the pair of sampling points unconnected after removal. Count the cumulative number of times each node appears in all minimum cut sets, and record it as the cumulative number of times the node appears; Divide the cumulative number of occurrences of each node by the total number of sampling point pairs to obtain the cut set occurrence frequency of that node; The calculation of physical capacity parameters includes: Obtain the corresponding physical channel width and set speed limit value for each node in the topology graph; Obtain the robot's feature width and nominal speed; The parallel throughput capability is obtained by dividing the physical channel width by the feature width and then rounding down. Divide the speed limit by the nominal speed to obtain the speed reduction factor, which is used to represent the proportion of the robot's speed reduction at that node relative to the nominal speed. The product of parallel throughput capacity and speed reduction factor is used as the physical capacity parameter; Step S2: After receiving the task, the robot calculates the task cut set, and calculates the intersection of the task cut set and the global key node set to form the task bottleneck node set; calculate the path redundancy for each node in the task bottleneck node set. Step S3: The robot broadcasts a node occupancy declaration including node identifier, estimated arrival time, estimated departure time, and path redundancy. Step S4: The robot receives node occupancy declarations from other robots. When a time window conflict is detected, the robot calculates the path redundancy difference between the two robots. If the path redundancy difference is greater than the preset redundancy threshold, the robot performs path replanning. Otherwise, it adjusts the time elasticity margin and achieves off-peak passage by adjusting the speed.

2. The robot interaction and collaboration method according to claim 1, characterized in that, The selection of the sampling point set includes: Count the total number of nodes in the topology diagram of the logistics environment; When the total number of nodes is less than 100, all nodes in the topology graph are selected as sampling points to form a sampling point set. When the total number of nodes is greater than or equal to 100, the topology graph is divided into several grid cells according to the spatial coordinates of the nodes. At least one node is randomly selected in each grid cell as a sampling point of that grid cell, and the total number of sampling points is counted. The target number of sampling points is obtained by calculating the square root of the total number of nodes and rounding it up. Check if the total number of sampling points has reached the target number of sampling points. If not, perform supplementary selection to form a set of sampling points.

3. The robot interaction and collaboration method according to claim 2, characterized in that, Step S2 involves calculating the path redundancy for each node in the task bottleneck node set, including: Select one node from the set of task bottleneck nodes as the current evaluation node; Obtain the starting node and target node from the task received by the robot; Starting from the initial node and ending at the target node, search for all shortest paths passing through the currently evaluated node in the topology graph, and count the number of paths as the number of direct paths. Temporarily remove the current evaluation node and all edges connected to it from the topology graph, and re-search for feasible paths from the starting node to the target node. Count the number of paths that do not pass through the current evaluation node as the number of detour paths. Add the number of direct paths to a constant 1 to obtain the base number of direct paths; divide the number of detour paths by the base number of direct paths to obtain the path redundancy of the current evaluation node.

4. The robot interaction and collaboration method according to claim 3, characterized in that, Step S4 includes the following steps: Step S41: The robot receives node occupancy declarations broadcast by other robots, and extracts the node identifier, the other party's estimated arrival time, the other party's estimated departure time, and the other party's path redundancy from the received node occupancy declarations; Step S42: When a node identifier belongs to the task bottleneck node set of its own robot, obtain its own estimated arrival time, its own estimated departure time, and its own path redundancy from its own robot's occupation declaration of that node; Step S43: When the other party's estimated departure time is greater than one's estimated arrival time and the other party's estimated arrival time is less than one's estimated departure time, a time window conflict is identified; Step S44: When there is a time window conflict, calculate the difference in path redundancy between the two parties; if the difference in path redundancy is greater than the preset redundancy threshold, remove the conflicting node in the topology map and the robot itself performs path replanning; otherwise, adjust the time elasticity margin and achieve off-peak passage by adjusting the speed.

5. The robot interaction and collaboration method according to claim 4, characterized in that, The adjustment time margin mentioned in step S44 is specifically as follows: Extract the path segment from the robot's current position to the conflict node. The path segment contains several sub-path segments. For each sub-path segment, determine the minimum speed and nominal speed based on the path attributes, and calculate the time difference between dividing the sub-path segment length by the minimum speed and dividing it by the nominal speed, which is used as the time extension of the sub-path segment. The total time elasticity margin is obtained by summing the time extension of all sub-path segments. The overlap of the time windows is calculated by subtracting the larger of the expected arrival times from the smaller of the expected departure times of both parties. If the total time margin is greater than the time window overlap, a speed reduction strategy will be adopted.

6. The robot interaction and collaboration method according to claim 5, characterized in that, The speed reduction strategy includes: The amount of time that needs to be delayed is calculated by subtracting one's own estimated arrival time from the other party's estimated departure time and adding a 2-second safety interval. Determine a new average speed based on the total path length of the robot from its current position to the conflict node and the amount of time required to delay it; Send a speed adjustment command to the robot's motion controller, travel at the new average speed, and rebroadcast after updating the expected arrival and departure times in the node occupancy declaration.

7. A robot interaction and collaboration system, characterized in that, For performing the robot interaction and collaboration method as described in claim 1, the robot interaction and collaboration system includes: The bottleneck modeling module is used to construct a topology graph of the logistics environment, calculate the cut set occurrence frequency and physical capacity parameter of the nodes, divide the cut set occurrence frequency by the physical capacity parameter to obtain the node congestion coefficient, and select the global key node set based on the node congestion coefficient. The bottleneck analysis module is used to calculate the task cut set after the robot receives the task, and to calculate the intersection of the task cut set and the global key node set to form the task bottleneck node set; and to calculate the path redundancy of each node in the task bottleneck node set. The occupancy broadcast module is used to broadcast a node occupancy declaration, including the node identifier, estimated arrival time, estimated departure time, and path redundancy. The conflict coordination module is used by the robot to receive node occupancy declarations from other robots. When a time window conflict is detected, it calculates the difference in path redundancy between the two robots. If the difference in path redundancy is greater than the preset redundancy threshold, the robot itself performs path replanning. Otherwise, it adjusts the time elasticity margin and achieves staggered passage by adjusting the speed.

8. A computer storage medium storing a computer program, characterized in that, When the computer program is executed, it implements the robot interaction and collaboration method as described in any one of claims 1 to 6.

Images