Dynamic capability grading traffic control method for heterogeneous robots

Abstract

The invention discloses a heterogeneous robot-oriented dynamic capability grading traffic management method, which comprises the following steps of: S1, establishing a unified scheduling platform, and receiving capability registration information of a heterogeneous robot; s2, defining a capability type of the robot, wherein the capability type is determined by a non-empty subset combination of a basic capability feature set and an expansion capability feature set; s3, the unified scheduling platform dynamically generates a decision scheme according to the registration capability type of the robot and the path conflict; s4, the queuing condition is monitored, the queue position is issued to the non-read-only robot, and the robot queues and waits at a certain interval and speed in sequence; s5, detecting and identifying a fault robot, delimiting a fault isolation area, and executing avoidance according to a capability type; s6, monitoring a deadlock condition, and executing a deadlock removing strategy according to a capability type; and S7, fusing robot sensor data, and cooperatively executing avoidance by a platform end and a local end. And through dynamic capability grading and a targeted scheduling strategy, the operation efficiency and safety of the heterogeneous robot and the overall performance of the system are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unified robot scheduling, and in particular to a dynamic ability grading traffic control method for heterogeneous robots. Background Art

[0002] With the large-scale application of multi-brand and multi-type robots in scenarios such as logistics warehousing and industrial manufacturing, when multi-brand and multi-type robots work together, there are problems such as inconsistent communication protocols and large differences in capabilities. The traditional traffic control system faces pain points, mainly including: different protocols and port opening degrees of multi-brand and multi-type robots, resulting in difficult data interaction, and some brand robots cannot open the ports for unified scheduling due to technical protection and cannot receive unified scheduling instructions, etc., so it is impossible to make multi-brand and multi-type robots achieve traffic control according to a single decision-making mechanism; the existing system can only handle robots of a single brand or with similar technical capabilities, and it is difficult to implement differential scheduling strategies for robots with different operation control capabilities (such as pause, avoidance, dynamic path replanning, etc.); the traditional avoidance mechanism relies on the local decision-making of robots, such as stopping and detouring, etc., and it is easy to have instruction conflicts or response delays, especially in high-density traffic scenarios, it is easy to cause deadlock phenomena, such as oncoming vehicles stopping and detouring at the same time, resulting in deadlock, etc.; when multi-brand and multi-type robots form a mixed formation to execute tasks, there is a lack of an effective hierarchical scheduling mechanism, and the advantages of robots with different capabilities cannot be fully utilized to achieve the overall optimal efficiency, etc. The defects of the existing technology lead to serious insufficient applicability of the robot traffic control system in multi-brand, multi-type, and high-dynamic complex scenarios, and it is difficult to make full use of the performance advantages of different robots, and it is difficult to meet the comprehensive requirements for traffic efficiency, safety, and system robustness in practical applications.

[0003] Therefore, it is necessary to develop a dynamic ability grading traffic control method for heterogeneous robots to solve the above defects. Summary of the Invention

[0004] Aiming at the technical problems existing in the above-mentioned existing technology, the purpose of this application is to provide a dynamic ability grading traffic control method for heterogeneous robots, and the specific steps are as follows: S1: Establish a unified scheduling platform to receive the ability registration information of heterogeneous robots, and the ability registration information at least includes the basic parameters, functional characteristics, communication protocols, etc. of the robots; S2: Define the ability types of robots, and the ability types are determined by the combination of non-empty subsets of the basic ability feature set and the extended ability feature set: The basic ability feature set at least includes: information reporting (I), pause instruction response (P), resume instruction response (R); The extended ability feature set at least includes: avoidance point navigation (A), dynamic path replanning (D), sensor data sharing (S), and environmental interaction (E); The ability classification rules include but are not limited to: If the ability feature subset contains {I}, but does not contain {P, R, A, D, S, E}, it is defined as read-only. If the ability feature subset contains {I, P, R}, but does not contain {A, D, S, E}, it is defined as basic. If the ability feature subset contains {I, P, R, A}, and at least contains one of {S, E}, it is defined as standard. If the ability feature subset contains {I, P, R, A, D}, and at least contains one of {S, E}, it is defined as intelligent. S3: The unified scheduling platform dynamically generates a decision-making plan according to the robot's registered ability type and path conflicts, specifically including: S3a: Prediction stage: Calculate the spatio-temporal conflict probability map of the robot encounter position, and perform avoidance according to the ability type, including but not limited to: Read-only type: Drive along the original path; Above basic type: Insert a pause instruction upstream of the conflict point; Above standard type: Insert a pause instruction upstream of the conflict point, or drive to the specified avoidance point for avoidance; Intelligent type: Insert a pause instruction upstream of the conflict point, or drive to the avoidance point for avoidance, or drive along the route re-planned by the unified scheduling platform to bypass the conflict area.

[0005] S3b: Conflict response stage: Real-time detect path conflicts and coordinate according to the conflict level, including but not limited to: Potential conflict: Reduce the speed of the robot, and high-priority tasks pass first; Emergency conflict: Send an emergency stop instruction to the conflicting robots that are not read-only, and generate a bypass route for intelligent robots.

[0006] S1 further includes that the unified scheduling platform processes heterogeneous systems through an adaptation layer, and the adaptation layer includes: a protocol conversion engine that converts the native protocols of each brand of robots into a unified communication protocol, an ability registry that dynamically records the robot's ability type, and a spatio-temporal unified coordinate system that fuses multi-source positioning data into a unified map.

[0007] S3a further includes: Taking the path intersection point as a node, calculate the time window for each robot to arrive; If the time windows of two robots overlap and the spatial distance is less than or equal to the safety distance, mark the conflict, and the safety distance is dynamically adjusted according to the positioning accuracy.

[0008] The S3b further includes: Real-time detection of path intersections. When the distance between a conflicting robot and the intersection is less than or equal to the dynamic safety distance, it is upgraded to an emergency conflict.

[0009] Preferably, this solution also includes queuing management: S4: In the queuing waiting area, the robot makes a reservation application to the unified scheduling platform for the right to use the resources ahead. The unified scheduling platform sorts the reservation applications according to the position order of the reserved robots. When a robot with a high priority is waiting for the right to use, the unified scheduling platform issues a queue position instruction to the subsequent non-read-only robots, and the robots queue up at a certain interval distance and speed in turn.

[0010] Preferably, this solution also includes S5 fault handling: The unified scheduling platform detects and identifies the faulty robot and demarcates a fault isolation area, specifically including: Generating a dynamic detour path for the intelligent robot and inserting a deceleration instruction for the upstream path for the non-read-only robot; When the faulty robot recovers, after repositioning, it verifies the integrity of the task and generates a continuous path.

[0011] Preferably, this solution also includes S6 deadlock management: S6a: Deadlock detection: The unified scheduling platform uses a deadlock detection algorithm to detect whether there is a deadlock situation in the robot system, including but not limited to: constructing a resource-robot relationship graph, establishing a dynamic relationship, a periodic evaluation mechanism, event-driven detection, path tracking analysis, and cyclic dependency judgment; S6b: Deadlock resolution: When a deadlock is detected, according to the sequence of robots and resources involved, select the nearest intelligent robot as the leading vehicle to generate a formation passing sequence; Designate an avoidance point or a straight avoidance area greater than or equal to the body safety length for the standard robot.

[0012] Preferably, this solution also includes S7 sensor fusion avoidance: Robots with the ability feature subset containing {S} share sensor feature-level data to the unified scheduling platform; Platform side: The unified scheduling fuses the data to construct a global dynamic obstacle map and performs hierarchical avoidance: Intelligent type: Dynamic detour; Standard type: Designate an avoidance point; When dense dynamic obstacles are detected, for robots with the ability feature subset containing {E}, activate area speed limit and audible and visual alarms; Local side: The robot local side performs avoidance actions according to the original sensor data.

[0013] Preferably, the protocol conversion engine supports bandwidth adaptive compression, switches to low-precision positioning data transmission during communication latency, and the spatio-temporal unified coordinate system performs timestamp synchronization at set time intervals.

[0014] Preferably, the feature-level data includes: obstacle bounding boxes, motion vectors, semantic labels with a confidence level greater than or equal to a preset value; when the density of dynamic obstacles is greater than a preset density, the platform divides a low-speed operation area with speed limits.

[0015] Compared with the prior art, the method for dynamic capacity grading traffic control of heterogeneous robots of the present invention has the following main advantages or beneficial effects: Effectively manage heterogeneous robot traffic through dynamic capacity grading and targeted scheduling; Formulate personalized decision-making schemes according to the differences in robot capabilities, and improve the system compatibility and task execution efficiency; Set up a path conflict prediction and response mechanism, which can adjust the robot behavior in real time, reduce the collision risk, and ensure the operation safety; Functions such as queuing management, fault handling, deadlock resolution, and sensor fusion avoidance further optimize the robot collaborative work process and enhance the system stability and reliability.

[0016] The solution of the present invention comprehensively improves the operation efficiency and intelligent level of multi-brand and multi-type robots in complex scenarios, and has significant innovation value and practical significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The figure shows a schematic flow diagram of the method of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention; It should be noted that the terms used herein are only for describing the specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, components or modules, assemblies and / or their combinations; It should be noted that the terms "including" and "having" and any variations thereof in the description, claims and above-mentioned drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices. It should be understood that the solution of the present invention can be implemented by a single or a combination of multiple of hardware, software or other devices. In the description of the following embodiments, the methods and steps of the present invention can be implemented by being stored in a storage device, including but not limited to hard disks, removable storage devices, magnetic disks, optical disks, etc. It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. Embodiment

[0019] As Figure 1 shown, a dynamic capability grading traffic control method for heterogeneous robots includes the following steps: S1: Build a unified scheduling platform: S1a: Receive capability registration information: The unified scheduling platform receives the capability registration information sent by heterogeneous robots through a communication network. This information at least includes the basic parameters of the robots (such as size, weight, maximum speed, etc.), functional characteristics (such as sensor type, navigation accuracy, etc.) and communication protocols (such as native protocols like TCP / IP, UDP, etc.).

[0020] S1b: Processing by the adaptation layer: Protocol conversion engine: Adopt a protocol conversion algorithm to convert the native protocols of robots of each brand into a unified communication protocol. For example, for brand A robots using the TCP / IP protocol and brand B robots using the UDP protocol, the protocol conversion engine, according to the preset conversion rules, realizes operations such as data format conversion and transmission control between the two protocols to ensure the compatibility and stability of communication between the unified scheduling platform and each robot.

[0021] Preferably, the protocol conversion engine monitors the communication bandwidth. When the bandwidth is insufficient or there is communication delay, it adaptively compresses the transmitted positioning data (such as high-precision maps, sensor image data, etc.) using an image compression algorithm (such as JPEG, H.264, etc.) to reduce the data volume and ensure the real-time nature of data transmission.

[0022] Capability registration table: Store the registered capability types of each robot in a dynamic recording manner. When a robot registers or updates its capability information, the capability registration table updates the record in real time to provide a basis for subsequent scheduling decisions.

[0023] Space - time unified coordinate system: Integrate multiple positioning data (such as lidar, camera data, etc.) into a unified map. Through coordinate conversion algorithms, convert the robot position information obtained from different positioning systems into a unified space - time coordinate system to ensure that the unified scheduling platform can accurately master and uniformly manage the positions of robots.

[0024] Preferably, the space - time unified coordinate system performs timestamp synchronization at set time intervals (such as every second or every 10 seconds, etc.), and adopts time synchronization algorithms (such as NTP, PTP and other protocols, etc.) to ensure the time consistency between each robot and the unified scheduling platform, and ensure the accuracy of time - related data such as robot position information and scheduling instructions.

[0025] S2: Define the types of robot capabilities: S2a: Determine the basic ability feature set and the extended ability feature set: The basic ability feature set includes but is not limited to: information reporting (I), pause instruction response (P), resume instruction response (R), etc. For example, a robot can report its own status information (such as position, speed, battery level, etc.) to the unified scheduling platform (I), and can respond to the pause (P) and resume (R) instructions sent by the platform; The extended ability feature set includes but is not limited to: avoidance point navigation (A), dynamic path replanning (D), sensor data sharing (S), environmental interaction (E), etc. For example, a robot can navigate to a specified avoidance point according to the platform instruction (A), can replan the path based on real - time environmental information or drive according to the path replanned by the platform (D), can share sensor data with other robots or the platform (S), and can interact with objects in the environment (such as shelves, elevators, etc.), sound and light alarm, etc. (E).

[0026] S2b: Ability classification rules: In this embodiment, in order to facilitate the distinction of different types of robots for subsequent different avoidance strategies, robots are classified according to the ability registration table, including but not limited to: Read - only type: If the ability feature subset contains {I}, but does not contain {P, R, A, D, S, E}, that is, the robot only has the information reporting (I) ability and does not have other abilities, it is classified as a read - only type. Such robots can only drive along the preset path, cannot execute complex avoidance and scheduling instructions, and the unified scheduling platform cannot control read - only robots; Basic type: If the ability feature subset contains {I, P, R}, but does not contain {A, D, S, E}, that is, the robot has the information reporting (I), pause instruction response (P), and resume instruction response (R) abilities, but does not have the abilities in the extended ability feature set, it is defined as a basic type. Such robots can perform simple pause / resume operations in case of conflicts; Standard type: If the subset of ability characteristics contains {I, P, R, A} and at least one of {S, E}, that is, the robot has the capabilities of information reporting (I), pause instruction response (P), resume instruction response (R), avoidance point navigation (A), and at least one of sensor data sharing (S) or environment interaction (E), it is a standard type. When a conflict occurs, such robots can, in addition to pausing, also drive to a designated avoidance point to avoid; Clever type: If the subset of ability characteristics contains {I, P, R, A, D} and at least one of {S, E}, that is, the robot has the capabilities of information reporting (I), pause instruction response (P), resume instruction response (R), avoidance point navigation (A), dynamic path replanning (D), and at least one of sensor data sharing (S) or environment interaction (E), it is a clever type. Such robots can perform the most flexible avoidance operations, such as driving along the route replanned by the platform and bypassing the conflict area; It should be noted that the above are only the ability types listed in this embodiment. Other ability types can be divided according to the needs of specific embodiments, and they have the same technical effects, which will not be elaborated here.

[0027] S3: The unified scheduling platform generates a dynamic decision-making plan based on the robot ability type and path conflict: S3a: Prediction stage: Calculate the spatio-temporal conflict probability map: Taking the path intersection as a node, establish a mathematical model to calculate the time window for each robot to reach the intersection: , Formula 1, where, t arrive : The time when the robot is expected to reach the intersection, calculated based on the robot's current position, speed, and path planning; Δt: Time deviation, determined considering factors such as robot speed fluctuation and positioning error, If the time windows of two robots overlap and the spatial distance is less than or equal to the safety distance, mark the conflict and dynamically adjust the safety distance: , Formula 2, where, k: Safety factor, set according to factors such as the robot's operating environment and speed, d precision : Positioning accuracy, determined by the performance of the positioning system.

[0028] Avoidance strategy execution: Execute avoidance according to the ability type, including but not limited to: Read-only type: Drive along the original path; Above basic type: Insert a pause instruction upstream of the conflict point; Above standard type: Insert a pause instruction upstream of the conflict point, or drive to a designated avoidance point to avoid; Intelligent type: Insert a pause instruction upstream of the conflict point, or drive to the avoidance point for avoidance, or drive along the route re-planned by the unified scheduling platform to bypass the conflict area. It should be noted that the avoidance strategy can be formulated according to the ability types of different robots, or different avoidance strategies from this solution can be formulated, which have the same technical effects and will not be elaborated here.

[0029] It should be noted that the path planning algorithm of the unified scheduling platform can adopt commonly used or optimized A*, D*, BFS, DFS, etc. in the field, and the appropriate algorithm can be selected according to the actual calculation situation.

[0030] S3b: Conflict response stage: Real-time detection and conflict escalation determination: Real-time detect the robot path intersection point. When the distance between the conflicting robot and the intersection point is less than or equal to the dynamic safety distance, it is upgraded to an emergency conflict. The calculation formula of the dynamic safety distance is the same as the above safety distance (Formula 2), but the k value can be appropriately increased in the case of an emergency conflict to improve safety and will not be elaborated here.

[0031] Coordination measure execution: Coordinate according to the conflict level, including but not limited to: Potential conflict: Reduce the speed of the robot, and the high-priority task passes first; Emergency conflict: Send an emergency stop instruction to the non-read-only conflicting robot and generate a bypass route for the intelligent robot; If the above coordination measures still cannot solve the problem, trigger an alarm and manually intervene to resolve the conflict.

[0032] S4: Queue management: In some operating scenarios, when queue robots all need or are about to apply for the same resource, they need to queue up: In the queuing area, the robot reserves and applies for the right to use the front resource (such as the right to pass through the area, the right to use the elevator, etc.) from the unified scheduling platform. The unified scheduling platform sorts the reservation applications according to the position order of the reserved robots. When the robot with a high priority is waiting for the right to use, the unified scheduling platform issues queue position instructions (position coordinates, pause, resume, etc.) to the subsequent non-read-only robots. The robots queue up at a certain interval distance and speed in turn to ensure the orderly progress of the queuing process and avoid congestion, deadlock, etc. caused by the subsequent entering robots bypassing, etc.; For the uncontrollable read-only robots, according to their status and position information, other non-read-only robots queue up before and after them. If congestion, deadlock, etc. are caused by the bypass of the read-only robot, trigger an alarm and manually intervene to resolve it.

[0033] S5: Fault handling: If a heterogeneous robot in the fleet has a fault, handle it according to the following process: S5a: Fault identification and isolation: The unified dispatching platform identifies faulty robots by monitoring robot status information (such as heartbeat signals, fault codes, etc.). When demarcating the fault isolation zone, it considers factors such as the position, size, and surrounding environment of the faulty robot to determine the range of the isolation zone; it generates dynamic detour paths for smart robots, uses path planning algorithms to avoid the fault isolation zone, and ensures the optimality of the path; it inserts upstream path deceleration instructions for non-read-only robots, and calculates reasonable deceleration values based on factors such as the distance and speed between the robot and the fault point to ensure that the robot passes safely or waits for the fault to be cleared; S5b: Fault recovery and task continuation: When the faulty robot recovers, the task integrity is verified after repositioning. By comparing the task information and location information before and after the fault, the remaining part of the task is determined and a continuation path is generated so that the robot can continue to perform the unfinished task from the fault point.

[0034] S6: Deadlock management: In multi-robot traffic control scenarios, deadlock may occur when multiple robots are unable to move due to path planning conflicts, uncoordinated avoidance actions, etc., and fall into a deadlock state where they wait for each other to avoid but do not take effective avoidance measures themselves, making it impossible to complete the established work process normally.

[0035] Therefore, in a multi-robot operation scenario, it is necessary to monitor the deadlock status and release the deadlock through the system, as follows: S6a: Deadlock detection: The unified scheduling platform uses a deadlock detection algorithm to detect whether the robot system is deadlocked, including whether there is a deadlock and the robot and resource sequence involved in the deadlock. In this embodiment, the deadlock detection algorithm is as follows. Those skilled in the art can also identify according to other deadlock detection algorithms according to computing requirements. The algorithm of this embodiment is as follows: (1) Construct a resource-robot relationship graph: Construct a directed graph G = (V, E), where the vertex set V contains robot vertices R = {R1, R2, …, Rn} and resource vertices H = {H1, H2, …, Hm}. The robot vertices represent the various robot entities in the system, and the resource vertices represent the various resources that the robot may need to access. (2) Establish dynamic relationships: By monitoring the interaction between robots and resources, the resource-robot relationship graph is dynamically updated. When a robot applies for or releases a resource, the corresponding edge is added or removed from the graph G. i Application Resource H j , add directed edge R i →H j ; If resource H j Assigned to robot R i , add a directed edge H j →R i;Define the edge where the robot applies for resources as the request edge, and the edge where resources are allocated to the robot as the allocation edge; (3) Periodic evaluation mechanism: Establish a periodic evaluation mechanism to analyze the resource-robot relationship graph regularly to identify potential deadlock states. The evaluation period can be adjusted according to the system scale and complexity to ensure timely detection of deadlock risks. The formula is as follows: , Formula Three, where, T min : The minimum value of the evaluation period, T max : The maximum value of the evaluation period, N: The number of robots in the system, α is the adjustment coefficient, N threshold : The threshold of the number of robots.

[0036] (4) Event-driven detection: Set up an event-driven detection mechanism. When key events such as a robot applying for or releasing resources and the task status changing occur, immediately start the deadlock detection process to ensure rapid response to system state changes and improve the timeliness of deadlock detection.

[0037] (5) Path tracing analysis: Use the depth-first search (DFS) algorithm to trace the paths of resource applications and allocations starting from each unexamined node in the resource-robot relationship graph. Initialize the set of visited nodes V visited = , Initialize DFS(v0), and the DFS recursive process is as follows: , Formula Four, where, V visited : The set of visited nodes, V0: The starting node, V: The current node, N(v): The set of all adjacent nodes of node V, : The difference set operation of the sets, indicating to select from the unvisited adjacent nodes, w: The adjacent node, DFS(v): The recursive process of depth-first search. If node v has been visited, return; otherwise, visit node v, mark it as visited, and recursively call DFS for all unvisited adjacent nodes v.

[0038] (6) Circular dependency judgment: If one of the robots indirectly or directly waits for the resources it already occupies, it indicates a deadlock risk. Trigger deadlock resolution and record the sequence of robots and resources involved as the input to the deadlock resolution strategy. The formula is as follows: , , Formula Five, where, P: The access path, v0, v1,..., v k : The nodes in the path, i, j: The position indices in the path. If there exists i < j and vi = v j indicates that in the path, node v j has been visited, indicating that there is a cycle. Cycle: The result of cycle dependency judgment. If the value is True, it means that a cycle dependency is found during the search process. If the value is False, it means that no cycle dependency is found.

[0039] Furthermore, before deadlock detection, resources are pre-classified to distinguish critical resources and non-critical resources. Higher priority is given to the application and allocation of critical resources, and stricter monitoring is carried out. In deadlock detection, a hierarchical detection mechanism is adopted. First, the robots and resource nodes related to critical resources are detected, and then the detection is extended to the entire system to improve the detection efficiency while ensuring the stable operation of critical system functions.

[0040] S6b: Deadlock resolution strategy: When a deadlock is detected, according to the sequence of robots and resources involved, the nearest intelligent robot is selected as the leading vehicle to generate a formation passage sequence. Based on factors such as the positions and speeds of each robot, the leading vehicle is determined, and a guiding path is planned for it to drive other robots to resolve the deadlock. Avoidance points or straight-line avoidance areas greater than or equal to the body safety length are preset for standard robots to ensure that standard robots have safe avoidance spaces and paths during deadlock resolution and to avoid new conflicts. If the unified scheduling platform cannot resolve the deadlock, an alarm is triggered and manual intervention is required to resolve it.

[0041] S7: Sensor fusion avoidance: In some embodiments, the unified scheduling platform can fuse the sensors uploaded by the robots and formulate a collaborative avoidance mechanism for the platform side and the local side, as follows: S7a: Sensor data sharing and fusion: Robots with the ability feature subset {S} share the feature-level data of the sensors to the unified scheduling platform, convert the data in different coordinate systems to the unified coordinate system, and the platform uses data fusion algorithms, such as the Kalman filter fusion algorithm, etc., to fuse the sensor data from different robots and construct a global dynamic obstacle map.

[0042] Furthermore, the feature-level data includes the obstacle border (a rectangle or irregular shape frame surrounded by coordinate points, indicating the position and range of the obstacle in space), the motion vector (a vector containing information such as the motion direction and speed of the obstacle), and semantic labels with a confidence level greater than or equal to the preset value (such as the confidence level for judging the obstacle category. If it is greater than the preset value, it is considered that the label is reliable and can be used for subsequent avoidance decisions), etc. The robots extract this feature-level data from the original sensor data through data processing algorithms.

[0043] S7b: Hierarchical avoidance execution: Platform - side hierarchical avoidance: Based on the global dynamic obstacle map, a hierarchical avoidance strategy is executed for robots of different ability types: Smart robots execute dynamic detours. According to real - time obstacle information and their own kinematic models, they plan detour paths in real - time to avoid obstacles. Standard robots drive to the designated avoidance point. The platform allocates a suitable avoidance point for the standard robot according to the obstacle position and the robot's ability, and guides it to drive to this point and wait. When detecting dense dynamic obstacles, area speed limit and audible and visual alarms are activated (for robots with the ability feature subset {E}). By reducing the robot's speed, safety is improved, and at the same time, audible and visual alarms remind surrounding pedestrians and vehicles to pay attention to avoidance.

[0044] Furthermore, when the density of dynamic obstacles is greater than the preset density, the platform divides a low - speed operation area with speed limits. In this area, the running speed of the robot is reduced to improve safety. The division basis is as follows: , Formula Six, where A low_speed : Low - speed operation area, D obs (x, y, z): The density of dynamic obstacles at the spatial point (x, y, z), D threshold : Preset density threshold.

[0045] Local - side avoidance: The robot's local - side, based on original sensor data such as lidar point - cloud data, camera image data, etc., uses the robot's own preset algorithms to quickly execute avoidance actions, such as an autonomous - navigation robot detouring around obstacles, a preset - track robot stopping and waiting when encountering obstacles, etc., to handle sudden close - range obstacles and ensure the real - time, flexible, and safe nature of avoidance.

[0046] It should be noted that in addition to the traffic control measures such as path conflict, queuing management, fault handling, deadlock resolution, and sensor - fusion avoidance listed in this embodiment, in the actual complex situation where multiple brands and types of robots are running simultaneously, there may be other conflicts. This embodiment cannot cover and explain all of them. Without departing from the principle of the present invention, using the traffic control method of dynamic ability grading is considered within the protection scope of the present invention.

[0047] The present invention can effectively improve the deficiencies of the existing technology. Through dynamic ability grading and targeted scheduling strategies, it effectively improves the operation efficiency, safety, and overall system performance of heterogeneous robots, enhances the compatibility, scalability, and intelligent level of the system, provides a reliable and efficient solution for multi - robot cooperation, and has great promotion value.

[0048] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A dynamic capability grading traffic control method for heterogeneous robots, characterized in that, Including: S1: Establish a unified scheduling platform to receive the capability registration information of heterogeneous robots. The capability registration information at least includes the basic parameters, functional characteristics, and communication protocols of the robots; S2: Define the capability types of the robots. The capability types are determined by the combination of non-empty subsets of the basic capability feature set and the extended capability feature set: The basic capability feature set at least includes: information reporting (I), pause instruction response (P), resume instruction response (R); The extended capability feature set at least includes: avoidance point navigation (A), dynamic path replanning (D), sensor data sharing (S), environment interaction (E); The capability division rules include but are not limited to: If the capability feature subset contains {I}, but does not contain {P, R, A, D, S, E}, it is defined as read-only type, If the capability feature subset contains {I, P, R}, but does not contain {A, D, S, E}, it is defined as basic type, If the capability feature subset contains {I, P, R, A}, and at least includes one of {S, E}, it is defined as standard type, If the capability feature subset contains {I, P, R, A, D}, and at least includes one of {S, E}, it is defined as intelligent type, S3: The unified scheduling platform dynamically generates a decision-making plan according to the registered capability types of the robots and path conflicts, specifically including: S3a: Prediction stage: Calculate the spatio-temporal conflict probability map of the robot encounter positions, and execute avoidance according to the capability types, including but not limited to: Read-only type: Drive according to the original path; Above basic type: Insert a pause instruction upstream of the conflict point; Above standard type: Insert a pause instruction upstream of the conflict point, or drive to the specified avoidance point for avoidance; Intelligent type: Insert a pause instruction upstream of the conflict point, or drive to the avoidance point for avoidance, or drive according to the route re-planned by the unified scheduling platform to bypass the conflict area; S3b: Conflict response stage: Real-time detect path conflicts and coordinate according to the conflict levels, including but not limited to: Potential conflict: Reduce the speed of the robot, and high-priority tasks pass first; Emergency conflict: Send an emergency stop instruction to the conflicting robots that are not read-only type, and generate a bypass route for the intelligent robots.

2. The dynamic capability grading traffic control method for heterogeneous robots according to claim 1, characterized in that The S1 further includes that the unified scheduling platform processes heterogeneous systems through an adaptation layer. The adaptation layer includes: a protocol conversion engine that converts the native protocols of each brand of robots into a unified communication protocol, a capability registry that dynamically records the capability types of the robots, and a spatio-temporal unified coordinate system that fuses multi-source positioning data into a unified map.

3. A dynamic capability grading traffic control method for heterogeneous robots according to claim 1, characterized in that, The S3a further includes: Taking the path intersection point as a node, calculate the time windows for each robot to arrive; If the time windows of two robots overlap and the spatial distance is less than or equal to the safety distance, mark it as a conflict. The safety distance is dynamically adjusted according to the positioning accuracy.

4. A dynamic capability grading traffic control method for heterogeneous robots according to claim 1, characterized in that, The S3b further includes: Real-time detect path intersection points. When the distance of the conflicting robots from the intersection point is less than or equal to the dynamic safety distance, it is upgraded to an emergency conflict.

5. A dynamic capability grading traffic control method for heterogeneous robots according to claim 1, characterized in that, Also including queuing management: S4: In the queuing waiting area, the robot makes a reservation application to the unified scheduling platform for the right to use the resources ahead. The unified scheduling platform ranks the reservation applications according to the position order of the reserved robots. When a robot with a higher priority is waiting for the right to use, the unified scheduling platform issues queue position instructions to the subsequent non-read-only robots, and the robots queue up at a certain interval distance and speed in turn.

6. The dynamic capability grading traffic control method for heterogeneous robots according to claim 1, characterized in that, It also includes S5 fault handling: The unified scheduling platform detects and identifies the faulty robot and demarcates a fault isolation area, which specifically includes: Generating a dynamic detour path for the intelligent robot and inserting an upstream path deceleration instruction for the non-read-only robot; When the faulty robot recovers, after repositioning, it verifies the integrity of the task and generates a continuous path.

7. A dynamic capability grading traffic control method for heterogeneous robots according to claim 1, characterized in that It also includes S6 deadlock management: S6a: Deadlock detection: The unified scheduling platform uses a deadlock detection algorithm to detect whether there is a deadlock situation in the robot system, including but not limited to: constructing a resource-robot relationship graph, establishing a dynamic relationship, a periodic evaluation mechanism, event-driven detection, path tracking analysis, and circular dependency judgment; S6b: Deadlock resolution: When a deadlock is detected, according to the sequence of robots and resources involved, select the nearest intelligent robot as the leading vehicle to generate a formation passage sequence; Designate an avoidance point or a straight-line avoidance area greater than or equal to the body safety length for the standard robot.

8. A dynamic capability grading traffic control method for heterogeneous robots according to claim 1, characterized in that It also includes S7 sensor fusion avoidance: Robots with the ability feature subset containing {S} share sensor feature-level data to the unified scheduling platform; Platform side: The unified scheduling fuses the data to construct a global dynamic obstacle map and performs hierarchical avoidance: Intelligent type: Dynamic detour; Standard type: Designate an avoidance point; When dense dynamic obstacles are detected, for robots with the ability feature subset containing {E}, activate area speed limit and audible and visual alarms; Local side: The robot local side performs avoidance actions according to the original sensor data.

9. A dynamic capability grading traffic control method for heterogeneous robots according to claim 2, characterized in that The protocol conversion engine supports bandwidth adaptive compression and switches to low-precision positioning data transmission during communication delays. The spatio-temporal unified coordinate system performs timestamp synchronization at regular time intervals.

10. A dynamic capability grading traffic control method for heterogeneous robots according to claim 8, characterized in that, The feature-level data includes: obstacle bounding box, running vector, semantic label with a confidence level greater than or equal to a preset value; when the density of dynamic obstacles is greater than the preset density, the platform demarcates a low-speed operation area with a speed limit.

Images