AGV Path Collision Prevention Method Based on Combining Static and Dynamic Routes

Abstract

This invention discloses an AGV path collision prevention method based on a combination of static and dynamic routes, relating to the fields of intelligent logistics and multi-robot collaborative scheduling. The method includes: dividing the road network into critical conflict zones and non-critical passage zones, and dynamically adjusting regional attributes using a traffic fluid dynamics model; planning global paths and sparsely reserving resources only for the critical conflict zones; employing a distributed strategy based on virtual pheromones for autonomous obstacle avoidance in non-critical passage zones; and using a game-theoretic bidding strategy to acquire real-time passage rights and pass through according to time slots when approaching critical conflict zones. This invention addresses the problems of excessive central scheduling computational load, high risk of local deadlock, and insufficient system adaptability to dynamic disturbances in large-scale AGV cluster environments.

Description

Technical Field

[0001] This invention relates to the field of intelligent logistics and multi-robot collaborative scheduling technology, and more specifically, to an AGV path anti-collision method based on a combination of static and dynamic routes. Background Technology

[0002] The application of Automated Guided Vehicle (AGV) clusters in industrial scenarios continues to expand, leading to increasingly complex road network environments. To meet the demands of large-scale, high-frequency material transportation, path planning and conflict avoidance technologies for multi-machine collaborative operations have become core supports for ensuring the overall throughput and operational efficiency of logistics systems. In high-density cluster environments, how to handle path contention among multiple machines in real time and robustly, and ensure that the system does not deadlock under dynamic disturbances, is a critical issue that the industry urgently needs to address.

[0003] Currently, mainstream AGV path collision avoidance solutions are mainly divided into two categories: centralized global scheduling and distributed local perception. Centralized solutions typically employ a time-sensitive window reservation algorithm based on spatiotemporal state space, with a central server planning non-conflicting trajectories for each AGV. Distributed solutions, on the other hand, focus on the autonomous decision-making of the onboard terminal, utilizing obstacle avoidance strategies such as artificial potential field methods or dynamic window methods, enabling the AGV to perform real-time obstacle avoidance and heading adjustments based on the surrounding environment perceived by sensors.

[0004] However, existing technologies still have significant limitations when dealing with large-scale, high-density scenarios. First, centralized scheduling faces severe scalability bottlenecks; the computational load of global spatiotemporal reservation increases exponentially with the number of AGVs, making it difficult to respond in real time to dynamic disturbances at the work site. Second, purely distributed strategies lack global coordination at intersections or narrow passages, easily leading to local deadlocks or avoidance oscillations, resulting in insufficient operational robustness. Furthermore, existing road network management models are mostly statically partitioned, unable to adaptively adjust control strategies based on real-time traffic flow patterns, and lack a refined allocation mechanism that balances individual fairness with overall system gain when dealing with competition for critical resources. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an AGV path anti-collision method based on a combination of static and dynamic routes. This method addresses the problems of excessive central scheduling computational load, high risk of local deadlock, and difficulty in balancing system scalability and operational robustness in large-scale AGV clusters under complex and highly dynamic environments by deeply coupling fluid dynamic state monitoring under a road network partitioning architecture, sparse reservation and game-theoretic bidding allocation in key conflict zones, and distributed avoidance of virtual pheromones in non-key passage zones.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The AGV path collision prevention method based on a combination of static and dynamic routes includes the following steps: dividing the road network into critical conflict zones and non-critical passage zones; monitoring the environmental status of the non-critical passage zones using a traffic fluid dynamics model and dynamically adjusting the zone attributes based on the monitoring results; planning a global static path for the target AGV's task request, identifying critical conflict zones in the global static path, and reserving the passage authorization resources corresponding to the critical conflict zones; when the target AGV is located in the non-critical passage zone, using a distributed strategy based on virtual pheromones to avoid it; when the target AGV approaches the critical conflict zone, acquiring the real-time right to use the passage authorization resources based on a game-theoretic bidding strategy and passing through according to the reserved time slot.

[0008] In a preferred embodiment, monitoring the environmental status of the non-critical passageway using a traffic fluid dynamics model includes: obtaining the AGV flow density within the non-critical passageway. and average speed According to the formula Calculate real-time traffic Reynolds number ,in For the length of the region feature, The equivalent viscosity coefficient of traffic flow.

[0009] In a preferred embodiment, the step of dynamically adjusting the regional attributes based on the monitoring results includes: if the real-time traffic Reynolds number is greater than a first threshold, then determining that the current region has entered a turbulent state, dynamically switching the non-critical traffic zone to a temporary critical conflict zone, and activating centralized traffic control; if the real-time traffic Reynolds number is less than a second threshold and the duration exceeds a preset smoothing period, then restoring the temporary critical conflict zone to a non-critical traffic zone; wherein, the second threshold is less than the first threshold.

[0010] In a preferred embodiment, the step of obtaining the real-time access rights to the access authorization resources based on a game-theoretic bidding strategy includes: obtaining virtual credit indicators of multiple AGVs competing for the same critical conflict zone, wherein the virtual credit indicators are weighted based on task priority, remaining power, and path length; and matching the virtual credit indicators using a VCG bidding mechanism to allocate the access authorization resources to the AGV with the highest system gain value.

[0011] In a preferred embodiment, the step of obtaining the real-time access rights to the access authorization resources based on a game-theoretic bidding strategy further includes a credit indicator deduction step: calculating the total delay loss caused by the AGV granted access to other AGVs participating in the bidding; and deducting a virtual credit indicator equivalent to the total delay loss from the account of the AGV granted access.

[0012] In a preferred embodiment, the distributed strategy based on virtual pheromones includes: each AGV releases repulsive pheromones in a local virtual map during its journey, and the release intensity of the repulsive pheromones is positively correlated with the current speed of the AGV; the AGV behind senses the pheromone gradient in the surrounding environment through onboard sensors, and superimposes an avoidance vector opposite to the direction of the pheromone gradient on the basis of the global static path.

[0013] In a preferred embodiment, the distributed strategy based on virtual pheromones further includes pheromone attenuation control: according to the formula Calculate the real-time concentration C(t) of the pheromone, where This is the initial release concentration. The evaporation coefficient is used to increase the pheromone removal rate when the AGV movement frequency in the area is lower than the preset frequency.

[0014] In a preferred embodiment, the reservation of access authorization resources corresponding to the critical conflict zone includes: adopting a one-time application strategy, that is, the AGV starts the task after obtaining access authorization resources for all critical conflict zones involved in its global static path.

[0015] In a preferred embodiment, the reservation of access authorization resources corresponding to the critical conflict zone includes: adopting a sequential application strategy, that is, each AGV applies for access authorization resources for the critical conflict zone in the order of the road network node number.

[0016] In a preferred embodiment, after obtaining the real-time access rights to the passage authorization resources based on a game-theoretic bidding strategy and passing through according to the reserved time slot, the method further includes a conflict recovery step: obtaining the avoidance time of the AGV in the non-critical passage area; calculating the estimated time offset of the target AGV to arrive at the next critical conflict area based on the avoidance time, and correcting the estimated arrival time accordingly; if the corrected estimated arrival time deviates from the reserved time slot, then the reservation queue for the critical conflict area is partially extended.

[0017] The technical effects and advantages of this invention's AGV path anti-collision method based on a combination of static and dynamic routes are as follows:

[0018] This invention constructs a scheduling architecture that combines road network zoned control with dynamic and static scheduling, and utilizes a traffic fluid dynamics model to achieve adaptive matching between the control mode and real-time flow conditions. By employing a strategy of sparse resource reservation only for key conflict areas, it significantly reduces the global computational load and spatiotemporal search pressure of the central scheduler, effectively solving the scalability challenge of large-scale cluster systems.

[0019] At the execution level, a distributed virtual pheromone avoidance mechanism endows AGVs with low-latency autonomous coordination capabilities in non-critical areas. Combined with a game-theoretic bidding allocation strategy in critical conflict areas, this achieves a deep decoupling of local flexibility and global orderliness. This not only fundamentally avoids the risk of multi-machine deadlock but also maximizes the overall traffic efficiency of the road network while ensuring task fairness, guaranteeing the system's high robustness even under highly dynamic disturbances. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the AGV path anti-collision method based on a combination of static and dynamic routes provided in an embodiment of the present invention.

[0021] Figure 2 This is a schematic diagram of the fluid dynamic phase transition and control intervention response curve provided in an embodiment of the present invention.

[0022] Figure 3 This is a schematic diagram illustrating the impact of the reservation mechanism on traffic efficiency under different densities provided in an embodiment of the present invention.

[0023] Figure 4 This is a schematic diagram of the spatial gradient field distribution of repulsive pheromones based on an exponential model, provided for an embodiment of the present invention.

[0024] Figure 5 This is a simulation diagram illustrating the convergence process of the local time slot delay adjustment algorithm provided in an embodiment of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0026] Example 1, Figure 1 This invention presents an AGV path anti-collision method based on a combination of static and dynamic routes, comprising the following steps:

[0027] S1 divides the road network into critical conflict zones and non-critical traffic zones.

[0028] In this embodiment, the road network environment is first digitally modeled and logically partitioned. The partitioning process follows a hierarchical judgment logic of "primarily based on geometric feature pre-planning, supplemented by statistical indicators," and specifically includes the following steps:

[0029] (1) Static preplanning based on physical topology features:

[0030] First, preliminary identification is performed based on the physical geometric attributes of each node in the road network. Crossroads, T-junctions, narrow one-way passages with a physical width that only allows a single AGV to pass through, and logistics stations performing high-frequency loading and unloading operations are directly identified and marked as Key Conflict Areas (KCAs). Correspondingly, straight corridors with bidirectional parallel passage capability and a width greater than a preset safety threshold (e.g., more than twice the standard width of an AGV) are pre-designated as Non-critical Access Areas (NCAs).

[0031] (2) Supplementary determination based on conflict sensitivity index:

[0032] To quantify the potential risks in areas of inconspicuous conflict (such as long corridors or temporary work junctions), a regional conflict sensitivity index is introduced. The preliminary planning results are revised. The regional conflict sensitivity index is mentioned. The calculation formula is as follows:

[0033] (1)

[0034] in, Indicators representing regional conflict sensitivity; This represents the topological degree of a road network node, i.e., the number of channel branches connecting that node; This represents the total number of AGV avoidance or lock-up conflicts that occurred in this area during the observation period; This indicates the set historical observation duration; and These represent the topological weight coefficient and the frequency influence factor, respectively.

[0035] In this embodiment, to ensure that the road network topology plays a dominant role in the initial partitioning, a weighting coefficient is set. The value range is 0.6-0.8. The value range is 0.2-0.4. When the calculated regional conflict sensitivity index... When the preset safety threshold is exceeded, even if the area has straight corridor characteristics, it is dynamically adjusted or supplemented from a non-critical passage area and defined as a critical conflict area to achieve a higher level of traffic control. To visually demonstrate the quantitative standards for area division, Table 1 provides the area attribute calibration parameters under different road network topologies in this embodiment:

[0036] Table 1

[0037]

[0038] (3) Spatial data storage and query optimization:

[0039] The topological information, physical coordinates, and regional category attributes of the aforementioned areas are managed by the host computer of the ground dispatch center and stored in real time in the electronic map spatial database of the dispatch management platform. In this embodiment, the spatial database is implemented using a PostGIS spatial plugin or Redis's GeoHash spatial index, leveraging its efficient R-Tree or radius query performance to meet the performance requirements of AGVs performing millisecond-level real-time location queries of their surrounding area attributes during high-speed travel.

[0040] This step provides a topological premise with engineering implementation reliability for the path planning of AGV groups by geometrically layering and labeling static areas and reinforcing dynamic indicators.

[0041] S2, using a traffic fluid dynamics model to monitor the environmental status of the non-critical passage area, and dynamically adjusting the area attributes based on the monitoring results.

[0042] In this embodiment, the real-time monitoring of the environmental status of the non-critical passageway using a traffic fluid dynamics model specifically includes the following steps:

[0043] (1) Acquisition and fusion of multidimensional operational data:

[0044] The scheduling system obtains the AGV traffic density in non-critical passage areas in real time. Average speed and velocity vector gradient Wherein, the flow density The average speed is calculated in real time by statistically analyzing the number of online AGVs per unit effective passage area within non-critical passage zones; It is obtained by fusing the wheel speed encoder data fed back by each AGV with the rate of change of the laser SLAM positioning coordinates.

[0045] Furthermore, the velocity vector gradient This is used to characterize the spatial differences in AGV flow velocity at different coordinate points within a region. In this embodiment, a method based on... Local early warning logic: when The modulus exceeds the preset coefficient of variation (e.g. When the real-time traffic Reynolds number reaches a certain level, it indicates that significant speed unevenness has occurred within the region (such as the initial signs of localized congestion). At this point, even with real-time traffic Reynolds numbers... If the first threshold for triggering switching is not reached, the scheduling system can also trigger a local early warning signal in advance to reduce the injection frequency of AGVs entering the area.

[0046] (2) Quantitative calculation of traffic Reynolds number:

[0047] The scheduling logic calculates real-time traffic Reynolds numbers using a built-in fluid dynamics analysis engine. The calculation formula is as follows:

[0048] (2)

[0049] To ensure the reproducibility of the solution, the key parameters in the formula are defined as follows:

[0050] Physical characteristic length Defined as the effective physical width of a non-critical passageway (analogous to the hydraulic diameter in fluid dynamics), rather than the longitudinal length of the passageway;

[0051] equivalent viscosity coefficient This characterizes the logical repulsion resistance generated by the AGV group due to maintaining a safe collision avoidance distance. In this embodiment, Minimum safe stopping distance from AGV It is directly proportional, and the calculation formula is as follows:

[0052] (3)

[0053] in, The first proportionality coefficient (preferred range is) ), This refers to the regional base load factor (corresponding to the base damping under no-load conditions). This is determined by the safe stopping distance. Incorporating this into the calculations allows the model to dynamically adapt to AGV models with different braking performance.

[0054] (3) Dynamic attribute switching and stability control:

[0055] Based on real-time calculated flow characteristics, the scheduling logic dynamically intervenes in the control level of the region:

[0056] Turbulence determination and centralized control: If the real-time traffic Reynolds number is... If the flow exceeds the first threshold, the current area is determined to be in a turbulent state. The non-critical passage area is dynamically switched to a temporary critical conflict area, and centralized traffic control is activated.

[0057] Steady flow recovery and hysteresis suppression: If the real-time traffic Reynolds number If the value is less than the second threshold and the duration exceeds the preset smoothing period, the temporary critical conflict area will be restored to a non-critical passage area; wherein the second threshold is less than the first threshold.

[0058] (4) Parameter configuration examples in typical scenarios:

[0059] For the convenience of those skilled in the art, in a typical application scenario of this embodiment (such as an e-commerce sorting center), the relevant parameters are set as follows:

[0060] First threshold: set to (Determined to have entered a turbulent flow state);

[0061] Second threshold: set to (Determined to have restored stable laminar flow);

[0062] Preset smoothing period: Set to The cycle is set using a sliding window filtering algorithm. Only when the flow characteristic value remains in a stable range within the set time will attribute rollback be executed, thereby effectively suppressing system control oscillations caused by instantaneous flow fluctuations by utilizing the hysteresis effect. Figure 2 The traffic flow Reynolds number was displayed. The dynamic response curve of regional traffic efficiency. In the simulation experiment, when... When the first threshold (e.g., 2300) is exceeded, the system observes turbulence in the local flow field, that is, the average speed of the AGV group decreases significantly and the frequency of avoidance surges; at this time, the attribute switch is triggered, and the regional throughput quickly recovers to a stable level under centralized control.

[0063] This step solves the technical problem of untimely and unstable control mode switching when static area division faces sudden high-density traffic flow by introducing a fluid dynamic phase transition mechanism with quantitative basis.

[0064] S3. Plan a global static path for the task request of the target AGV, identify key conflict areas in the global static path, and reserve the access authorization resources corresponding to the key conflict areas.

[0065] In this embodiment, after receiving a task request, the central dispatch and management platform first calculates the optimal topology path on a digital electronic map using the A* algorithm or Dijkstra's algorithm. To achieve refined management and control of road network resources, the central server implements access control by managing a resource pool containing the passage status of all critical conflict zones (KCAs). The passage permission corresponding to each critical conflict zone is logically mapped to a mutex lock or semaphore with atomic operation characteristics.

[0066] (1) Adaptive switching logic of reservation strategy:

[0067] To ensure that the system can maintain efficient operation under different workloads, this embodiment introduces a load intensity determination factor. To achieve dynamic switching of the reservation strategy, the calculation formula is as follows:

[0068] (4)

[0069] in, This indicates the total number of AGVs currently active online in the road network; This represents the total number of critical conflict zones defined in the road network. The scheduling system presets a handover threshold. (For example, the range of values ​​is) ):

[0070] Low-load scenarios This strategy employs a one-time request strategy, meaning the target AGV must successfully lock all mutually exclusive resources along the entire path before starting. This strategy ensures the continuity of task execution by breaking the "hold and wait" condition caused by deadlock.

[0071] High load scenarios The system employs a sequential request strategy. This strategy means that at the starting point of the path, the AGV sends all reservation requests related to KCA nodes on that path in ascending order of their unique node numbers. The scheduling center processes the request queue based on the number order rather than the physical arrival order, logically eliminating resource contention loops. Figure 3 This chart compares the performance of average task waiting latency under different reservation strategies as network operation density changes. The chart shows the performance difference between the one-time request strategy and the sequential request strategy under different operation densities. It can be seen that as network operation density increases, the sequential request strategy significantly suppresses the growth rate of waiting latency compared to the one-time request strategy, effectively improving the system's average task turnaround efficiency in high-concurrency scenarios. Through this differentiated reservation mechanism, the system reduces the average task turnaround time in high-concurrency scenarios by approximately 35% while ensuring deadlock-free operation.

[0072] (2) Quantitative assessment and decision-making regarding expected delays in route bookings:

[0073] To scientifically predict the impact of reservation behavior on the overall timeliness of the road network, this embodiment defines the expected route reservation delay. The cost of the path (i.e., the cost of traversing the path) is used as an evaluation metric for path optimization. Its calculation formula is as follows:

[0074] (5)

[0075] in, Indicates the total estimated delay (in seconds) when the target AGV executes the current path reservation task. Indicates the first Conflict probability coefficients for key conflict zones; Indicates the first The current queue length of pending tasks in key conflict zones; Indicates the first Average physical transit time through key conflict zones; Indicates the first The resource turnover frequency of key conflict zones per unit of time.

[0076] In this embodiment, the resource turnover frequency Statistical analysis of the past using the sliding window method Within minutes (e.g.) The frequency parameter is obtained by counting the number of times authorized resources are released in the critical conflict zone, thus ensuring that the frequency parameter can truly reflect the real-time throughput capability of the node.

[0077] (3) Path decision-making logic based on evaluation indicators:

[0078] After obtaining multiple alternative paths, the scheduling system calculates the expected delay for each path. The values ​​are then sorted, and the best value is selected. The path with the smallest value is selected as the final execution path. Furthermore, if all calculated candidate paths... If all values ​​exceed the preset congestion safety limit, the current road network is determined to be oversaturated, and the scheduling system will temporarily suspend the start of the target AGV's task until... The value will fall back to a safe range to avoid ineffective resource consumption and potential systemic congestion.

[0079] This step significantly improves the parallel traffic efficiency of the road network under complex task flows while ensuring absolute safety of multi-machine operations through load-driven strategy switching, transparent frequency statistics, and closed-loop decision-making based on latency expectations.

[0080] S4, when the target AGV is located in the non-critical passage area, a distributed strategy based on virtual pheromones is adopted for avoidance.

[0081] During the specific obstacle avoidance process, each AGV releases repulsive pheromones in a local virtual map while traveling, and the intensity of the repulsive pheromone release is positively correlated with the current speed of the AGV. In this embodiment, the virtual grid map is stored on an edge computing node or a distributed vehicle database, and millisecond-level pose and concentration data synchronization is achieved through a 5G network or high-frequency Wi-Fi.

[0082] (1) Mapping and release of pheromone concentration:

[0083] The vehicle controller synchronizes its real-time pose and speed information to a shared virtual grid map. Specifically, the initial release concentration of pheromones... The calculation satisfies the following mapping relationship:

[0084] (6)

[0085] in, This indicates the instantaneous speed of the AGV at the moment of release; As a second proportionality coefficient, in a preferred embodiment, The range of values ​​is Through this mapping relationship, a faster AGV will generate a higher repulsive potential field strength around its trajectory, thus creating a more spacious logical safety buffer for high-speed vehicles.

[0086] (2) Collision avoidance control based on data fusion and vector synthesis:

[0087] In a multi-machine interactive environment, the rear AGV performs deep data fusion, superimposing the physical obstacle information collected by the vehicle-mounted LiDAR or ultrasonic sensors with the pheromone gradient information parsed from the virtual map in multiple dimensions.

[0088] Specifically, the onboard computing unit calculates the avoidance vector by performing spatial analysis on a local virtual map. The avoidance vector is calculated according to the following formula:

[0089] (7)

[0090] in, This represents the pheromone concentration gradient at the current coordinate point; This is the preset avoidance gain constant; To prevent extremely small offset constants with a denominator of zero (such as...) ).

[0091] Should As a compensation parameter pointing in the direction of decreasing pheromone concentration, it is input in real time to the underlying PID motion control algorithm or model predictive control (MPC) framework, enabling the AGV to achieve smooth arc-shaped navigation while maintaining the global forward trend, thus avoiding efficiency losses caused by frequent stops.

[0092] (3) Pheromones decay and environmental purification control:

[0093] To ensure the timeliness of avoidance information, the scheduling logic implements strict pheromone decay control. The formula for calculating the real-time pheromone concentration is as follows:

[0094] (8)

[0095] in, Indicates elapsed time Real-time pheromone concentration afterward; Indicates the initial release concentration of pheromones; Indicates the evaporation coefficient of pheromones; This indicates the time elapsed since the release.

[0096] At the dynamic adjustment level, when the AGV movement frequency within the area is lower than the preset frequency, the system increases the evaporation coefficient. (For example, to Increase the speed by 1.5-2 times (to the original value) to accelerate the removal of outdated path information. This ensures that new vehicles entering the area can make path decisions based on the cleanest map environment, avoiding invalid heading deflections caused by "ghost obstacles". Figure 4 A spatial profile of the virtual pheromone concentration around the AGV was depicted. The figure shows that the pheromone intensity decreases exponentially with increasing distance, and the gradient is steepest in front of the AGV's movement. Experiments demonstrate that by superimposing the avoidance vector onto the global path, the radius of curvature of the AGV's local bypass path increases significantly, and the average avoidance speed loss decreases from 100% in the stop-and-wait mode to below 15%.

[0097] This step utilizes a biomimetic model combined with a precise vector compensation algorithm to achieve low-latency local autonomous coordination, significantly reducing the computational pressure on the central server while improving traffic flow in non-critical passage areas.

[0098] S5, when the target AGV approaches the critical conflict zone, it obtains the real-time right to use the passage authorization resource based on a game-theoretic bidding strategy and passes through according to the reserved time slot.

[0099] When the target AGV approaches the decision physical boundary of the critical conflict zone (KCA) (e.g., within 5m of the intersection center point), the scheduling management platform executes a refined resource competition decision. During this process, the real-time right to use the passage authorization resource is acquired based on a game-theoretic bidding strategy. Specific implementation details are as follows:

[0100] (1) Normalized calculation of Virtual Credit Index (VCI):

[0101] The scheduling platform obtains virtual credit indicators for multiple AGVs competing for the same critical conflict zone. These virtual credit indicators are weighted based on task priority, remaining battery power, and path length. To eliminate the influence of different metrics on weight allocation, the virtual credit indicators... The calculation is performed using the normalized formula as follows:

[0102] (9)

[0103] in, A virtual credit metric representing the AGVs participating in the bidding; This represents the priority coefficient of the current task. The highest priority nominal value defined by the system; This represents the remaining battery percentage. This is the energy adjustment factor (used to prevent numerical divergence caused by an excessively small denominator). This represents the remaining path length. This is the nominal value of the longest task path in the road network; , and The weights are preset. This normalization process ensures that all parameters are on the same order of magnitude, improving the fairness of the bidding logic.

[0104] (2) Resource allocation and deduction based on VCG mechanism:

[0105] The scheduling logic uses the VCG bidding mechanism for matching, allocating passage authorization to the system gain value (i.e., The AGV with the highest sum of virtual credit indicators. Correspondingly, a credit indicator deduction step is also included: calculating the total delay loss caused by the approved AGV to other participating AGVs; and deducting an equivalent amount of virtual credit indicators from the AGV's account. .

[0106] (10)

[0107] In Equation 10, m is the total number of AGVs participating in the bidding, w is the index of the winning AGV, and j is the traversal index; to meet the high real-time requirements of the embedded system, the additional waiting time... Quickly obtain the average turnover frequency based on this key conflict zone using a table lookup method. The queuing position of each competing AGV can be directly retrieved, which greatly reduces the real-time floating-point calculation pressure on the scheduling server.

[0108] (3) Credit score replenishment and system dynamic conservation:

[0109] To ensure the dynamic conservation of credit scores within the system and prevent AGV accounts from "going bankrupt" due to frequent bidding, this embodiment adds a credit score replenishment logic. When an AGV completes a transportation task and successfully unloads goods, the system replenishes the credit limit based on its execution efficiency. :

[0110] (11)

[0111] in, The base recovery value; Efficiency reward coefficient; The standard estimated time for this task; This represents the actual total time spent. This mechanism maintains the economic balance of the "virtual currency" within the system by incentivizing efficient task completion. Figure 5This is a convergence simulation diagram of the estimated arrival time deviation after the conflict recovery step is triggered. The figure shows the change of the overall time axis offset with the number of dynamic scheduling compensation cycles; it can be seen that under the algorithm intervention, the initial deviation decreases rapidly cycle by cycle, and enters the set reservation tolerance threshold band around the 5th compensation cycle. The result eventually approaches zero, verifying the effectiveness and convergence speed of the local time slot delay adjustment algorithm in eliminating time axis offset.

[0112] (4) Conflict recovery and time slot extension adjustment:

[0113] This step also includes a conflict recovery step to address time offsets during dynamic operation. In specific implementation, the avoidance time of the AGV in the non-critical passage area is first obtained. This time is obtained by accumulating the difference between the actual travel time of the AGV when performing virtual pheromone avoidance actions and the original uniform speed planning time.

[0114] Subsequently, the scheduling platform calculates the estimated time offset of the target AGV's arrival at the next critical conflict zone based on the avoidance time, and corrects the estimated arrival time accordingly. The formula for correcting the estimated arrival time is as follows:

[0115] (12)

[0116] in, This indicates the revised estimated time of arrival at the next key conflict zone; This indicates the original scheduled entry time of the AGV during the global static path planning phase; This indicates that the AGV is the [number]th ... The time consumed by each distributed avoidance action; This indicates the total number of evasive maneuvers that occurred before reaching the next critical conflict zone.

[0117] After obtaining the revised estimated time, the scheduling platform performs a time slot deviation determination. If the revised estimated arrival time deviates from the scheduled time slot, i.e., when the time offset... If the preset tolerance threshold is exceeded, the system determines that the current reservation is invalid. The tolerance threshold is set based on the physical properties of the critical conflict zone, typically taken as the average time it takes for the AGV to traverse that critical conflict zone. (For example, set as) Once a deviation is detected, the scheduling platform will adjust the local time slot of the reservation queue in the critical conflict area. Specifically, the system will automatically trigger dynamic time slot shifting to... Using this as a baseline, the access authorization window for the AGV and its subsequent affected vehicles in the KCA resource pool is simultaneously corrected.

[0118] This step maximizes the overall system throughput while ensuring the fairness of large-scale AGV cluster operations through a game-theoretic bidding model with parameter normalization and a time slot correction mechanism with tolerance protection.

[0119] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0120] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0121] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0122] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0123] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0124] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An AGV path anti-collision method based on a combination of static and dynamic routes, characterized in that, Includes the following steps: The road network is divided into critical conflict zones and non-critical traffic zones; The environmental conditions of the non-critical passage areas are monitored using a traffic fluid dynamics model, and the area attributes are dynamically adjusted based on the monitoring results. For the task request of the target AGV, a global static path is planned, key conflict areas in the global static path are identified, and access authorization resources corresponding to the key conflict areas are reserved. When the target AGV is located in the non-critical passage area, a distributed strategy based on virtual pheromones is adopted for avoidance. When the target AGV approaches the critical conflict zone, it acquires the real-time right to use the passage authorization resource based on a game-theoretic bidding strategy and passes through according to the reserved time slot.

2. The method according to claim 1, characterized in that, The monitoring of the environmental status of the non-critical passageway using a traffic fluid dynamics model includes: Obtain the AGV flow density in the non-critical passage area. and average speed ; According to the formula Calculate real-time traffic Reynolds number ,in For the length of the region feature, The equivalent viscosity coefficient of traffic flow.

3. The method according to claim 2, characterized in that, The dynamic adjustment of regional attributes based on monitoring results includes: If the real-time traffic Reynolds number is greater than the first threshold, it is determined that the current area has entered a turbulent state, the non-critical passage zone is dynamically switched to a temporary critical conflict zone, and centralized traffic control is activated. If the real-time traffic Reynolds number is less than the second threshold and the duration exceeds the preset smoothing period, the temporary critical conflict zone will be restored to a non-critical passage zone; wherein the second threshold is less than the first threshold.

4. The method according to claim 1, characterized in that, The method of obtaining the real-time right to use the access authorization resource based on a game-theoretic bidding strategy includes: The virtual credit index of multiple AGVs competing for the same critical conflict zone is obtained, and the virtual credit index is obtained by weighting based on task priority, remaining power and path length. The virtual credit index is matched using the VCG bidding mechanism, and the access authorization resources are allocated to the AGV with the highest system gain value.

5. The method according to claim 4, characterized in that, It also includes the credit score deduction process: Calculate the total delay loss caused by the approved AGV to other AGVs participating in the bidding; A virtual credit metric equivalent to the total delay loss is deducted from the account of the AGV that has been granted access.

6. The AGV path anti-collision method based on a combination of static and dynamic routes according to claim 1, characterized in that, The distributed strategy based on virtual pheromones includes: Each AGV releases repulsive pheromones in a local virtual map during its operation, and the intensity of the release of the repulsive pheromones is positively correlated with the current speed of the AGV. The AGV behind senses the pheromone gradient in the surrounding environment through onboard sensors and superimposes an avoidance vector opposite to the direction of the pheromone gradient on the global static path.

7. The method according to claim 6, characterized in that, It also includes pheromone decay control: According to the formula Calculate the real-time concentration C(t) of the pheromone, where This is the initial release concentration. The coefficient of evaporation; When the AGV movement frequency in the area is lower than the preset frequency, the pheromone removal rate is increased by increasing the evaporation coefficient.

8. The method according to claim 1, characterized in that, The reservation of access authorization resources corresponding to the key conflict zone includes: A one-time application strategy is adopted, that is, the AGV starts the task after obtaining the passage authorization resources of all key conflict areas involved in its global static path.

9. The method according to claim 1, characterized in that, The reservation of access authorization resources corresponding to the key conflict zone includes: A sequential application strategy is adopted, that is, each AGV applies for passage authorization resources in key conflict areas in the order of road network node number.

10. The method according to claim 1, characterized in that, After obtaining the real-time access rights to the access authorization resources based on a game-theoretic bidding strategy and passing through according to the reserved time slot, the process also includes a conflict resolution step: Obtain the time taken for the AGV to avoid obstacles in the non-critical passage area; Based on the avoidance time, calculate the estimated time offset of the target AGV to arrive at the next critical conflict zone, and correct the estimated arrival time accordingly; If the revised estimated arrival time deviates from the reserved time slot, the reservation queue for the critical conflict area will be adjusted by extending the local time slot.

Images