Real-time traffic-oriented agv cooperative avoidance path planning method

Abstract

The application discloses a kind of AGV cooperative path planning methods for real-time traffic, it is related to AGV path planning technical field, comprising: based on AGV state data constructs global traffic map, and global traffic map is dynamically updated, the dynamic updating includes the weight update based on AGV density value and average speed;Based on the global traffic map after updating, using path planning algorithm generates AGV global optimization cooperative path set;Cooperative path set is converted into control instruction, and control instruction is dynamically compensated by load sensing, and corrected control instruction is obtained and issued to AGV;The application utilizes global traffic dynamic updating to combine cooperative path optimization and the dynamic control of load sensing, realizes the global traffic balance of multiple AGVs and real-time congestion prevention, solves the problem that path conflict is frequent under high load, local congestion cannot be actively adjusted and the overall efficiency is reduced by single AGV optimization.

Description

Technical Field

[0001] This invention relates to the field of AGV path planning technology, and more specifically, to an AGV collaborative congestion avoidance path planning method oriented towards real-time traffic conditions. Background Technology

[0002] In automated logistics and manufacturing systems, the collaborative operation efficiency of multiple AGVs (Automated Guided Vehicles) is crucial. With the deepening of applications and the expansion of cluster scale, path conflicts and traffic congestion have become core issues restricting system performance. Existing AGV scheduling systems, in terms of traffic control, have evolved from early static rules such as fixed paths and area locks to dynamic path planning based on real-time computation. However, these methods generally suffer from the problem of optimizing only a local perspective and passive management strategies, leading to a sharp drop in system efficiency under high loads and failing to achieve true global collaboration and congestion prevention. Existing technologies have the following drawbacks:

[0003] 1. Existing systems largely rely on predefined, static traffic rules, such as area interlocks, time windows, and one-way loops. These rules are fixed during the system design phase and cannot perceive or adapt to dynamically changing road conditions during operation. When a road segment becomes congested due to AGV aggregation, slow movement, or temporary obstacles, the system cannot proactively and promptly adjust the paths of subsequent AGVs, causing AGVs to "blindly" enter congested areas, exacerbating traffic problems and creating a "deadlock" or "livelock" state.

[0004] 2. Most dispatching systems, when planning routes, only use "shortest path" or "shortest time" as optimization objectives, planning independently and sequentially for individual AGVs. This approach lacks a coordinated consideration of the overall traffic flow of the system. The system cannot obtain and comprehensively utilize the traffic efficiency of all road segments (such as AGV density, average speed, queue length, etc.) in real time as a basis for planning. Therefore, it cannot achieve a balanced distribution of traffic flow from a global perspective, and local bottleneck sections become the bottlenecks in the overall efficiency of the system.

[0005] 3. When conflicts (such as deadlocks or head-on encounters) occur, existing systems typically employ a "response-based" resolution mechanism, meaning they only address the issue by simply "stopping and waiting" or "randomly yielding" when the conflict has already occurred or is about to occur. This "blocking first and then fixing later" approach not only causes AGV operation interruptions and wasted time, significantly reducing system throughput, but also increases energy consumption and mechanical wear due to frequent start-ups and shutdowns. It is essentially a passive and inefficient remedial measure, rather than a proactive prevention strategy. To address the above problems, this invention proposes a solution. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an AGV collaborative congestion avoidance path planning method oriented to real-time traffic conditions. By combining global traffic condition dynamic updates with collaborative path optimization and load-aware dynamic control, the method achieves global traffic balance and real-time congestion prevention for multiple AGVs, solving the problems of frequent path conflicts under high load, inability to actively adjust local congestion, and overall efficiency decline caused by single AGV optimization.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] This application provides an AGV collaborative congestion avoidance path planning method for real-time traffic conditions. The method includes: constructing a global traffic map based on AGV status data and dynamically updating the global traffic map. The dynamic update includes weighted updates based on AGV density values ​​and the average speed of each road segment. Constructing the global traffic map based on AGV status data includes time synchronization calibration of each AGV to obtain calibrated AGV status data. The AGV density value is used to characterize the congestion level of each road segment. Based on the updated global traffic map, a path planning algorithm is used to generate a globally optimized collaborative path set for AGVs. The path planning algorithm includes a cost function based on static path length and dynamic path weights, and combines path load and traffic flow for collaborative optimization. The collaborative path set is converted into control commands, and the control commands are dynamically compensated through load sensing to obtain corrected control commands, which are then sent to the AGVs. The load sensing dynamic compensation includes acquiring AGV load data and inertial state data, calculating acceleration correction values ​​and steering angle correction values, and dynamically adjusting the control commands through a PID control algorithm.

[0009] In one embodiment, constructing a global road condition map based on AGV status data includes: performing time synchronization calibration on each AGV to obtain calibrated AGV status data; constructing a global road condition map containing initial weights of path edges based on the calibrated AGV status data and static map information; calculating the AGV density value and average speed of each path edge based on the global road condition map; and dynamically adjusting the initial weights of the corresponding path edges based on the AGV density value and average speed to update the global road condition map.

[0010] In one embodiment, time synchronization calibration is performed on each AGV to obtain calibrated AGV status data. Specifically, this involves: constructing a two-layer clock model including a global clock layer and a local clock layer; broadcasting a global time synchronization signal packet, with AGV nodes recording their local reception time and calculating the time offset; continuously collecting offset samples to form a time series and calculating the time drift trend; constructing a drift compensation function based on the time drift trend to correct the local clock time; calculating and updating the offset based on the correction result, and integrating them to form a global time consistency matrix; calculating the matrix variance and determining whether the time synchronization accuracy is stable. If unstable, the system performs broadcast frequency adjustment and generates synchronization status feedback information, dynamically adjusting the broadcast period based on the synchronization status feedback information.

[0011] In one embodiment, the global road condition map is dynamically updated, including: smoothing the AGV density value and average speed, and calculating the smoothed value of the AGV density value and the smoothed value of the average speed; calculating the updated dynamic path weight based on the difference between the smoothed value of the AGV density value and the smoothed value of the average speed and the standard data, combined with the initial weight of the path and the weight adjustment coefficient; and comparing the rate of change of the AGV density value and the average speed with a preset threshold, and increasing or decreasing the weight adjustment coefficient based on the comparison result.

[0012] In one embodiment, based on the updated global traffic map, a path planning algorithm is used to generate a collaborative path set for global AGV optimization, including: performing single AGV path planning based on the global traffic map, calculating the optimal path for each AGV using a cost function consisting of static path length and dynamic path weight; and performing collaborative optimization based on the optimal paths of each AGV, combined with path load and traffic indicators, to generate the final path set.

[0013] In one embodiment, the method of generating a globally optimized collaborative path set for AGVs using a path planning algorithm further includes: obtaining the actual AGV density value and actual average speed of each path segment based on the current optimal path of each AGV, calculating the flow load coefficient and performing normalization processing, and identifying high-load path segments; dynamically adjusting the AGV path distribution according to the cost changes of high-load path segments; monitoring the rate of change of the global average flow load coefficient, and determining that the system flow has reached a stable equilibrium state when the rate of change is less than a preset threshold, and outputting the final collaborative path set.

[0014] In one embodiment, the method of generating a globally optimized collaborative path set for AGVs using a path planning algorithm further includes: correcting the dynamic path weights through a traffic load coefficient, substituting them into a cost function, and obtaining a system cost matrix; and performing global path collaborative optimization based on the cost matrix to minimize the total system cost function.

[0015] In one embodiment, the cooperative path set is converted into control commands, specifically by: receiving a cooperative path set consisting of a start point, an end point, and a sequence of path nodes; performing curve fitting on each path node sequence to generate a smooth, continuous curve; calculating the state data of each sampling point based on the smooth, continuous curve, and obtaining velocity and acceleration parameters through curve differentiation; and discretizing the path into control command units, each command unit containing path segment coordinates, target velocity, acceleration, and steering angle.

[0016] In one embodiment, the control commands are dynamically compensated through load sensing to obtain corrected control commands, which are then sent to the AGV. Specifically, the load data and inertial state data of the AGV are acquired; based on the load data and inertial state data, acceleration correction values ​​and steering angle correction values ​​are calculated; based on the acceleration correction values ​​and steering angle correction values, the control commands of the AGV are dynamically adjusted through a PID control algorithm to obtain corrected control commands.

[0017] In one embodiment, the control commands of the AGV are dynamically adjusted using a PID control algorithm based on the acceleration correction value and the steering angle correction value to obtain the corrected control commands. Specifically, this involves: establishing an independent speed control system and a steering control system based on a PID framework; calculating the speed error and steering error of the control system; generating speed correction commands and steering correction commands respectively using a PID algorithm based on the speed error and steering error; adjusting the speed and steering control commands of the AGV according to the speed correction commands and steering correction commands; and sending the corrected control commands to the AGV's drive system for execution.

[0018] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0019] By collecting AGV status data in real time and constructing a dynamically updated global road condition map, combined with an adaptive synchronization algorithm based on a two-layer clock model, time deviations in multi-AGV systems are eliminated, achieving high-precision time synchronization. In path planning, an improved A* algorithm is used to combine static path length and dynamic path weight for optimal single-AGV path planning, and high-load paths are dynamically adjusted based on the flow load coefficient to achieve global flow balance and congestion suppression. At the control execution level, smooth and continuous paths are generated through curve fitting, and combined with load-aware acceleration and steering correction, and PID dual-channel control, accurate trajectory tracking and stable control of AGVs under different loads and dynamic operating conditions are achieved. Overall, this scheme significantly improves the path planning accuracy, scheduling efficiency, flow balance, and motion control robustness of multi-AGV systems, effectively reducing congestion risks and ensuring safe and reliable system operation. Attached Figure Description

[0020] Figure 1This is a schematic diagram of a collaborative AGV congestion avoidance path planning method based on real-time traffic conditions, provided in an embodiment of this application. Detailed Implementation

[0021] 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.

[0022] Reference Figure 1 As shown in the diagram, this invention provides a flowchart of an AGV collaborative congestion avoidance path planning method based on real-time traffic conditions, which includes the following steps:

[0023] S1. Real-time acquisition of AGV status data, construction of a global road condition map and dynamic updates, the dynamic updates including weighted updates based on AGV density value and average speed.

[0024] The AGV status information is collected in real time through onboard sensors (LiDAR, IMU, etc.). The AGV status data includes current coordinates, operating speed, acceleration, direction of travel, task status, and road segment identification, and is continuously reported to the central server via a wireless communication network. The real-time data collection process involves periodic data reporting based on the communication link between the wireless communication network and the dispatch center, forming an AGV status dataset indexed by timestamps.

[0025] In this embodiment, AGV status data is acquired in real time, a global road condition map is constructed, and dynamically updated. The dynamic update includes weighted updates based on AGV density values ​​and average speeds, specifically:

[0026] An adaptive synchronization algorithm based on a two-layer clock model is used to dynamically compensate for the time offset of each AGV, achieve time synchronization calibration, and obtain calibrated AGV status data.

[0027] Based on the calibrated AGV status data, combined with the static map information inside the warehouse (including path topology, node location and traffic constraints), a global road condition map is constructed in the form of a node-edge graph.

[0028] In the global traffic map, nodes represent intersections or task points, edges represent passable paths, and each path edge is assigned an initial weight, which is used to reflect the traffic efficiency of the road segment.

[0029] Based on the global road condition map, the real-time AGV coordinates are projected onto the corresponding path units, and the number of AGVs on each path within a unit time window is counted.

[0030] The AGV density value of a path is obtained by calculating the ratio of the number of AGVs to the path length. The AGV density value is used to characterize the congestion level of each road segment and serves as an important parameter for subsequent dynamic map updates.

[0031] Within a unit time window, the instantaneous speed values ​​of all AGVs on the same path are counted, and the average speed is calculated.

[0032] The average speed is used to characterize the traffic speed characteristics of the path, and the specific calculation formula is as follows:

[0033]

[0034] In the formula, For average speed, Let be the instantaneous speed value of the i-th AGV. This refers to the number of AGVs.

[0035] The initial weights of corresponding paths in the global road condition map are adjusted based on the AGV density value and average speed, and the global road condition map is dynamically updated.

[0036] Furthermore, an adaptive synchronization algorithm based on a two-layer clock model dynamically compensates for the time offset of each AGV, achieving time synchronization calibration and obtaining calibrated AGV status data, specifically:

[0037] In a multi-AGV system, a two-layer clock model is constructed, comprising a global clock layer and a local clock layer. The global clock layer is maintained by the scheduling center and provides a unified reference time for the system. The local clock layer is maintained by each AGV node, which records its own local clock time. ;

[0038] The dispatch center periodically broadcasts a global time synchronization signal packet to all AGV nodes. The signal packet includes the current system unified reference time and synchronization sequence number identifier.

[0039] Each AGV node immediately records the local reception time and calculates the time offset upon receiving a signal packet;

[0040] The specific formula for calculating the time offset is as follows:

[0041]

[0042] In the formula, Initial time offset Let be the local clock time of the i-th AGV.

[0043] During system operation, each AGV continuously collects several offset samples of signal packets to form a time series, and uses the sliding time window method to calculate the time drift trend.

[0044] The specific formula for calculating the time drift trend is as follows:

[0045]

[0046] In the formula, The drift trend of the i-th AGV is shown below. Let k be the time offset of the i-th AGV during the k-th sampling. Let be the time offset of the (k-1)th sample of the i-th AGV. is the interval of the kth sampling period.

[0047] Based on the time drift trend, a drift compensation function is constructed. The drift compensation function is used to smooth the drift trend of the AGV local clock and generate a continuous and stable correction amount to adjust the local clock time so that it is closer to the global clock.

[0048] The drift compensation function is calculated using the following formula:

[0049]

[0050] In the formula, For drift compensation amount, The drift compensation amount at the previous moment. The smoothing coefficient determines the degree to which the current drift trend affects the compensation.

[0051] Correct the local clock time according to the drift compensation function;

[0052] The specific formula for calculating the corrected local clock time is as follows:

[0053]

[0054] In the formula, To correct the local clock time.

[0055] Calculate and correct the local clock update offset based on the corrected local clock time;

[0056] The specific formula for calculating the update offset is as follows:

[0057]

[0058] In the formula, To update the offset.

[0059] If the absolute value of the updated offset is less than the preset synchronization tolerance, the current node is considered to have achieved effective synchronization.

[0060] If the absolute value of the updated offset is greater than the preset synchronization tolerance, the smoothing coefficient in the drift compensation function is adjusted according to the offset direction and the drift trend of the previous period. If the drift trend of the current period is the same as the drift trend of the previous period and the updated offset increases, the smoothing coefficient is increased; if the drift trend of the current period is the same as the drift trend of the previous period and the updated offset decreases, the smoothing coefficient is maintained or finely adjusted; if the drift trend of the current period is opposite to the drift trend of the previous period, the smoothing coefficient is decreased.

[0061] Integrate the update offsets of all AGVs to form a global time consistency matrix;

[0062] Calculate the variance of the matrix and make a judgment. If the variance is lower than the set standard variance threshold, it means that the overall time synchronization accuracy of the system has reached a stable state. If the variance exceeds the set standard variance threshold, the system will perform broadcast frequency adjustment.

[0063] The specific calculation formula for the broadcast frequency adjustment is as follows:

[0064]

[0065] In the formula, The adjusted broadcast frequency, Original broadcast frequency, This is for adjusting the coefficient.

[0066] Based on the matrix judgment results, global synchronization status feedback information is generated and transmitted back to the scheduling center. The scheduling center dynamically adjusts the broadcast period according to the feedback information to maintain the optimal synchronization effect under different network delays and communication loads, thereby realizing the system's adaptive clock synchronization control to network fluctuations and its own drift.

[0067] The feedback information refers to the variance of the matrix, the update offset, and the adjusted broadcast frequency.

[0068] It's important to note that in multi-AGV systems, clock drift exists in the sensors, positioning modules, and communication devices of each vehicle. Combined with wireless communication network latency and jitter, this leads to discrepancies in the timestamps of the status data reported by different AGVs. When the number of AGVs reaches dozens or even hundreds, these time discrepancies accumulate, causing time misalignment in the calculated AGV density and average speed within the same time window. This means the actual vehicle status is out of sync with the statistical data, affecting the real-time performance and accuracy of the global road condition map. Traditional time synchronization protocols such as NTP or PTP struggle to completely eliminate jitter and drift in this wireless network environment, exacerbating the problem.

[0069] To address this issue, the aforementioned adaptive synchronization algorithm based on a two-layer clock model constructs a global clock layer and a local clock layer within the system. It combines periodic global time broadcasting, time offset calculation, drift trend modeling, and an exponential smoothing drift compensation function to achieve dynamic correction and real-time synchronization of the local clocks of each AGV. Simultaneously, by dynamically adjusting the smoothing coefficient α, constructing a global time consistency matrix, and adaptively adjusting the broadcast frequency based on variance results, a closed-loop control mechanism is formed. The advantages of this technical solution include: firstly, it significantly improves the time synchronization accuracy of each AGV, ensuring consistent data reporting times from different vehicles; secondly, it eliminates time misalignment caused by multi-source data delays, guaranteeing the real-time performance and accuracy of AGV density values ​​and average speed statistics; and thirdly, it enhances the system's robustness to network fluctuations and node drift, enabling dynamic updates and reliable scheduling of the global road condition map, and providing an accurate time reference for multi-AGV collaborative control.

[0070] Furthermore, the initial weights of the corresponding paths in the global road condition map are adjusted based on the AGV density value and average speed, and the global road condition map is dynamically updated, specifically as follows:

[0071] The AGV density value and average speed are smoothed by using the exponential weighted moving average method to calculate the smoothed values ​​of AGV density value and average speed.

[0072] The specific formula for calculating the smoothed value of the AGV density is as follows:

[0073]

[0074] In the formula, This is a smoothed value for AGV density. This represents the AGV density value. These are the filter coefficients. This is the smoothed value of the AGV density value from the previous moment.

[0075] The specific formula for calculating the average speed smoothing value is as follows:

[0076]

[0077] In the formula, This is the average speed smoothing value. This is the smoothed average velocity value from the previous moment.

[0078] The initial weights of the path are updated based on the differences between the smoothed values ​​of AGV density and average speed and the standard data, resulting in updated dynamic path weights.

[0079] The updated dynamic path weight is calculated using the following formula:

[0080]

[0081] In the formula, For the updated dynamic path weights, As the initial weights, This is the standard AGV density value. For standard average speed, and , respectively, are the weighting adjustment coefficients.

[0082] The weight adjustment coefficient is dynamically adjusted based on the rate of change of AGV density and average speed. If the rate of change of the current AGV density or average speed exceeds the preset threshold, the weight adjustment coefficient is increased to improve the response speed. If the rate of change is less than the preset threshold, the weight adjustment coefficient is decreased to suppress weight oscillation.

[0083] It should be noted that the initial weights are updated using AGV density and average speed, and these two parameters themselves change continuously with vehicle driving status, network latency, and AGV scheduling fluctuations. If the weight update frequency is too high or updates are made directly based on the original data, the path weights are prone to frequent oscillations on the global road condition map, making it difficult for the system to converge stably. Conversely, if the update frequency is too low or filtering is insufficient, the map cannot reflect actual road condition changes in a timely manner, resulting in weight lag and affecting path scheduling and optimization decisions.

[0084] By applying an exponentially weighted moving average (EWMA) smoothing process to the AGV density values ​​and average speeds, the impact of instantaneous fluctuations on weight updates is reduced. Secondly, the initial weights of the path are updated based on the smoothed values, and the update magnitude is automatically increased or decreased according to the rate of change of AGV density values ​​and average speeds through dynamic adjustment of the weight adjustment coefficients. This enables rapid response to sudden congestion and suppression of oscillations in stable road conditions. It improves the stability and reliability of the global road condition map, allowing it to quickly reflect changes in vehicle density and speed when road conditions change abruptly, thus gradually stabilizing the map weights and supporting reliable scheduling and path optimization decisions for multi-AGV systems.

[0085] S2, based on a global traffic map, employs a path planning algorithm. The primary objective is to plan the shortest path for a single AGV, while the advanced objectives are to prevent congestion and balance traffic flow. This generates a globally optimized collaborative path set for AGVs. The path planning algorithm includes a cost function based on static path length and dynamic path weight to achieve collaborative optimization and balancing of global traffic flow in the system. The path planning algorithm also includes an improved A* algorithm.

[0086] In this embodiment, based on a global traffic map, a path planning algorithm is used. The primary objective is to plan the shortest path for a single AGV, while the higher-level objectives are congestion prevention and traffic balancing. This generates a globally optimized set of cooperative paths for the AGVs, specifically:

[0087] Single AGV path planning is performed based on a global road condition map. The path planning calculates the optimal path based on a cost function constructed from static path length and dynamic path weight.

[0088] The calculation of the optimal path specifically involves: using an improved A* algorithm, inputting the AGV's start and end points, the node-edge topology of the global road condition map, the static path length and dynamic path weight of each edge; calculating the comprehensive cost value for each passable edge using a cost function; using the comprehensive cost value from the current start point to the end point as the heuristic function of the improved A* algorithm, searching for the path with the minimum comprehensive cost value from the start point to the end point, until the optimal path with the minimum total cost is found.

[0089] The cost function is calculated using the following formula:

[0090]

[0091] In the formula, The comprehensive cost of path e, Let be the static path length of path e. The dynamic path weight for path e. The weighting coefficient for static path length. This represents the dynamic path weight coefficient.

[0092] Based on the optimal path of each AGV, collaborative optimization is performed by combining path load and traffic.

[0093] Furthermore, based on the optimal path of each AGV, collaborative optimization is performed by combining path load and traffic, specifically as follows:

[0094] Based on the current optimal path for each AGV, obtain the actual AGV density value and actual average speed on each path segment;

[0095] The flow load factor of each path segment is calculated by using the actual AGV density value and the actual average speed. The flow load factor is used to quantify the load ratio of the path segment in the current cycle.

[0096] The specific formula for calculating the flow load factor is as follows:

[0097]

[0098] In the formula, For flow load factor, This represents the actual AGV density value. For the maximum permissible density, This is the actual average speed. , These are the weighting coefficients.

[0099] The flow load factor is normalized and high-load path segments are identified. If the normalized flow load factor exceeds the preset load threshold, the path segment is marked as a congested path, i.e. a high-load path segment.

[0100] The dynamic path weight of the current road segment is corrected by the traffic load factor. The corrected dynamic path weight is then substituted into the cost function to recalculate the comprehensive cost value and generate a comprehensive cost matrix for all AGV paths in the entire system.

[0101] The specific calculation formula for the corrected dynamic path weight is as follows:

[0102]

[0103] In the formula, The corrected dynamic path weights, The normalized flow load factor is... For the system's average normalized load, This is the dynamic adjustment coefficient.

[0104] Global path collaborative optimization is performed based on the cost matrix. An improved A* algorithm is adopted to minimize the total system cost function. During the optimization process, the AGV path distribution is dynamically adjusted according to the cost changes of high-load path segments to achieve global traffic rebalancing.

[0105] In the optimization process, the AGV path distribution is dynamically adjusted based on the cost changes of high-load path segments. This can be understood as follows: in a multi-AGV system, high-load areas are characterized by high AGV density and low average speed, corresponding to a larger dynamic path weight for the path segments. The system uses this information to automatically adjust the AGV distribution during path planning to achieve traffic balance. Specifically, in each iteration of path optimization, the dynamic path weights of each path segment in the global traffic map are adaptively corrected based on the normalized traffic load coefficient and the global average load. Increasing the weight of high-load paths leads to an increase in total cost, while decreasing the weight of low-load paths leads to a decrease in total cost. This guides some AGVs to avoid high-load paths and choose low-load paths in subsequent A* searches. This process is repeated in each iteration. If a path is still high-load, its weight is further increased, forcing AGVs to choose alternative routes. The iteration continues until the overall system load coefficient approaches the global average, achieving a spatial redistribution of the AGV group, "unloading" congested road segments, and balancing the overall system traffic. Although the total path length may increase slightly, the traffic efficiency and system stability are significantly improved, achieving global traffic rebalancing.

[0106] During continuous optimization iteration, the rate of change of the global average flow load coefficient is monitored. When the rate of change is less than the preset load change threshold, the system flow is determined to have reached a stable equilibrium state.

[0107] The final optimized set of collaborative paths is output, achieving global collaborative control with shortest path, load balancing, and congestion adaptive regulation.

[0108] It should be noted that by quantifying static path length and dynamic path weight into a unified cost function based on the global traffic map, and combining it with the improved A* algorithm to perform optimal path planning for a single AGV, and by calculating the path traffic load coefficient in real time and dynamically adjusting the weight of high-load path segments in the multi-AGV system, the global path distribution is iteratively optimized. This achieves a balanced spatial distribution of the AGV group and congestion suppression, which not only ensures that each AGV can complete its task with the shortest path, but also effectively reduces the overall congestion risk of the system, improves traffic efficiency and the stability of task completion, and achieves an organic combination of individual optimization and global traffic collaborative optimization.

[0109] S3 converts the collaborative path set into control commands, and dynamically compensates for the control commands through load perception to obtain corrected control commands and send them to the AGV.

[0110] In this embodiment, the cooperative path set is converted into control commands, specifically as follows:

[0111] Receive the final optimized set of cooperative paths, where each path consists of a start point, an end point, and a sequence of path nodes, and includes the estimated transit time or timing information for each path segment;

[0112] Curve fitting is performed on each path node sequence to convert the discrete node sequence into a smooth continuous curve. The curve is fitted using cubic spline curves, Bézier curves, or minimum Jerk curves to reduce turning transition errors and path oscillations.

[0113] Based on the smooth and continuous curve, the state data of the curve at each sampling point is calculated. The velocity and acceleration are obtained by the first and second derivatives of the curve, respectively, to provide continuous motion parameters for the generation of subsequent control commands. The state data includes tangent direction, curvature, tangent velocity, and tangent acceleration.

[0114] Speed, acceleration, and curvature are discretized into control instruction units that can be executed by the AGV. Each instruction includes the starting coordinates of the path segment, the ending coordinates, the target speed, the acceleration, and the turning angle.

[0115] Furthermore, the control commands are dynamically compensated through load sensing to obtain corrected control commands, specifically:

[0116] The load data and inertial state data of the AGV are obtained. The load data includes the weight of the goods carried by the AGV, the load distribution, and the center of gravity position. The inertial state data includes the current speed, acceleration, turning angle, and inertial torque of the AGV.

[0117] Based on the load data and inertial state data, the influence of the load on the AGV acceleration is calculated, and the acceleration correction value is obtained.

[0118] The specific formula for calculating the acceleration correction value is as follows:

[0119]

[0120] In the formula, This is the acceleration correction value. The additional force caused by the load, This refers to the total mass of the AGV.

[0121] Based on load data and inertial state data, the influence of load on AGV steering angle is calculated, and the steering angle correction value is obtained.

[0122] The specific formula for calculating the steering angle correction value is as follows:

[0123]

[0124] In the formula, This is the steering angle correction value. This is the steering gain coefficient. For speed, The wheelbase of the AGV. For the quality of goods.

[0125] Based on the acceleration correction value and the steering angle correction value, the control command of the AGV is dynamically adjusted through the PID control algorithm to obtain the corrected control command.

[0126] Furthermore, based on the acceleration correction value and the steering angle correction value, the control command of the AGV is dynamically adjusted through a PID control algorithm to obtain the corrected control command, specifically:

[0127] Based on the acceleration correction value and the steering angle correction value, two independent control systems are established under the PID control framework. The control systems include speed control and steering control.

[0128] For speed control, a target speed is set, and the speed error is calculated by comparing the difference between the target speed and the actual speed.

[0129] For steering control, a target steering angle is set, and the steering error is calculated by comparing the difference between the target steering angle and the actual steering angle.

[0130] Based on speed error and steering error, a PID algorithm is used to calculate the controller's output signal, which includes speed correction command and steering correction command.

[0131] The specific calculation formula for the speed correction command is as follows:

[0132]

[0133] In the formula, For speed correction instructions, For speed error, This represents the cumulative velocity error from time 0 to the current time t. The proportional gain for speed control, For the integral gain of speed control, The differential gain is for speed control.

[0134] The specific calculation formula for the steering correction command is as follows:

[0135]

[0136] In the formula, To redirect the correction instruction, For the proportional gain of steering control, For the integral gain of steering control, The differential gain for steering control, For steering error, This represents the cumulative steering error from time 0 to the current time t.

[0137] Based on the speed correction command and the steering correction command, the speed and steering control commands of the AGV are adjusted respectively to obtain the corrected speed control command and the corrected steering control command.

[0138] The specific calculation formula for the revised speed control command is as follows:

[0139]

[0140] In the formula, The revised speed control command. For the target speed, For time step.

[0141] The specific calculation formula for the revised steering control command is as follows:

[0142]

[0143] In the formula, This is the revised steering control command. Turn towards the target.

[0144] The revised control commands are sent to the AGV's drive system for execution, ensuring that the AGV can maintain accurate trajectory tracking under different load conditions and avoid problems such as excessive acceleration or unstable steering caused by load changes.

[0145] It should be noted that by transforming the cooperative path set into executable continuous control commands and combining it with a load-aware dynamic compensation mechanism, adaptive control of the AGV under different load, speed, and inertia conditions is achieved. Its advantages are: generating smooth, continuous paths through curve fitting effectively reduces path oscillations and steering transition errors; real-time correction of acceleration and steering angle based on load data ensures that control commands reflect true motion characteristics; and dual-channel dynamic adjustment of speed and steering via a PID algorithm achieves precise and stable motion control. This method significantly improves the trajectory tracking accuracy and control robustness of the AGV under complex working conditions, avoids speed fluctuations and steering instability caused by load changes, thereby enhancing the overall safety and reliability of the system.

[0146] 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.

[0147] 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.

[0148] 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.

[0149] 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.

[0150] 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.

[0151] 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. A collaborative congestion avoidance path planning method for AGVs oriented towards real-time traffic conditions, characterized in that, include: A global traffic map is constructed based on AGV status data and dynamically updated. The dynamic update includes weighted updates based on AGV density values ​​and average traffic speeds of each road segment. Constructing a global traffic map based on AGV status data includes time synchronization calibration of each AGV to obtain calibrated AGV status data. The AGV density value is used to characterize the congestion level of each road segment. The dynamic updating of the global traffic map includes: Smooth the AGV density value and average speed, and calculate the smoothed value of the AGV density value and the smoothed value of the average speed. Based on the differences between the smoothed values ​​of AGV density and average speed and the standard data, and combined with the initial weight and weight adjustment coefficient of the path, the updated dynamic path weight is calculated. The weight adjustment coefficient is increased or decreased based on the comparison results by comparing the AGV density value and the rate of change of average speed with a preset threshold. Based on the updated global road condition map, a path planning algorithm is used to generate a globally optimized collaborative path set for AGVs. The path planning algorithm includes a cost function based on static path length and dynamic path weight, and combines path load and traffic for collaborative optimization. The cooperative path set is converted into control commands, and the control commands are dynamically compensated through load sensing to obtain corrected control commands, which are then sent to the AGV. The load sensing dynamic compensation includes acquiring the AGV's load data and inertial state data, calculating acceleration correction values ​​and steering angle correction values, and dynamically adjusting the control commands through a PID control algorithm.

2. The AGV collaborative congestion avoidance path planning method based on real-time traffic conditions according to claim 1, characterized in that, The construction of a global road condition map based on AGV status data includes: Perform time synchronization calibration on each AGV to obtain the calibrated AGV status data; Based on the calibrated AGV status data and static map information, a global road condition map containing the initial weights of path edges is constructed. Based on the global road condition map, calculate the AGV density value and average speed of each path edge; Based on the AGV density value and average speed, the initial weights of the corresponding path edges are dynamically adjusted, and the global road condition map is updated.

3. The AGV collaborative congestion avoidance path planning method based on real-time traffic conditions according to claim 2, characterized in that, The step of performing time synchronization calibration on each AGV to obtain calibrated AGV status data specifically involves: Construct a two-layer clock model including a global clock layer and a local clock layer; Broadcast a global time synchronization signal packet; the AGV node records the local received time and calculates the time offset. Continuously acquire offset samples to form a time series and calculate the time drift trend; A drift compensation function is constructed based on the time drift trend to correct the local clock time; Calculate the update offset based on the correction results, and integrate them to form a global time consistency matrix; Calculate the matrix variance and determine whether the time synchronization accuracy is stable. If it is unstable, the system will adjust the broadcast frequency and generate synchronization status feedback information. The broadcast period will be dynamically adjusted based on the synchronization status feedback information.

4. The AGV cooperative congestion avoidance path planning method based on real-time traffic conditions according to claim 1, characterized in that, The method, based on the updated global road condition map, uses a path planning algorithm to generate a globally optimized collaborative path set for AGVs, including: Single AGV path planning is performed based on a global road condition map, and the optimal path for each AGV is calculated using a cost function consisting of static path length and dynamic path weight. Based on the optimal path of each AGV, the final path set is generated by combining path load and flow indicators for collaborative optimization.

5. The AGV cooperative congestion avoidance path planning method based on real-time traffic conditions according to claim 4, characterized in that, The method of generating a globally optimized collaborative path set for AGVs using a path planning algorithm also includes: Based on the current optimal path of each AGV, the actual AGV density value and actual average speed of each path segment are obtained, the flow load coefficient is calculated and normalized, and high-load path segments are identified. Dynamically adjust AGV path distribution based on changes in cost of high-load path segments; Monitor the rate of change of the global average flow load coefficient. When the rate of change is less than a preset threshold, determine that the system flow has reached a stable equilibrium state and output the final cooperative path set.

6. The AGV cooperative congestion avoidance path planning method based on real-time traffic conditions according to claim 5, characterized in that, The method of generating a globally optimized collaborative path set for AGVs using a path planning algorithm also includes: The dynamic path weight is corrected by the traffic load factor, and then substituted into the cost function to obtain the system cost matrix. Global path collaborative optimization is performed based on the cost matrix to minimize the total system cost function.

7. The AGV cooperative congestion avoidance path planning method based on real-time traffic conditions according to claim 1, characterized in that, The process of converting the cooperative path set into control commands specifically involves: Receives a set of cooperative paths consisting of a start point, an end point, and a sequence of path nodes; Curve fitting is performed on each path node sequence to generate a smooth, continuous curve; Based on the smooth continuous curve, the state data of each sampling point is calculated, and the velocity and acceleration parameters are obtained by curve differentiation. The path is discretized into control command units, each of which contains path segment coordinates, target velocity, acceleration, and steering angle.

8. The AGV cooperative congestion avoidance path planning method based on real-time traffic conditions according to claim 7, characterized in that, The process of dynamically compensating for control commands through load sensing to obtain corrected control commands and then sending them to the AGV involves the following steps: Acquire the load data and inertial state data of the AGV; Based on the load data and inertial state data, calculate the acceleration correction value and the steering angle correction value; The specific formula for calculating the acceleration correction value is as follows: In the formula, This is the acceleration correction value. The additional force caused by the load, The total mass of the AGV; The specific formula for calculating the steering angle correction value is as follows: In the formula, This is the steering angle correction value. This is the steering gain coefficient. For speed, The wheelbase of the AGV. For the quality of goods; Based on the acceleration correction value and the steering angle correction value, the control command of the AGV is dynamically adjusted through the PID control algorithm to obtain the corrected control command.

9. The AGV cooperative congestion avoidance path planning method based on real-time traffic conditions according to claim 8, characterized in that, The process involves dynamically adjusting the AGV's control commands using a PID control algorithm based on the acceleration and steering angle correction values ​​to obtain the corrected control commands. Specifically: Establish independent speed control and steering control systems based on the PID framework; Calculate the speed error and steering error of the control system; Based on speed error and steering error, speed correction commands and steering correction commands are generated respectively using a PID algorithm; Adjust the AGV's speed and steering control commands according to the speed correction command and steering correction command; The revised control commands are sent to the AGV's drive system for execution.

Images