RGV dolly avoidance system and method

Abstract

The invention relates to the technical field of logistics, in particular to an RGV dolly avoidance system and method.The RGV dolly avoidance method is characterized in that a multi-mode sensing module integrates data of a laser radar, a camera and a millimeter wave radar to form an environment sensing report, and an intelligent decision-making module rapidly assesses threats and generates an avoidance strategy based on a deep reinforcement learning algorithm; the dynamic path planning module adopts an improved ant colony algorithm and plans a safe path in combination with obstacle parameters and kinematic characteristics of a trolley, the precise control module drives a steering system to execute the path through a motion control algorithm, and the redundant communication module adopts wireless / wired multi-link to guarantee real-time communication, map the running state in real time and early warn faults. Data support is provided for system optimization, and all the modules cooperate to achieve the functions of environment perception, decision planning, accurate control and state monitoring.

Description

Technical Field

[0001] This invention relates to the field of logistics technology, and in particular to an RGV (Rotation Recycling Vehicle) obstacle avoidance system and method. Background Technology

[0002] RGV carts, as key equipment for automated logistics transportation, play an important role in modern industrial production.

[0003] Currently, existing obstacle avoidance systems have several shortcomings. On the one hand, single-sensor technology is prone to failure in complex environments such as strong light or dense fog, resulting in incomplete information acquisition. On the other hand, obstacle avoidance algorithms based on fixed rules struggle to adapt to dynamically changing obstacles, easily leading to traffic congestion and inefficiency. In pedestrian-vehicle mixed-traffic lanes or high-density warehouse scenarios, existing systems struggle to handle sudden obstacles and dynamic targets in real time, and the lack of deep collaborative optimization between system modules makes it difficult to achieve efficient dynamic path planning and precise control. Currently, there is a lack of intelligent obstacle avoidance solutions fully adapted to complex industrial scenarios such as corrugated cardboard production and high-density warehousing. Summary of the Invention

[0004] The purpose of this invention is to provide an RGV (Road Transport Vehicle) obstacle avoidance system and method, which aims to solve the problem that existing RGV obstacle avoidance systems are unable to adapt to dynamically changing obstacles, easily leading to traffic congestion.

[0005] To achieve the above objectives, in a first aspect, the present invention provides an RGV (Automated Guided Vehicle) obstacle avoidance system, comprising a multimodal perception module, an intelligent decision-making module, a dynamic path planning module, a precise control module, a redundant communication module, a digital twin monitoring module, an intelligent scheduling module, and a system integration interface module. The multimodal perception module, the intelligent decision-making module, the dynamic path planning module, and the precise control module are connected sequentially, and the redundant communication module and the digital twin monitoring module are connected to each other. The intelligent scheduling module is connected to the dynamic path planning module and the system integration interface module, and the system integration interface module is connected to the redundant communication module and the digital twin monitoring module.

[0006] The multimodal perception module is used to collect environmental data around the RGV vehicle;

[0007] The intelligent decision-making module is used to assess threats and formulate avoidance strategies based on perceived data;

[0008] The dynamic path planning module is used to generate specific avoidance paths based on the avoidance strategy;

[0009] The precision control module is used to control the RGV vehicle to accurately avoid obstacles according to the avoidance path;

[0010] The redundant communication module is used to ensure the stability and real-time performance of communication between modules;

[0011] The digital twin monitoring module is used to monitor the system's operating status in real time, provide early warnings of potential faults and offer repair suggestions, and record operating data.

[0012] The intelligent scheduling module is used to optimize the scheduling plan and coordinate the collaborative operation of multiple vehicles based on production tasks and vehicle status.

[0013] The system integration interface module is used to achieve seamless integration with production line and warehouse management systems.

[0014] The multimodal sensing module includes a lidar sensing unit, a visual sensing unit, a millimeter-wave radar sensing unit, a data preprocessing unit, and a data fusion unit.

[0015] The lidar sensing unit is used to detect the distance and position of obstacles and acquire high-precision point cloud data;

[0016] The visual perception unit is used to capture images through a camera and identify the shape and color features of the target;

[0017] The millimeter-wave radar sensing unit is used to detect the target's speed and trajectory.

[0018] The data preprocessing unit is used to synchronize the timestamps and unify the format of the raw data from each sensor;

[0019] The data fusion unit is used to fuse data from different sensors to ensure information integrity and consistency.

[0020] The intelligent decision-making module includes an environment modeling unit, a threat assessment unit, a decision generation unit, and a learning optimization unit.

[0021] The environmental modeling unit is used to establish a dynamic model of the surrounding environment based on the perception data and simulate the movement trend of obstacles.

[0022] The threat assessment unit is used to predict the obstacle threat level based on the environmental model;

[0023] The decision generation unit is used to generate avoidance strategies using deep reinforcement learning algorithms;

[0024] The learning optimization unit is used to update and optimize the parameters of the reinforcement learning model, thereby improving the intelligence and adaptability of the policy.

[0025] The dynamic path planning module includes a global path planning unit, a local path adjustment unit, a smoothing processing unit, a dynamic avoidance unit, and a deadlock detection and recovery unit.

[0026] The global path planning unit is used to generate an initial path based on map information and task requirements;

[0027] The local path adjustment unit is used to adjust the local path according to real-time obstacle information;

[0028] The smoothing unit is used to optimize path smoothness and reduce frequent turning of the vehicle;

[0029] The dynamic obstacle avoidance unit is used to adjust the path in real time to ensure that the vehicle avoids dynamic obstacles;

[0030] The deadlock detection and recovery unit is used to detect and recover path planning deadlock states.

[0031] The precision control module includes a control command parsing unit, a trajectory tracking unit, a feedback adjustment unit, and a power control unit.

[0032] The control command parsing unit is used to parse the control commands transmitted from the path planning module and convert them into execution parameters;

[0033] The trajectory tracking unit is used to track the deviation between the actual trajectory and the planned trajectory of the vehicle in real time;

[0034] The feedback adjustment unit is used to adjust the control parameters according to the deviation to ensure execution accuracy;

[0035] The power control unit is used to directly control the steering motor and drive motor of the vehicle to achieve precise obstacle avoidance.

[0036] The redundant communication module includes a main communication link unit, a redundant link switching unit, a protocol adaptation unit, and a data verification unit.

[0037] The main communication link unit is used for data transmission during normal operation;

[0038] The redundant link switching unit is used to automatically switch to the backup link when the primary link fails.

[0039] The protocol adaptation unit is used to support seamless switching between multiple protocols;

[0040] The aforementioned data verification unit is used to verify the transmitted data to ensure data integrity and correctness.

[0041] Secondly, an RGV (Automated Guided Vehicle) obstacle avoidance method, used in the RGV obstacle avoidance system described in the first aspect, includes the following steps:

[0042] The RGV vehicle collects environmental data around it through a multimodal perception module, including the position, speed, and shape information of obstacles.

[0043] The collected data is transmitted sequentially to the intelligent decision-making module through a redundant communication module. The intelligent decision-making module assesses the threat level based on the perceived data and formulates corresponding avoidance strategies.

[0044] The dynamic path planning module receives the avoidance strategy, generates a specific avoidance path, and transmits the path information to the precision control module.

[0045] The precision control module controls the RGV vehicle to perform precise obstacle avoidance based on the avoidance path;

[0046] The intelligent scheduling module optimizes the scheduling plan and coordinates the collaborative operation of multiple vehicles based on production tasks and vehicle status.

[0047] The system integration interface module enables seamless integration with production line and warehouse management systems;

[0048] The digital twin monitoring module monitors the system's operating status in real time, provides early warnings of potential faults and offers repair suggestions, and records operating data for subsequent analysis and optimization.

[0049] This invention discloses an RGV (Remotely Reinforced Vehicle) obstacle avoidance system. First, a multimodal perception module uses sensors such as LiDAR, cameras, and millimeter-wave radar to collect real-time environmental data of the RGV's surroundings, including the position, speed, and shape of obstacles. This data is integrated to form an environmental perception report, which is then transmitted to an intelligent decision-making module. The intelligent decision-making module uses deep reinforcement learning algorithms to analyze and process the perception data, quickly assessing the threat level and formulating an avoidance strategy. Next, a dynamic path planning module, based on the avoidance strategy, uses path planning algorithms such as improved ant colony optimization to generate a new, safe driving path for the RGV, fully considering the position and size of surrounding obstacles and the vehicle's own kinematic characteristics. A precision control module, following the planned path, uses precise motion control algorithms to manipulate the RGV's drive and steering systems, enabling the vehicle to travel stably along the planned path and achieve precise obstacle avoidance. Throughout the process, a redundant communication module establishes multiple communication links, such as wireless and wired networks, to ensure the stability and real-time nature of information transmission between modules. When the main communication link fails, the redundant links quickly switch, ensuring the system's normal operation. Meanwhile, the digital twin monitoring module establishes a virtual digital model corresponding to the actual RGV vehicle and its obstacle avoidance system, mapping the vehicle's operating status and obstacle avoidance process in real time. This allows for the early detection of potential fault risks, timely issuance of warning signals, and recording of operational data, providing a basis for subsequent system optimization and fault analysis. This solves the problem of existing RGV obstacle avoidance systems being unable to adapt to dynamically changing obstacles, which can easily lead to traffic congestion. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a schematic diagram of an RGV vehicle obstacle avoidance system provided by the present invention.

[0052] Figure 2 This is a schematic diagram of a multimodal sensing module.

[0053] Figure 3 This is a schematic diagram of the intelligent decision-making module.

[0054] Figure 4 This is a schematic diagram of the dynamic path planning module.

[0055] Figure 5 This is a schematic diagram of the precision control module.

[0056] Figure 6 This is a schematic diagram of a redundant communication module.

[0057] Figure 7 This is a schematic diagram of a digital twin monitoring module.

[0058] Figure 8 This is a schematic diagram of the intelligent scheduling module.

[0059] Figure 9 This is a schematic diagram of the system integration interface module.

[0060] Figure 10 This is a flowchart of an RGV vehicle obstacle avoidance method provided by the present invention.

[0061] In the diagram: 1-Multimodal perception module, 2-Intelligent decision-making module, 3-Dynamic path planning module, 4-Precise control module, 5-Redundant communication module, 6-Digital twin monitoring module, 7-Intelligent scheduling module, 8-System integration interface module, 11-LiDAR perception unit, 12-Visual perception unit, 13-Millimeter-wave radar perception unit, 14-Data preprocessing unit, 15-Data fusion unit, 21-Environmental modeling unit, 22-Threat assessment unit, 23-Decision generation unit, 24-Learning optimization unit, 31-Global path planning unit, 32-Local path adjustment unit, 33-Smoothing unit, 34-Dynamic avoidance unit, 35-Dead Lock detection and recovery unit, 41-control command parsing unit, 42-trajectory tracking unit, 43-feedback adjustment unit, 44-power control unit, 51-main communication link unit, 52-redundant link switching unit, 53-protocol adaptation unit, 54-data verification unit, 61-real-time mapping unit, 62-status monitoring unit, 63-fault diagnosis unit, 64-optimization suggestion unit, 71-task allocation unit, 72-task allocation unit, 73-scheduling optimization unit, 71-multi-vehicle collaboration unit, 81-data conversion unit, 82-protocol conversion unit, 83-interface adaptation unit, 84-system monitoring unit. Detailed Implementation

[0062] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0063] Please see Figures 1 to 9 In a first aspect, the present invention provides an RGV (Automated Guided Vehicle) obstacle avoidance system, comprising a multimodal perception module 1, an intelligent decision-making module 2, a dynamic path planning module 3, a precise control module 4, a redundant communication module 5, a digital twin monitoring module 6, an intelligent scheduling module 7, and a system integration interface module 8. The multimodal perception module 1, the intelligent decision-making module 2, the dynamic path planning module 3, and the precise control module 4 are connected in sequence and are respectively connected to the redundant communication module 5 and the digital twin monitoring module 6. The intelligent scheduling module 7 is connected to the dynamic path planning module 5 and the system integration interface module 8. The system integration interface module 8 is respectively connected to the redundant communication module 5 and the digital twin monitoring module 6.

[0064] The multimodal perception module 1 is used to collect environmental data around the RGV vehicle;

[0065] The intelligent decision-making module 2 is used to assess threats and formulate avoidance strategies based on perceived data;

[0066] The dynamic path planning module 3 is used to generate specific avoidance paths based on the avoidance strategy;

[0067] The precision control module 4 is used to control the RGV vehicle to accurately avoid obstacles according to the avoidance path;

[0068] The redundant communication module 5 is used to ensure the stability and real-time performance of communication between modules;

[0069] The digital twin monitoring module 6 is used to monitor the system's operating status in real time, provide early warnings of potential faults and offer repair suggestions, and record operating data.

[0070] In this embodiment, firstly, the multimodal perception module 1 uses sensors such as LiDAR, cameras, and millimeter-wave radar to collect real-time environmental data of the RGV vehicle, including information such as the position, speed, and shape of obstacles. This data is integrated to form an environmental perception report and transmitted to the intelligent decision-making module 2. The intelligent decision-making module 2 uses deep reinforcement learning algorithms to analyze and process the perception data, quickly assessing the threat level and formulating avoidance strategies. Next, the dynamic path planning module 3, based on the avoidance strategy, uses path planning algorithms such as improved ant colony algorithms to generate a new, safe driving path for the RGV vehicle, fully considering factors such as the position and size of surrounding obstacles and the vehicle's own kinematic characteristics. The precision control module 4, following the planned path, uses precise motion control algorithms to operate the RGV vehicle's drive and steering systems, enabling the vehicle to travel stably along the planned path and achieve precise avoidance. Throughout the process, the redundant communication module 5 constructs multiple communication links, such as wireless networks and wired networks, to ensure the stability and real-time nature of information transmission between modules. When the main communication link fails, the redundant links quickly switch to ensure normal system operation. Meanwhile, the digital twin monitoring module 6 establishes a virtual digital model corresponding to the actual RGV vehicle and its obstacle avoidance system. This model maps the vehicle's operating status and obstacle avoidance process in real time, enabling early detection of potential fault risks, timely issuance of warning signals, and recording of operational data. This provides a basis for subsequent system optimization and fault analysis. Consequently, it solves the problem of existing RGV obstacle avoidance systems being unable to adapt to dynamically changing obstacles, which can easily lead to traffic congestion.

[0071] Furthermore, the multimodal sensing module 1 includes a lidar sensing unit 11, a visual sensing unit 12, a millimeter-wave radar sensing unit 13, a data preprocessing unit 14, and a data fusion unit 15;

[0072] The lidar sensing unit 11 is used to detect the distance and position of obstacles and acquire high-precision point cloud data;

[0073] The visual perception unit 12 is used to capture images through the camera and identify the shape and color features of the target;

[0074] The millimeter-wave radar sensing unit 13 is used to detect the target's speed and trajectory.

[0075] The data preprocessing unit 14 is used to synchronize the timestamps and unify the format of the raw data from each sensor.

[0076] The data fusion unit 15 is used to fuse data from different sensors to ensure information integrity and consistency.

[0077] In this embodiment, the lidar sensing unit 11 is a lidar, installed at the front of the RGV vehicle. It rotates at a speed of 10-20 revolutions per second to emit laser pulses, receives reflected light to calculate the distance and position of obstacles, acquires point cloud data, and then uses a clustering algorithm to distinguish different obstacles, achieving accurate detection. Its advantages lie in high precision and anti-interference, but it is limited by environmental factors such as weather and dust. The visual sensing unit 12 is a color camera, installed on the side of the vehicle, capturing images at 30 frames per second. After image preprocessing, it uses a convolutional neural network (CNN) to identify the shape and color features of the target, such as squares, blue, etc. This unit can provide rich visual information to assist in identifying obstacle types, but it is greatly affected by lighting and occlusion, and needs to be combined with other sensing units. The millimeter-wave radar sensing unit 13 is a millimeter-wave radar, installed at the rear of the vehicle. It emits millimeter waves and receives reflected signals, calculates the target speed and trajectory based on the Doppler effect, and its speed measurement accuracy can reach 0.01 m / s. It has all-weather working capability and can be used in complex weather conditions, but its resolution is relatively low, and its ability to detect stationary targets is limited. The data preprocessing unit 14 transmits data from each sensor to the data preprocessing unit 14 for timestamp synchronization and format unification. During timestamp synchronization, all data is aligned to the system timestamp, with errors controlled within milliseconds. Format unification standardizes different data formats into intermediate formats, such as converting LiDAR point cloud data to XYZ format and image data to grayscale, providing standardized data for subsequent fusion. The data fusion unit 15 uses a Kalman filter algorithm to fuse the preprocessed multi-source data. The Kalman filter predicts the current obstacle state, and then the LiDAR, visual, and millimeter-wave radar data are fused to update the state, reducing errors and uncertainties. The fused data contains complete information such as obstacle position, speed, shape, and color. Verification shows that compared to a single sensor, the fused perception accuracy is improved by 40%-60%, providing a reliable basis for the intelligent decision-making module 2.

[0078] Furthermore, the intelligent decision-making module 2 includes an environment modeling unit 21, a threat assessment unit 22, a decision generation unit 23, and a learning optimization unit 24;

[0079] The environmental modeling unit 21 is used to establish a dynamic model of the surrounding environment based on the perception data and simulate the movement trend of obstacles.

[0080] The threat assessment unit 22 is used to predict the obstacle threat level based on the environmental model;

[0081] The decision generation unit 23 is used to generate an avoidance strategy using a deep reinforcement learning algorithm;

[0082] The learning optimization unit 24 is used to update and optimize the parameters of the reinforcement learning model to improve the intelligence and adaptability of the policy.

[0083] In this embodiment, the environment modeling unit 21 constructs a dynamic model of the vehicle's surrounding environment using the Monte Carlo localization algorithm based on data from the multimodal perception module 1. This model includes information such as obstacle positions and speeds, and is combined with map data for precise localization. The model is updated 10 times per second. This unit provides basic support for subsequent threat assessment and decision generation. However, in complex dynamic environments, the accuracy and real-time performance of the model may be affected. The threat assessment unit 22, based on the dynamic model constructed by the environment modeling unit 21, uses an assessment algorithm based on Dempster-Shafer evidence theory to predict the threat level of obstacles. This algorithm comprehensively considers factors such as the distance, speed, and direction of movement of obstacles to calculate the degree of threat posed by obstacles to the RGV vehicle. Threat levels are divided into low, medium, and high levels, and the assessment results provide important basis for the decision generation unit 23. In cases where obstacle movement is complex or information is incomplete, the accuracy of the threat assessment may decrease. The decision generation unit 23 generates avoidance strategies using a deep reinforcement learning algorithm. Employing a Deep Q-Network (DQN) algorithm, it takes perceived data and obstacle threat levels as input, and the vehicle's safety and traffic efficiency as reward signals to train the vehicle's avoidance strategies in different scenarios. Based on real-time data, the decision generation unit 23 quickly selects the optimal avoidance strategy and controls the vehicle to perform the corresponding operation. Deep reinforcement learning algorithms require a large amount of training data to ensure the reliability of the strategy; when facing rare or entirely new scenarios, further learning and adaptation may be necessary. The learning optimization unit 24 is responsible for updating and optimizing the deep reinforcement learning model parameters. It uses experience replay technology and a target network to stabilize the training process, collects the vehicle's operational data in various scenarios, including successful and failed avoidance cases, and continuously adjusts the network parameters. By combining online learning and offline optimization, the learning optimization unit 24 can improve the intelligence and adaptability of the strategy. Model optimization is a continuous process that requires constantly collecting new data and training to adapt to environmental changes and changes in the vehicle's own performance.

[0084] Furthermore, the dynamic path planning module 3 includes a global path planning unit 31, a local path adjustment unit 32, a smoothing processing unit 33, a dynamic avoidance unit 34, and a deadlock detection and recovery unit 35.

[0085] The global path planning unit 31 is used to generate an initial path based on map information and task requirements;

[0086] The local path adjustment unit 32 is used to adjust the local path according to real-time obstacle information;

[0087] The smoothing unit 33 is used to optimize path smoothness and reduce frequent turning of the vehicle;

[0088] The dynamic obstacle avoidance unit 34 is used to adjust the path in real time to ensure that the vehicle avoids dynamic obstacles;

[0089] The deadlock detection and recovery unit 35 is used to detect and recover the path planning deadlock state.

[0090] In this embodiment, the global path planning unit 31 generates an initial path based on map information and task requirements using an improved algorithm. A dynamic weight adjustment mechanism is introduced, comprehensively considering factors such as path distance, obstacle distribution, and traffic flow, with the weight coefficients dynamically adjusted. This unit provides the vehicle with overall navigation direction, but when the dynamic environment changes significantly, the global path may need frequent replanning. The local path adjustment unit 32 adjusts the local path using an artificial potential field method based on real-time obstacle information. By calculating the repulsive force of obstacles and the attractive force of the target point, the vehicle is guided to avoid obstacles. The local path adjustment unit 32 can quickly respond to changes in the surrounding environment, but in complex and narrow passages, the artificial potential field method may encounter local minima, leading to path planning difficulties. The smoothing unit 33 optimizes the path using a Bezier curve algorithm, reducing frequent turning by the vehicle. The smoothing unit 33 can improve the stability and comfort of the vehicle's driving, but when the path curvature changes significantly, smoothing may conflict with obstacle avoidance requirements, requiring further balancing. The dynamic obstacle avoidance unit 34 adjusts the path in real time to ensure the vehicle avoids dynamic obstacles. It employs a model predictive control (MPC) algorithm to monitor the movement status of dynamic obstacles in real time. The dynamic obstacle avoidance unit 34 can effectively handle sudden situations involving dynamic obstacles, but the difficulty of dynamic avoidance increases significantly when the obstacle's trajectory is difficult to predict. The deadlock detection and recovery unit 35 monitors the path planning status, detects deadlock states, and uses a backtracking algorithm to recover the path planning. The deadlock detection and recovery unit 35 can, to some extent, prevent the vehicle from getting stuck, but in complex environments, deadlock recovery may take a considerable amount of time, affecting the vehicle's passage efficiency.

[0091] Furthermore, the precision control module 4 includes a control command parsing unit 41, a trajectory tracking unit 42, a feedback adjustment unit 43, and a power control unit 44;

[0092] The control command parsing unit 41 is used to parse the control commands transmitted from the path planning module and convert them into execution parameters.

[0093] The trajectory tracking unit 42 is used to track the deviation between the actual trajectory and the planned trajectory of the vehicle in real time.

[0094] The feedback adjustment unit 43 is used to adjust the control parameters according to the deviation to ensure execution accuracy.

[0095] The power control unit 44 is used to directly control the steering motor and drive motor of the vehicle to achieve precise obstacle avoidance.

[0096] In this embodiment, the control command parsing unit 41 receives control commands from the dynamic path planning module 3, parses them into execution parameters such as steering angle and speed, and converts the path planning results into specific control signals based on the vehicle's kinematic model and dynamic characteristics. The control command parsing unit 41 provides basic parameters for subsequent precise control, but accuracy loss or parameter mismatch may occur during the parsing process. The trajectory tracking unit 42 tracks the deviation between the vehicle's actual trajectory and the planned trajectory in real time, employing a PID control algorithm to adjust control parameters to make the vehicle's trajectory as close as possible to the planned trajectory. The trajectory tracking unit 42 effectively improves the vehicle's control accuracy, but the accuracy of trajectory tracking may be affected when the vehicle is subjected to external interference or changes in the power system performance. The feedback adjustment unit 43 adjusts the control parameters according to the deviation to ensure execution accuracy, using an extended Kalman filter (EKF) algorithm to estimate the vehicle's state in real time. The feedback adjustment unit 43 enhances the robustness of the control system, but the effect of feedback adjustment may be limited when the system model is inaccurate or noise is high. The power control unit 44 directly controls the steering motor and drive motor of the vehicle to achieve precise obstacle avoidance. It employs vector control technology to precisely adjust the motor speed and torque. While the power control unit 44 ensures the vehicle's motion control performance, the stability of the power control may be challenged when motor parameters change or the load suddenly changes.

[0097] Furthermore, the redundant communication module 5 includes a main communication link unit 51, a redundant link switching unit 52, a protocol adaptation unit 53, and a data verification unit 54.

[0098] The main communication link unit 51 is used for data transmission during normal operation;

[0099] The redundant link switching unit 52 is used to automatically switch to the backup link when the primary link fails.

[0100] The protocol adaptation unit 53 is used to support seamless switching between multiple protocols;

[0101] The data verification unit 54 is used to verify the transmitted data to ensure data integrity and correctness.

[0102] In this embodiment, the main communication link unit 51 uses a 5G network as the data transmission link during normal operation, leveraging its high bandwidth and low latency characteristics to ensure the stability and real-time performance of data transmission between modules. The main communication link unit 51 can meet the communication needs of the vehicle in high-speed movement and complex environments, but communication quality may degrade when 5G signal coverage is insufficient or interfered with. The redundant link switching unit 52 monitors the status of the main communication link in real time. When the main link fails, it quickly switches to the backup link, with the switching time controlled within seconds. The redundant link switching unit 52 improves system reliability, but brief data loss or communication interruptions may occur during the switching process. The protocol adaptation unit 53 supports seamless switching between multiple communication protocols, such as TCP / IP and CAN, automatically selecting the appropriate protocol for data transmission based on different communication scenarios and module requirements. The protocol adaptation unit 53 enhances system compatibility, but low protocol conversion efficiency or data format mismatch may occur when switching between different protocols. The data verification unit 54 uses a CRC check algorithm to verify the transmitted data, ensuring data integrity and correctness, with a verification cycle of 10 times per second. The data verification unit 54 can effectively reduce the data transmission error rate, but in high data traffic or high noise environments, the burden of data verification may increase, affecting communication efficiency.

[0103] Furthermore, the digital twin monitoring module 6 includes a real-time mapping unit 61, a status monitoring unit 62, a fault diagnosis unit 63, and an optimization suggestion unit 64;

[0104] The real-time mapping unit 61 is used to map the physical system's operating state to the digital model in real time.

[0105] The status monitoring unit 62 is used to monitor the operating parameters of the vehicle and the module;

[0106] The fault diagnosis unit 63 is used to analyze monitoring data and diagnose potential faults;

[0107] The optimization suggestion unit 64 is used to generate system optimization suggestions based on the diagnostic results.

[0108] In this embodiment, the real-time mapping unit 61 uses lightweight digital twin technology to map the operating status of the physical RGV vehicle and its obstacle avoidance system to a digital model in real time, updating 10 times per second. This unit can detect potential faults in advance and provide early warnings and repair suggestions. However, due to limitations in computing resources and modeling accuracy, real-time mapping may experience delays or information loss in highly complex scenarios. The status monitoring unit 62 monitors the status parameters of the vehicle and its modules in real time, such as temperature, voltage, and rotational speed, using sensor network technology, with a monitoring frequency of 50 times per second. The status monitoring unit 62 can detect anomalies promptly, but the complexity and cost of the monitoring system increase when there are numerous or widely distributed sensors. The fault diagnosis unit 63 uses a machine learning-based fault diagnosis algorithm to analyze the monitoring data and diagnose potential fault causes. The fault diagnosis unit 63 achieves rapid and accurate fault diagnosis by training a fault feature extraction model. The accuracy of fault diagnosis may be affected when fault features are not obvious or when data noise is high. Based on the diagnostic results, the optimization suggestion unit 64 provides system optimization suggestions using a knowledge base and optimization algorithms. For example, when the vehicle's power system load is too high, it suggests adjusting the task scheduling strategy. However, the feasibility and effectiveness of the optimization suggestions depend on the completeness of the knowledge base and the performance of the optimization algorithm.

[0109] Furthermore, the intelligent scheduling module 7 includes a task allocation unit 71, a task allocation unit 72, a scheduling optimization unit 73, and a multi-vehicle coordination unit 74;

[0110] The task allocation unit 71 is used to allocate transportation tasks according to production needs.

[0111] The task allocation unit 72 is used to monitor and manage production line resources;

[0112] The scheduling optimization unit 73 is used to optimize the scheduling plan using genetic algorithms and other methods.

[0113] The multi-vehicle coordination unit 74 is used to coordinate the operation of multiple vehicles and avoid conflicts.

[0114] In this embodiment, the task allocation unit 71 acquires the task requirements of the production system, performs a comprehensive evaluation based on the vehicle status (position, load, task priority), and rationally allocates transportation tasks using a weighted round-robin algorithm. The task allocation results are then updated to the vehicle control module to ensure efficient task execution. The task allocation unit 72 monitors production line resources (vehicles, charging piles, loading / unloading points) in real time, records their location, status, and usage, analyzes resource utilization efficiency, predicts supply and demand changes, and dynamically adjusts resource allocation strategies to ensure task execution. The fitness optimization unit collects task and resource data, initializes a population of scheduling schemes, evaluates the merits of schemes based on a fitness function, selects superior schemes using a genetic algorithm for crossover and mutation operations, and generates an optimized scheduling plan to improve task completion efficiency. The multi-vehicle collaboration unit 74 establishes a vehicle communication network, shares location and speed information in real time, adjusts speed and path based on a consensus algorithm, avoids dynamic obstacles and traffic conflicts, and ensures safe and efficient multi-vehicle operations.

[0115] Furthermore, the system integration interface module 8 includes a data conversion unit 81, a protocol conversion unit 82, an interface adaptation unit 83, and a system monitoring unit 84;

[0116] The data conversion unit 81 is used to convert data formats;

[0117] The protocol conversion unit 82 is used to adapt to different communication protocols;

[0118] The interface adaptation unit 83 is used to provide a standardized interface;

[0119] The system monitoring unit 84 is used to monitor the operating status of the integrated system.

[0120] In this embodiment, the data conversion unit 81 receives raw data, identifies the format type, parses and reassembles the data structure according to the target format, and uses an interpolation algorithm to unify data precision, thus completing the data conversion task. The protocol conversion unit 82 receives communication data, parses the source protocol type, extracts the payload, maps it to the target protocol data format, and repackages it into a data frame conforming to the target protocol standard, achieving seamless conversion between different protocol data. The interface adaptation unit 83 parses the parameter specifications and semantics of received system interface requests, converts the parameters to the format required by the target system, and simulates the target system interface behavior for testing and verification, ensuring the accuracy and reliability of interface adaptation. The system monitoring unit 84 collects the integrated system's operating data in real time, analyzes the system status according to preset rules, assesses the system health, promptly detects anomalies and issues warnings, providing monitoring support for the stable operation of the system.

[0121] Secondly, an RGV (Automated Guided Vehicle) obstacle avoidance method, used in the RGV obstacle avoidance system described in the first aspect, includes the following steps:

[0122] S1 collects environmental data around the RGV vehicle through the multimodal perception module 1, including the position, speed, and shape information of obstacles;

[0123] Specifically, the multimodal perception module 1 utilizes devices such as LiDAR, visual sensors, and millimeter-wave radar to acquire real-time information about the vehicle's surrounding environment. LiDAR determines the distance and location of obstacles by emitting laser pulses and measuring reflection time; visual sensors capture images and identify the shape and color of obstacles; and millimeter-wave radar detects the speed and trajectory of obstacles. After preprocessing, this data is integrated using data fusion technology to form unified environmental perception data, providing comprehensive and accurate information for subsequent decision-making.

[0124] S2 transmits the collected data sequentially to the intelligent decision-making module 2 through the redundant communication module 5. The intelligent decision-making module assesses the threat level based on the perceived data and formulates corresponding avoidance strategies.

[0125] Specifically, redundant communication module 5 employs diverse communication links (such as 5G, wired networks, etc.) to ensure the stability and real-time performance of data transmission. The perceived data first undergoes preliminary processing to remove noise and redundant information, and then is sent to intelligent decision-making module 2 via the communication module. Intelligent decision-making module 2 uses deep reinforcement learning algorithms, combined with real-time environmental data, to assess the threat level of current obstacles to the vehicle's operation. Based on the threat level, the decision-making module formulates corresponding avoidance strategies, such as adjusting the driving direction, speed, or path planning, to ensure the vehicle safely and efficiently avoids obstacles.

[0126] The S3 dynamic path planning module 3 receives the avoidance strategy, generates a specific avoidance path, and transmits the path information to the precision control module 4;

[0127] Specifically, after receiving the obstacle avoidance strategy from the intelligent decision-making module 2, the dynamic path planning module 3 uses improved ant colony optimization and other path planning algorithms to comprehensively consider the obstacle's position and speed, the vehicle's kinematic constraints, and the traffic rules of the production environment to generate a safe and efficient obstacle avoidance path. This path not only avoids obstacles but also optimizes path length and travel time, reducing unnecessary detours. The generated path information is then transmitted to the precision control module 4 via the communication module to guide the vehicle's precise obstacle avoidance.

[0128] S4 Precision Control Module 4 controls the RGV vehicle to perform precise avoidance according to the avoidance path;

[0129] Specifically, after receiving path information from the dynamic path planning module 3, the precision control module 4 parses the control commands and converts them into specific execution parameters, such as steering angle and speed adjustment. The control module employs a feedforward-feedback combined model predictive control algorithm to monitor the deviation between the vehicle's actual position and the planned path in real time, and dynamically adjusts the control parameters to ensure the vehicle travels along the planned path with centimeter-level accuracy, precisely completing obstacle avoidance maneuvers. Simultaneously, the control module continuously provides feedback on the vehicle's operating status to ensure the coordination and stability of the entire system.

[0130] The S5 intelligent scheduling module 7 optimizes the scheduling plan and coordinates the collaborative operation of multiple vehicles based on production tasks and vehicle status.

[0131] Specifically, the intelligent scheduling module first receives tasks from the production management system and parses key information, while continuously monitoring the position, load, battery level, and operating status of each RGV (Remotely Rated Vehicle). Based on task priority and vehicle status, it uses optimization algorithms such as genetic algorithms to formulate an initial scheduling plan, determining the task and path for each vehicle. When multiple vehicles are operating in the same area, the module adjusts the vehicle's speed and direction in real time by setting virtual tracks and safety distances to avoid collisions and congestion. During task execution, the module monitors task progress and vehicle status in real time. If any abnormalities occur, the scheduling plan is immediately redesigned to ensure efficient task completion.

[0132] The S6 system integration interface module 8 enables seamless integration with production line and warehouse management systems;

[0133] Specifically, the system integration interface module 8 is responsible for data format conversion and communication protocol adaptation between the RGV system and the production line and warehouse management systems. For example, it converts the TCP / IP protocol used internally by the RGV system to the Profibus-DP protocol commonly used by production line equipment, ensuring accurate data transmission and interpretation. It provides standardized interfaces, reducing the complexity of integration with other systems, and simultaneously synchronizes the status information between the RGV system and the production line and warehouse management systems in real time, ensuring information consistency between the systems. Furthermore, this module monitors for faults during system integration. When communication is interrupted or data transmission errors occur, it automatically executes diagnostic procedures and attempts to restore the connection, ensuring stable system operation.

[0134] The S7 digital twin monitoring module 6 monitors the system's operating status in real time, provides early warnings of potential faults and offers repair suggestions, and records operating data for subsequent analysis and optimization.

[0135] Specifically, the digital twin monitoring module 6 constructs a virtual digital model corresponding to the actual RGV vehicle and its obstacle avoidance system. Through real-time mapping technology, this module maps the vehicle's operating status, the working parameters of each module, and key data during the obstacle avoidance process onto the digital model in real time. The status monitoring unit 62 continuously tracks the operating parameters of the vehicle and each module, such as temperature, voltage, and speed. Utilizing big data analysis and machine learning algorithms, the fault diagnosis unit 63 can identify potential fault risks in advance and issue timely warning signals. Based on the diagnostic results, combined with historical data and the system knowledge base, the optimization suggestion unit 64 provides specific suggestions for system optimization and adjustment, while recording detailed operating data to provide strong support for subsequent system improvements and fault analysis.

[0136] Beneficial effects:

[0137] I. The RGV (Remotely Reinforced Vehicle) obstacle avoidance system and method provided by this invention improves the obstacle avoidance success rate: the multimodal perception module can accurately perceive the surrounding environment, providing sufficient data support for intelligent decision-making; the intelligent decision-making module, based on deep reinforcement learning algorithms, can quickly formulate the optimal obstacle avoidance strategy; the path generated by the dynamic path planning module is safe and reliable; and the precision control module ensures that the vehicle can accurately execute obstacle avoidance actions. These closely coordinated components greatly improve the obstacle avoidance success rate of the RGV in complex environments.

[0138] II. The RGV vehicle obstacle avoidance system and method provided by this invention enhances system stability: the redundant communication module effectively avoids system paralysis caused by communication failures, thus enhancing the stability of the entire obstacle avoidance system. The digital twin monitoring module can monitor the system's operating status in real time, provide early warnings of potential faults, and ensure the long-term stable operation of the system.

[0139] Third, the RGV vehicle obstacle avoidance system and method provided by this invention improves the level of intelligence: the obstacle avoidance system adopts advanced deep reinforcement learning algorithms, improved ant colony algorithms and other artificial intelligence technologies, which enables the RGV vehicle to have a high degree of autonomous decision-making and autonomous obstacle avoidance capabilities, improves the overall intelligence level of the system, and reduces the dependence on human intervention.

[0140] IV. The RGV vehicle obstacle avoidance system and method provided by this invention are adaptable to various environments: the multimodal perception module integrates data from multiple sensors, enabling the system to adapt to different lighting conditions, weather conditions, and complex industrial environments. Whether in strong light, dense fog, or scenarios involving both pedestrians and vehicles, the system can operate stably and achieve reliable obstacle avoidance functionality.

[0141] V. The RGV vehicle obstacle avoidance system and method provided by this invention optimizes operating efficiency: Through precise path planning and accurate control, the RGV vehicle can maintain a high operating speed and efficiency while avoiding obstacles, reducing parking time and path detours caused by obstacle avoidance, and improving the operating efficiency of the entire logistics system.

[0142] The above description is merely a preferred embodiment of the RGV vehicle obstacle avoidance system and method of the present invention, and should not be construed as limiting the scope of the present invention. Those skilled in the art can understand that all or part of the processes of the above embodiments can be implemented, and equivalent changes made in accordance with the claims of the present invention are still within the scope of the invention.

Claims

1. An RGV vehicle avoidance system, characterized by, The system comprises a multi-modal perception module, an intelligent decision-making module, a dynamic path planning module, a precise control module, a redundant communication module, a digital twin monitoring module, an intelligent scheduling module, and a system integration interface module. The multi-modal perception module is configured to collect surrounding environment data of the RGV. The intelligent decision-making module is configured to evaluate threats and develop avoidance strategies based on the perception data. The dynamic path planning module is configured to generate specific avoidance paths based on the avoidance strategies. The precise control module is configured to control the RGV to avoid obstacles accurately based on the avoidance paths. The redundant communication module is configured to ensure the stability and real-time performance of communication between modules. The digital twin monitoring module is configured to monitor the system's running state in real time, provide early warning and repair suggestions for potential faults, and record running data. The intelligent scheduling module is configured to optimize scheduling plans and coordinate multi-vehicle collaborative work based on production tasks and vehicle states. The system integration interface module is configured to achieve seamless integration with production lines and warehouse management systems.

2. The RGV avoidance system of claim 1, wherein: The multi-modal perception module comprises a laser radar perception unit, a visual perception unit, a millimeter wave radar perception unit, a data preprocessing unit, and a data fusion unit. The laser radar perception unit is configured to detect obstacle distance and position and obtain high-precision point cloud data. The visual perception unit is configured to capture images through a camera and identify target shape and color features. The millimeter wave radar perception unit is configured to detect target speed and motion trajectory. The data preprocessing unit is configured to synchronize timestamps and unify formats for raw data from various sensors. The data fusion unit is configured to fuse data from different sensors to ensure information integrity and consistency.

3. The RGV avoidance system of claim 1, wherein: The intelligent decision-making module comprises an environment modeling unit, a threat assessment unit, a decision-making generation unit, and a learning optimization unit. The environment modeling unit is configured to establish a dynamic model of the surrounding environment based on perception data and simulate obstacle movement trends. The threat assessment unit is configured to predict obstacle threat levels based on the environment model. The decision-making generation unit is configured to generate avoidance strategies using deep reinforcement learning algorithms. The learning optimization unit is configured to update and optimize reinforcement learning model parameters to improve strategy intelligence and adaptability.

4. The RGV avoidance system of claim 1, wherein: The dynamic path planning module comprises a global path planning unit, a local path adjustment unit, a smoothing processing unit, a dynamic avoidance unit, and a deadlock detection and recovery unit. The global path planning unit is configured to generate an initial path according to map information and task requirements. The local path adjustment unit is configured to adjust a local path according to real-time obstacle information. The smoothing processing unit is configured to optimize path smoothness and reduce frequent turning of the trolley. The dynamic avoidance unit is configured to adjust the path in real time to ensure that the trolley avoids dynamic obstacles. The deadlock detection and recovery unit is configured to detect and recover from a path planning deadlock state.

5. The RGV trolley avoidance system of claim 1, wherein The precision control module includes a control instruction analysis unit, a trajectory tracking unit, a feedback adjustment unit, and a power control unit. The control instruction analysis unit is configured to analyze control instructions from the path planning module and convert them into execution parameters. The trajectory tracking unit is configured to track deviations between actual and planned trajectories of the trolley in real time. The feedback adjustment unit is configured to adjust control parameters based on deviations to ensure execution accuracy. The power control unit is configured to directly control trolley steering and drive motors to achieve precise avoidance.

6. The RGV trolley avoidance system of claim 1, wherein The redundant communication module includes a main communication link unit, a redundant link switching unit, a protocol adaptation unit, and a data verification unit. The main communication link unit is configured for data transmission during normal operation. The redundant link switching unit is configured to automatically switch to a backup link in the event of a main link failure. The protocol adaptation unit is configured to support seamless switching between multiple protocols. The data verification unit is configured to verify transmitted data to ensure data integrity and correctness.

7. A RGV vehicle avoiding method for the RGV vehicle avoiding system of any one of claims 1-6, characterized in that, The method includes the following steps: Collecting RGV trolley surrounding environment data, including obstacle position, speed, and shape information, through a multi-modal perception module; Transferring collected data to an intelligent decision-making module through a redundant communication module, which evaluates threat levels based on perception data and develops corresponding avoidance strategies. A dynamic path planning module receives avoidance strategies, generates specific avoidance paths, and transmits path information to a precision control module. The precision control module controls the RGV trolley to avoid obstacles precisely based on avoidance paths. An intelligent scheduling module optimizes scheduling plans based on production tasks and trolley states to coordinate multi-trolley collaborative work. A system integration interface module enables seamless integration with production lines and warehouse management systems. A digital twin monitoring module monitors system operation in real time, provides early warnings and repair recommendations for potential faults, and records operation data for subsequent analysis and optimization.

Images