Logistics robot travel path planning method based on digital twin technology

Abstract

The invention relates to the technical field of intelligent logistics and path planning, and discloses a digital twinborn technology-based logistics robot path planning method, which comprises the steps of constructing a digital twinborn model consistent with a storage environment, identifying logistics path key nodes and obstacles, generating a plurality of candidate paths and carrying out simulation verification, and screening an optimal path according to simulation feedback, and finally issuing the path to the robot for execution. By establishing a virtual and physical two-way synchronous environment model, efficient simulation and prediction of a logistics path are realized, so that the intelligent level of robot scheduling and the stability of path advancing are improved, and the method is suitable for complex logistics scenes such as multi-robot cooperation and dynamic path optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent logistics and path planning, and particularly to a method for planning the path of a logistics robot based on digital twin technology. Background Art

[0002] Currently, in an automated logistics warehousing system, robots, as the core execution units, undertake key tasks such as handling, sorting, and path scheduling. Although traditional path planning methods such as A* and Dijkstra can be used for static path calculation, they have problems such as unstable paths, high execution delays, and weak environmental adaptability when facing dynamic obstacles, real-time scheduling conflicts, and complex task intersections.

[0003] At the same time, existing scheduling systems mostly rely on local sensor information or static maps, lacking the global perception ability of the overall warehousing dynamics, unable to predict and adjust the robot's behavior in real time through simulation, and prone to causing traffic jams, path conflicts, or task failures.

[0004] Therefore, it is urgent to introduce a dynamic perception and path simulation mechanism based on digital twin to achieve real-time planning and intelligent control of the robot's path, and improve the operation efficiency and stability of the logistics system. Summary of the Invention

[0005] In order to achieve the above invention purpose, the present invention provides the following technical solutions: A method for planning the path of a logistics robot based on digital twin technology, including the following steps:

[0006] S1. Construct a digital twin model, perform digital mapping on the actual logistics warehousing environment, and collect real-time perception data in the environment;

[0007] S2. According to the digital twin model and the collected data, identify key nodes and obstacle information on the logistics path, and generate a structured path data set;

[0008] S3. Based on the structured path data set, generate candidate path trajectories, and verify the feasibility of the candidate paths in the digital twin simulation environment;

[0009] S4. According to the simulation feedback results, calculate the operation efficiency indicators of each path, and select the optimal path;

[0010] S5. Send the optimal path to the logistics robot, execute path navigation, and continuously optimize the digital twin model by real-time transmitting path deviation data.

[0011] Preferably, in step S1, the construction of the digital twin model includes:

[0012] Deploy multiple types of sensors in the logistics warehousing environment to collect raw physical information, where the sensors include lidar for obtaining the spatial structure contour, RGB-D cameras for capturing cargo images and depth data, and UWB base stations and tags for personnel and robot positioning and tracking;

[0013] Fuse the obtained spatial contour data with the preset warehousing layout data, construct a spatial grid model of the actual warehousing environment through a three-dimensional reconstruction algorithm, and synchronously generate a virtual space coordinate system;

[0014] Synchronously implant various static devices and dynamic entities in the actual warehousing system into the virtual space. The static devices include shelves, loading and unloading ports, and control terminals, and the dynamic entities include robots, manual handlers, and mobile shelves to achieve a full-scale mapping of the physical environment;

[0015] After the above construction is completed, the digital twin system will continuously monitor the data streams reported by various sensors, and through the correction of the difference between the physical environment state and the virtual model state, realize the real-time dynamic update of the twin model, providing an accurate spatial basis for subsequent path analysis.

[0016] Preferably, in step S2, the process of identifying key nodes and obstacles on the logistics path includes:

[0017] In the digital twin environment, based on the warehousing layout and robot task scheduling rules, number each channel in the warehouse, extract the starting and ending nodes at both ends of each channel, and take the nodes at the identified intersection, turning, and corner positions as high-frequency scheduling nodes;

[0018] Introduce the data streams real-time feedback by the sensors into the model. The data streams include traffic density maps, moving entity speed distributions, frequency of occurrence of piled-up objects, etc. For each path node area, define an obstacle scoring function O(x,y) to evaluate its potential blocking degree, specifically as follows:

[0019] O(x,y) = ω1·D(x,y) + ω2·V(x,y) -1 + ω3·F(x,y);

[0020] Where:

[0021] D(x,y) represents the traffic density value at node (x,y);

[0022] V(x,y) represents the average speed of moving entities at the node, with the unit of m / s;

[0023] F(x,y) represents the frequency of occurrence of piled-up objects per unit time at the node;

[0024] ω1, ω2, ω3 are empirical adjustment weights, satisfying ω1 + ω2 + ω3 = 1;

[0025] When O(x, y) exceeds the obstacle score threshold, the node is determined to be a potential obstacle area and set as a low-priority node.

[0026] Combined with ground markings, virtual fence definitions, and dynamic human-machine interaction area division information, further label the nodes in the path with temporary obstacles, soft isolation, and dangerous areas, and add non-passable labels;

[0027] Finally, a structured path dataset is formed, including: path number, connected nodes, estimated passing width, passing frequency statistical value, obstacle prediction level, passing feasibility level fields, which are used to guide subsequent path combinations.

[0028] Preferably, in step S3, the generating candidate paths and simulating and verifying include:

[0029] Screen passable node pairs from the structured path dataset, and construct multiple candidate paths from the starting point to the ending point according to the task start and end points. The path combination logic refers to the node passing weights for dynamic weighted sorting;

[0030] Map each path combination to the digital twin virtual environment and execute the path feasibility verification process: including path turning radius inspection, robot dynamics model constraint simulation, path conflict simulation;

[0031] During the simulation process, according to the path reachability, passing status markings, and energy consumption model feedback at each time step, determine whether the path has the execution conditions. If any section of the path has a non-passable marking during the simulation process, eliminate the path;

[0032] All candidate paths that meet the passing conditions are included in the feasible path set, and each path records its total path segment length, estimated passing time, dynamic obstacle bypassing quantity, and average path smoothness score.

[0033] Preferably, in step S4, the process of screening the optimal path according to the simulation feedback results includes:

[0034] Based on the feasible path set obtained in step S3, analyze the operation indicators of each path in turn. The analysis dimensions include the total path length L, simulation passing time T, energy consumption E required for the path, number of potential conflicts, and path stability score;

[0035] During the comprehensive evaluation process, set the path scoring function R as follows:

[0036]

[0037] Where:

[0038] L is the total path length, in meters;

[0039] Let \(T\) be the path simulation passing time, with the unit of seconds;

[0040] Let \(E\) be the energy consumption required for the robot to simulate the path, with the unit of joules;

[0041] Let \(S\) be the path smoothness and safety score. The value of \(S\) is a real number between 0 and 1. The larger the value, the more stable the path;

[0042] \(\alpha\), \(\beta\), and \(\gamma\) are adjustable weight parameters, satisfying \(\alpha+\beta+\gamma = 1\), and can be set according to the warehousing priority scheduling strategy.

[0043] All candidate paths participate in the scoring calculation. Finally, the path with the highest score is selected as the optimal path, and its path segment information, node order, and task execution priority are recorded for path distribution and execution.

[0044] Preferably, in the step S5, the process of distributing the path to the robot and executing it includes:

[0045] Convert the optimal path determined in step S4 into a path instruction sequence that can be parsed by the robot control module, including a movement instruction MOVE, a turning instruction ROTATE, and a deceleration and docking instruction STOP AT. Each instruction is marked with the path segment, execution speed, and action duration;

[0046] Through a wireless scheduling system connected to the robot communication module, the wireless scheduling system supports the WIFI protocol, send the complete path control sequence to the target robot side, and start the path execution process;

[0047] During the path execution process of the robot, the robot uses the built-in UWB positioning system to obtain its absolute position in the warehouse in real time. According to the current path segment and the expected trajectory, compare the execution position with the deviation range to determine whether fine-tuning, replanning and other operations are needed;

[0048] When the robot deviates from the path by more than the preset deviation threshold, that is, trigger the path deviation feedback mechanism, synchronize the current position, speed, and obstacle change data back to the digital twin model platform, and record the deviation trend data for continuously optimizing the subsequent twin mapping and path generation algorithms.

[0049] Compared with the prior art, the beneficial effects of the present invention are:

[0050] Improve the real-time performance and accuracy of path planning: By integrating multi-source data such as lidar, RGB-D camera, and UWB positioning, a real-time updated digital twin warehousing environment is constructed, which can comprehensively reflect the dynamic changes of the physical space, effectively support the path planning requirements of the robot in high-frequency scheduling scenarios, and significantly improve the accuracy and stability of the traveling path.

[0051] Implement comprehensive intelligent evaluation of path selection: The present invention constructs a multi-parameter weighted function based on four dimensions of path length, travel time, energy consumption, and safety score, and conducts quantitative evaluation in combination with the simulation results to achieve intelligent screening of candidate paths, avoid the limitations of single-factor decision-making, and effectively improve the overall operation efficiency of robot scheduling.

[0052] Support closed-loop dynamic correction of robot path execution: During the path execution process, the robot can locate its own position in real time. If deviation occurs, it triggers a path deviation feedback mechanism and feeds back key parameters to the digital twin system to continuously optimize the subsequent twin model and path generation strategy, realizing the closed-loop control ability from "path planning" to "path adaptive execution". Description of the Drawings

[0053] Figure 1 It is a schematic diagram of the method step flow provided by this application;

[0054] Figure 2 It is a schematic diagram of the flow of Step 1 provided by this application;

[0055] Figure 3 It is a schematic diagram of the flow of Step 2 provided by this application;

[0056] Figure 4 It is a schematic diagram of the flow of Step 3 provided by this application;

[0057] Figure 5 It is a schematic diagram of the flow of Step 4 provided by this application;

[0058] Figure 6 It is a schematic diagram of the flow of Step 5 provided by this application;

[0059] Figure 7 It is a scenario diagram of the path planning of the logistics robot provided by this application. Detailed Implementation Modes

[0060] In order to enable those skilled in the art of the present technology to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0061] Refer to Figures 1-6 , the embodiments of the present invention provide a method for path planning of a logistics robot based on digital twin technology, including the following steps:

[0062] Step 1: Construct a digital twin model, perform digital mapping on the actual logistics warehousing environment, and collect real-time perception data in the environment.

[0063] In Step 1, first deploy a variety of sensor devices in the logistics warehousing environment, including lidar, RGB-D cameras, and UWB base stations. The lidar is used to scan the spatial structure of the warehousing area and output accurate two-dimensional and three-dimensional point cloud data; the RGB-D camera synchronously collects cargo images and depth information; the UWB base stations and tags are respectively fixed in the warehousing area and on moving entities to achieve high-precision real-time positioning and tracking of manual handling personnel and robots.

[0064] The system fuses the acquired lidar point cloud data with the preset warehousing CAD layout map, generates a complete spatial grid model through a three-dimensional reconstruction algorithm, and constructs a digital twin virtual coordinate system. In the virtual space, static devices such as shelves, loading and unloading ports, and terminal equipment are implanted according to the actual deployment information, and at the same time, dynamic entity models such as robots and personnel are added according to the sensor recognition results. All real-time data collected by the sensors is transmitted into the twin system through middleware, and the system dynamically corrects in real time according to the state difference between the physical environment and the virtual model, providing an accurate spatial environment basis for subsequent path analysis.

[0065] Step 2: According to the digital twin model and the collected data, identify the key nodes and obstacle information on the logistics path, and generate a structured path dataset.

[0066] In Step 2, the system automatically numbers and classifies the internal path channels of the warehouse according to the warehousing path planning rules. The intersection points, turning points, and cross points at both ends of the channels are identified as high-frequency scheduling nodes, and these nodes usually form the key relay points of the task path. Combining the task scheduling frequency, the historical running trajectory of the robot, and the warehousing logistics strategy, the system preliminarily screens out the passable path framework.

[0067] Subsequently, the system introduces the data streams uploaded by the sensors in real time, including the channel traffic density heat map, the moving target speed distribution map, and the obstacle accumulation frequency map. Through density threshold analysis and machine learning recognition, the nodes in potential congestion or obstacle areas are determined and set as low-priority or non-passable nodes. At the same time, using the ground marking information and virtual fence settings, the dynamic human-machine interaction areas or dangerous area nodes are further marked, and finally a structured path dataset containing information such as path numbers, passing attributes, and obstacle levels is formed.

[0068] Step 3: Based on the structured path dataset, generate candidate travel paths and verify the feasibility of the candidate paths in the digital twin simulation environment.

[0069] In step three, the system extracts the starting and ending node pairs that meet the conditions from the structured path dataset according to the task requirements, and constructs multiple candidate path combinations based on the connection weights between the nodes. The reference factors for path construction include node passage scores, turn radius reachability, and path conflict historical data, etc. The system establishes a candidate path sequence for each path combination and conducts simulation mapping and path execution drills in the digital twin environment.

[0070] During the simulation process, the system performs simulations on each segment of the path based on the robot kinematic model, including judging whether it meets the corner limit, whether there are obstacle interferences or path conflicts. The simulation is executed in a time-step advancing manner. At each step, the path passage status markers and energy consumption estimates are analyzed. If any path segment does not meet the passable conditions, the path is eliminated. Finally, the system retains all candidate paths that meet the operating conditions, and respectively records the path length, estimated time consumption, path score, and the number of obstacle detours to form a set of feasible paths.

[0071] Step four: Calculate the operation efficiency indicators of each path according to the simulation feedback results, and select the optimal path.

[0072] In step four, based on the set of feasible paths obtained in step S3, the operation indicators of each path are analyzed in turn. The analysis dimensions include the total path length L, the simulation passage time T, the energy consumption E required for the path, the number of potential conflicts, and the path stability score.

[0073] During the comprehensive evaluation process, the path scoring function R is set as follows:

[0074]

[0075] Where:

[0076] L is the total path length, in meters;

[0077] T is the path simulation passage time, in seconds;

[0078] E is the energy consumption required for the robot to simulate the path, in joules;

[0079] S is the path smoothness and safety score. The value of S is a real number between 0 and 1, and the larger the value, the more stable the path;

[0080] α, β, and γ are adjustable weight parameters, satisfying α + β + γ = 1, and can be set according to the warehouse priority scheduling strategy.

[0081] All candidate paths participate in the scoring calculation. Finally, the path with the highest score is selected as the optimal path, and its path segment information, node order, and task execution priority are recorded for path distribution and execution.

[0082] Step 5: Send the optimal path to the logistics robot, execute path navigation, and continuously transmit path deviation data in real time to continuously optimize the digital twin model.

[0083] In Step 5, the system converts the path into a control instruction sequence recognizable by the robot control module segment by segment, including instructions such as MOVE (forward), ROTATE (turn), STOP AT (decelerate and dock), etc. Each instruction clearly specifies the corresponding path segment, execution speed, and duration. The scheduling system transmits the instruction sequence to the target robot through the WIFI communication module to start the path execution process.

[0084] During the execution process, the robot uses its built-in UWB module for real-time positioning to obtain its precise coordinates within the warehousing area and compares them with the expected positions of the path segments. If the deviation does not exceed the threshold, it continues to move forward; if a large deviation occurs, the system will trigger the path deviation feedback mechanism to report the robot's current position, speed, and sudden obstacle data to the digital twin platform in real time. Based on this, the system dynamically corrects the current task or the path of subsequent tasks to achieve closed-loop scheduling and continuous optimization of the travel plan.

[0085] Reference Figure 7 , in specific applications, relying on the digital twin platform to conduct a complete modeling and simulation of the logistics warehousing environment, collecting environmental status data through multi-source sensing devices, and combining path planning algorithms to achieve efficient path generation and execution. The specific process is as follows:

[0086] First, deploy lidar, RGB-D cameras, and UWB positioning systems at the warehousing site to collect spatial structure data, cargo images and depth information, and the position data of personnel and robots in the warehousing environment. Then, fuse the collected data with the pre-set warehousing layout information in the system to construct a virtual space grid model. This model synchronously maps entity objects such as shelves, loading and unloading ports, robots, and manual operators, and realizes dynamic correction of the model through real-time sensing.

[0087] Subsequently, the system numbers the logistics path channels according to the warehousing task allocation logic, marks various intersection nodes, obstacle points, dangerous areas, and temporary isolation areas, and generates a structured path data set. The data fields include information such as path passage scores and obstacle prediction levels. Based on the above data, the system combines multiple paths between the task start point and the end point, simulates and evaluates their passage stability, energy consumption, and execution reliability, and conducts a comprehensive evaluation through the following path scoring function:

[0088]

[0089] Among them, L is the path length, T is the passage time, E is the energy consumption, S is the path stability score, and α, β, γ are weight coefficients.

[0090] Finally, the path with the highest score is selected, disassembled into a standard instruction sequence and sent to the target robot. During the execution process, the robot uses the UWB module for real-time positioning and matches it with the path. When a deviation occurs, the system will immediately trigger a feedback mechanism for model correction and path replanning to ensure that the task is completed on the optimal path.

[0091] It should be noted that, without conflict, the embodiments in the present invention and the features and technical solutions in the embodiments can be combined with each other.

[0092] Obviously, the above-described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. The drawings show preferred embodiments of the present invention, but do not limit the patent scope of the present invention. The present invention can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosed content of the present invention more thorough and comprehensive. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions recorded in the foregoing specific embodiments, or perform equivalent replacements for some of the technical features. Any equivalent structure directly or indirectly using the content of the specification and drawings of the present invention in other related technical fields is equally within the scope of the patent protection of the present invention.

Claims

1. A path planning method for a logistics robot based on digital twin technology, characterized in that, It includes the following steps: S1. Construct a digital twin model to digitally map the actual logistics warehousing environment and collect real-time perception data in the environment; S2. According to the digital twin model and the collected data, identify the key nodes and obstacle information on the logistics path and generate a structured path data set; S3. Based on the structured path data set, generate candidate travel paths and verify the feasibility of the candidate paths in the digital twin simulation environment; S4. According to the simulation feedback results, calculate the operation efficiency indicators of each path and screen out the optimal path; S5. Send the optimal path to the logistics robot to execute path navigation and continuously transmit path deviation data in real time to continuously optimize the digital twin model.

2. The method for planning the travel path of a logistics robot based on digital twin technology according to claim 1, in step S1, the construction of the digital twin model includes: Deploy multiple types of sensors in the logistics warehousing environment to collect original physical information, where the sensors include lidar for obtaining the spatial structure contour, RGB-D cameras for capturing cargo images and depth data, and UWB base stations and tags for realizing the positioning and tracking of personnel and robots; Fuse the obtained spatial contour data with the preset warehousing layout data, construct a spatial grid model of the actual warehousing environment through a 3D reconstruction algorithm, and synchronously generate a virtual space coordinate system; Synchronously implant various static devices and dynamic entities in the actual warehousing system into the virtual space. The static devices include shelves, loading and unloading ports, and control terminals, and the dynamic entities include robots, manual handlers, and mobile shelves to achieve a full-scale mapping of the physical environment; After the above construction is completed, the digital twin system will continuously monitor the data streams reported by various sensors, and through the difference correction between the physical environment state and the virtual model state, realize the real-time dynamic update of the twin model, providing an accurate spatial basis for subsequent path analysis.

3. The method for planning the travel path of a logistics robot based on digital twin technology according to claim 1, in the step S2, the process of identifying the key nodes and obstacles on the logistics path includes: In the digital twin environment, based on the warehousing layout and the robot task scheduling rules, number each passageway in the warehouse, extract the starting and ending nodes at both ends of each passageway, and regard the nodes at the identified intersection, turning, and corner positions as high-frequency scheduling nodes; Introduce the data streams real-time feedback by the sensors into the model. The data streams include traffic density maps, moving entity speed distributions, debris occurrence frequencies, etc. For each path node area, define an obstacle scoring function O(x,y) to evaluate its potential blocking degree, specifically as follows: O(x,y) = ω1·D(x,y) + ω2·V(x,y) -1 + ω3·F(x,y); Where: D(x,y) represents the traffic density value at the node (x,y); V(x,y) represents the average speed of the moving entity at the node, with the unit of m / s; F(x,y) represents the frequency of debris occurrence per unit time at the node; ω1, ω2, ω3 are empirical adjustment weights, satisfying ω1 + ω2 + ω3 = 1; When O(x,y) exceeds the obstacle scoring threshold, the node is determined as a potential obstacle area and set as a low-priority node. Combined with ground marking, virtual fence definition, and dynamic human-machine interaction area division information, further label the nodes in the path where there are temporary obstacles, soft isolation, and dangerous areas, and add non-passable labels; Finally, a structured path dataset is formed, including: path number, connected nodes, estimated passing width, passing frequency statistical value, obstacle prediction level, passing feasibility level fields, which are used to guide subsequent path combination.

4. According to the method for path planning of a logistics robot based on digital twin technology described in claim 1, in step S3, the generation of candidate paths and simulation verification include: Screen passable node pairs from the structured path dataset, and based on the start and end points of the task, construct multiple candidate paths from the start point to the end point. The path combination logic refers to the passing weights of the nodes for dynamic weighted sorting; Map each path combination to the digital twin virtual environment and execute the path feasibility verification process: including path turning radius inspection, robot dynamics model constraint simulation, and path conflict simulation; During the simulation process, based on the path reachability, passing status marking, and energy consumption model feedback at each time step, determine whether the path meets the execution conditions. If any section of the path has a non-passable mark during the simulation process, then remove this path; All candidate paths that meet the passing conditions are included in the set of feasible paths, and each path records its total path segment length, estimated passing time, number of detours around dynamic obstacles, and average path smoothness score.

5. According to the method for path planning of a logistics robot based on digital twin technology described in claim 1, in step S4, the process of screening the optimal path according to the simulation feedback results includes: Based on the set of feasible paths obtained in step S3, analyze the operation indicators of each path in turn. The analysis dimensions include the total path length L, simulation passing time T, energy consumption E required for the path, number of potential conflicts, and path stability score; During the comprehensive evaluation process, set the path scoring function R as follows: Where: L is the total path length, in meters; T is the path simulation passing time, in seconds; E is the energy consumption required for the robot to simulate the path, in joules; S is the path smoothness and safety score, and the value is a real number between 0 and 1. The larger the value, the more stable the path; α, β, γ are adjustable weight parameters, satisfying α + β + γ = 1, and can be set according to the warehouse priority scheduling strategy. All candidate paths participate in the scoring calculation. Finally, select the path with the highest score as the optimal path, and record its path segment information, node order, and task execution priority for path distribution and execution.

6. According to the method for path planning of a logistics robot based on digital twin technology described in claim 1, in step S5, the process of distributing the path to the robot and executing it includes: Convert the optimal path determined in step S4 into a path instruction sequence that can be parsed by the robot control module, including move instruction MOVE, turn instruction ROTATE, decelerate and stop instruction STOP AT, and each instruction marks the path segment, execution speed, and action duration; Through a wireless scheduling system connected to the robot communication module, the wireless scheduling system supports the WIFI protocol, sends the complete path control sequence to the target robot side, and starts the path execution process; During the path execution process, the robot obtains its absolute position in the warehouse in real time through the built-in UWB positioning system, and based on the current path segment and the expected trajectory, compares the execution position with the deviation range to determine whether operations such as fine-tuning and replanning are required; When the robot deviates from the path by more than the preset offset threshold, the path deviation feedback mechanism is triggered, and the current position, speed, and obstacle change data are synchronized back to the digital twin model platform, and the deviation trend data is recorded for continuously optimizing the subsequent twin mapping and path generation algorithms.

Images