Intelligent unmanned forklift system and method based on visual navigation

Abstract

The invention relates to the technical field of intelligent unmanned forklifts, in particular to an intelligent unmanned forklift system and method based on visual navigation, and the system comprises a forklift main body module, a binocular vision module, an image preprocessing unit, a binocular distance measurement module, a track prediction module, a pedestrian area detection module, a pedestrian area center calculation module and an obstacle avoidance module. The image preprocessing unit is connected with the binocular vision module, the binocular distance measurement module is connected with the image preprocessing unit, the track prediction module is connected with the binocular distance measurement module, and the pedestrian rectangular area detection module is connected with the track prediction module. The technical problems that in the complex environment, under the conditions that light changes, shelters exist and the like, the performance is limited, the positioning precision is reduced, and even safety accidents are possibly caused are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent unmanned forklifts, and particularly to an intelligent unmanned forklift system and method based on visual navigation. Background Art

[0002] With the booming development of the logistics and warehousing industry, automation and intelligence have become important forces driving the transformation and upgrading of this industry. Intelligent unmanned forklifts, as a key part of the automated logistics solution, are increasingly widely used, bringing unprecedented efficiency improvements to warehousing operations. These forklifts can autonomously complete tasks such as cargo handling and stacking without human control, greatly reducing the labor burden and improving the operation efficiency. During the use of intelligent unmanned forklifts, the navigation system of intelligent unmanned forklifts is crucial, and the navigation system mainly relies on sensors such as lidar and ultrasonic sensors.

[0003] However, when using a single sensor such as lidar and ultrasonic sensors for navigation, in complex environments, such as when there are light changes and obstacles, the performance will be limited, resulting in a decrease in positioning accuracy and even potentially causing safety accidents. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent unmanned forklift system and method based on visual navigation, aiming to solve the technical problem that when the existing intelligent unmanned forklift system uses a single sensor for navigation, in complex environments, such as when there are light changes and obstacles, the performance will be limited, resulting in a decrease in positioning accuracy and even potentially causing safety accidents.

[0005] To achieve the above purpose, an intelligent unmanned forklift system based on visual navigation adopted by the present invention includes a forklift body module, a binocular vision module, an image preprocessing unit, a binocular ranging module, a trajectory prediction module, a pedestrian area detection module, a pedestrian area center calculation module, and an obstacle avoidance module. The binocular vision module is installed on the forklift body module. The image preprocessing unit is connected to the binocular vision module. The binocular ranging module is connected to the image preprocessing unit. The trajectory prediction module is connected to the binocular ranging module. The pedestrian rectangular area detection module is connected to the trajectory prediction module. The pedestrian center calculation module is connected to the pedestrian rectangular area detection module. The obstacle avoidance module is connected to the pedestrian area center calculation module;

[0006] The forklift body module serves as a carrier for performing physical tasks such as handling and stacking;

[0007] The binocular vision module is used to provide stereoscopic vision perception and provide basic data for subsequent image processing and ranging;

[0008] The image preprocessing unit receives the image information captured by the binocular vision module and performs preprocessing to improve the image quality;

[0009] The binocular ranging module uses the binocular vision principle and combines the preprocessed image information to calculate the exact distance of the objects in the environment;

[0010] The trajectory prediction module predicts the motion trajectories of the objects in the environment based on the distance information calculated by the binocular ranging module, providing data support for the obstacle avoidance decision-making of the forklift main body module;

[0011] The pedestrian area detection module is used to detect the pedestrian area in the image, mark it as a rectangular box, and at the same time transmit it to the pedestrian area center calculation module to calculate the center point position of the pedestrian area;

[0012] The obstacle avoidance module receives the pedestrian center point position information, determines whether the forklift needs to avoid pedestrians, and formulates corresponding obstacle avoidance strategies.

[0013] Among them, the image preprocessing unit includes a data collection module, a denoising module, a contrast enhancement module, an edge detection module, and an image output module. The data collection module is connected to the binocular vision module. The denoising module is connected to both the data collection module and the contrast enhancement module. The edge detection module is connected to the contrast enhancement module. The image output module is connected to the edge detection module.

[0014] Among them, the intelligent unmanned forklift system based on visual navigation further includes an environment perception module, which is also installed on the forklift main body module;

[0015] The environment perception module uses environment perception sensors, such as lidar or ultrasonic sensors, to work in cooperation with the binocular vision module to further improve the perception ability of complex environments.

[0016] Among them, the intelligent unmanned forklift system based on visual navigation further includes a path planning module, and the path planning module is connected to the obstacle avoidance module.

[0017] Among them, the pedestrian area center calculation module is a Yolov8 pedestrian detection algorithm improved based on GhostNet, and at the same time combines object detection and segmentation methods to calculate the center of the pedestrian area.

[0018] GhostNet is a lightweight neural network structure. By introducing lightweight convolution operations, it reduces the computational complexity of the model while maintaining the detection accuracy. GhostNet is integrated into the Yolov8 model and used for pedestrian detection and segmentation.

[0019] Subsequently, by precisely segmenting the detected pedestrian area, the central position of the pedestrian is calculated; this can improve the accuracy and real-time performance of dynamic obstacle avoidance, enabling the system to better understand the position and behavior of pedestrians in the environment, and thus make more reasonable obstacle avoidance decisions.

[0020] Among them, the intelligent unmanned forklift system based on visual navigation further includes an interaction module and an operation terminal. The interaction module is connected to the obstacle avoidance module, and the operation terminal is connected to the interaction module.

[0021] Among them, the intelligent unmanned forklift system based on visual navigation further includes an adaptive optimization module, and the adaptive optimization module is connected to the path planning module.

[0022] The adaptive optimization module adopts an improved adaptive genetic algorithm (AGA), which is specifically used for path planning and scheduling of unmanned forklifts in complex environments. Through multiple technological innovation points, it solves the limitations of traditional genetic algorithms in practical applications, enabling it to operate efficiently in a dynamic environment with multiple objectives and multiple constraints.

[0023] Among them, the intelligent unmanned forklift system based on visual navigation further includes a safety monitoring module, and the safety monitoring module is also installed on the forklift main body module;

[0024] The monitoring module is used to monitor the operating status of the forklift and the surrounding environment in real time. Once an abnormal situation is detected, it immediately issues an alarm and takes corresponding safety measures.

[0025] Among them, the intelligent unmanned forklift system based on visual navigation further includes a cooperative operation module, and the cooperative operation module is connected to the forklift main body module.

[0026] The present invention also provides a method for using an intelligent unmanned forklift based on visual navigation, which is applied to the intelligent unmanned forklift system based on visual navigation as described above.

[0027] It includes the following steps:

[0028] Using the binocular vision module installed on the forklift main body module, perform stereo vision perception on the surrounding environment to capture image information;

[0029] The image preprocessing unit receives the image information captured by the binocular vision module and performs preprocessing to improve the image quality;

[0030] The binocular ranging module uses the binocular vision principle and combines the preprocessed image information to calculate the precise distance of objects in the environment;

[0031] The trajectory prediction module predicts the movement trajectories of objects in the environment based on the distance information calculated by the binocular ranging module. Meanwhile, the pedestrian area detection module detects the pedestrian area in the image, marks it as a rectangular box, and transmits it to the pedestrian area center calculation module to calculate the center point position of the pedestrian area;

[0032] The obstacle avoidance module receives the pedestrian center point position information, determines whether the forklift needs to avoid pedestrians, formulates corresponding obstacle avoidance strategies, and controls the forklift main body module to perform obstacle avoidance operations.

[0033] In the specific use of an intelligent unmanned forklift system and method based on visual navigation according to the present invention, the present invention first uses the binocular vision module installed on the forklift main body module to perform stereo vision perception on the surrounding environment and capture image information; the image preprocessing unit receives the image information captured by the binocular vision module and performs preprocessing to improve the image quality; the binocular ranging module uses the binocular vision principle and combines the preprocessed image information to calculate the accurate distances of objects in the environment; the trajectory prediction module predicts the movement trajectories of objects in the environment based on the distance information calculated by the binocular ranging module. Meanwhile, the pedestrian area detection module detects the pedestrian area in the image, marks it as a rectangular box, and transmits it to the pedestrian area center calculation module to calculate the center point position of the pedestrian area; the obstacle avoidance module receives the pedestrian center point position information, determines whether the forklift needs to avoid pedestrians, formulates corresponding obstacle avoidance strategies, and controls the forklift main body module to perform obstacle avoidance operations. In this way, the technical problem in the prior art that when an intelligent unmanned forklift system uses a single sensor for navigation, its performance will be limited in complex environments such as light changes and the presence of obstacles, resulting in a decrease in positioning accuracy and even potential safety accidents can be solved.

[0034] By introducing the binocular vision module and combining the image preprocessing unit and the binocular ranging module, the present invention can achieve stereo perception of the surrounding environment. Compared with a single sensor, the binocular vision system can provide richer depth information and more accurate distance measurement, effectively cope with complex environments such as light changes and the presence of obstacles, and significantly improve the navigation accuracy and stability of the forklift in complex scenarios.

[0035] The addition of the pedestrian area detection module and the pedestrian area center calculation module enables the system to detect and track the pedestrian position in real time and accurately calculate the center position of the pedestrian area. This function is crucial for ensuring pedestrian safety, especially in scenarios with dense crowds or dynamic changes. Combined with the obstacle avoidance module, the system can plan the obstacle avoidance path in advance, effectively avoid potential collisions with pedestrians, and greatly improve safety.

[0036] The trajectory prediction module provides the forklift with a farther field of view and a more accurate decision-making basis by predicting the movement trajectories of pedestrians and obstacles. This helps the forklift make more reasonable path planning and speed control in complex environments, reduce sudden stops or detours caused by unexpected situations, and improve operation efficiency.

[0037] Through the collaborative work of multiple modules, adaptive adjustment to environmental changes is achieved. Whether it is the intensity of light, the occlusion situation, or the uncertainty of pedestrian behavior, the system can maintain a high navigation accuracy and obstacle avoidance ability through the mutual cooperation of each module, enhancing the robustness and adaptability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0039] Figure 1 is the principle block diagram of the first embodiment of the present invention.

[0040] Figure 2 is the principle block diagram of the second embodiment of the present invention.

[0041] 101 - Forklift main body module, 102 - Binocular vision module, 103 - Image preprocessing unit, 104 - Binocular ranging module, 105 - Trajectory prediction module, 106 - Pedestrian area detection module, 107 - Pedestrian area center calculation module, 108 - Obstacle avoidance module, 109 - Environment perception module, 110 - Path planning module, 111 - Data collection module, 112 - Denoising module, 113 - Contrast enhancement module, 114 - Edge detection module, 115 - Image output module, 201 - Interaction module, 202 - Operation end, 203 - Adaptive optimization module, 204 - Safety monitoring module, 205 - Collaborative operation module, 206 - Energy management module, 207 - Login verification module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The following will describe in detail the embodiments of the present invention. The examples of the embodiments are shown in the drawings. The embodiments described below with reference to the drawings are exemplary and are intended to explain the present invention, and should not be construed as a limitation to the present invention.

[0043] The first embodiment of the present application is as follows:

[0044] Please refer to Figure 1 , Figure 1 is the principle block diagram of the first embodiment of the present invention.

[0045] The present invention provides an intelligent unmanned forklift system based on visual navigation, which includes a forklift body module 101, a binocular vision module 102, an image preprocessing unit 103, a binocular ranging module 104, a trajectory prediction module 105, a pedestrian area detection module 106, a pedestrian area center calculation module 107, an obstacle avoidance module 108, an environment perception module 109 and a path planning module 110. The image preprocessing unit 103 includes a data collection module 111, a denoising module 112, a contrast enhancement module 113, an edge detection module 114 and an image output module 115. The foregoing solution solves the technical problem that when the intelligent unmanned forklift system in the prior art uses a single sensor for navigation, its performance will be limited in complex environments, such as when there are light changes and obstacles, resulting in a decrease in positioning accuracy and even possible safety accidents.

[0046] For this specific embodiment, the forklift body module 101 serves as a carrier for performing physical tasks such as handling and stacking.

[0047] The binocular vision module 102 is used to provide stereoscopic vision perception and provide basic data for subsequent image processing and ranging.

[0048] The image preprocessing unit 103 receives the image information captured by the binocular vision module 102 and performs preprocessing to improve the image quality.

[0049] The binocular ranging module 104 uses the binocular vision principle and combines the preprocessed image information to calculate the exact distance of the objects in the environment.

[0050] The trajectory prediction module 105 predicts the movement trajectories of the objects in the environment based on the distance information calculated by the binocular ranging module 104, and provides data support for the obstacle avoidance decision of the forklift body module 101.

[0051] The pedestrian area detection module 106 is used to detect the pedestrian area in the image, mark it as a rectangular frame, and at the same time transmit it to the pedestrian area center calculation module 107 to calculate the center point position of the pedestrian area.

[0052] The obstacle avoidance module 108 receives the pedestrian center point position information, determines whether the forklift needs to avoid pedestrians, and formulates corresponding obstacle avoidance strategies.

[0053] Among them, the binocular vision module 102 is installed on the forklift main body module 101. The image preprocessing unit 103 is connected to the binocular vision module 102. The binocular ranging module 104 is connected to the image preprocessing unit 103. The trajectory prediction module 105 is connected to the binocular ranging module 104. The pedestrian rectangular area detection module is connected to the trajectory prediction module 105. The pedestrian center calculation module is connected to the pedestrian rectangular area detection module. The obstacle avoidance module 108 is connected to the pedestrian area center calculation module 107. In specific use, the present invention first uses the binocular vision module 102 installed on the forklift main body module 101 to perform stereo vision perception on the surrounding environment and capture image information. The image preprocessing unit 103 receives the image information captured by the binocular vision module 102 and performs preprocessing to improve the image quality. The binocular ranging module 104 uses the binocular vision principle and combines the preprocessed image information to calculate the precise distance of the objects in the environment. The trajectory prediction module 105 predicts the movement trajectory of the objects in the environment based on the distance information calculated by the binocular ranging module 104. At the same time, the pedestrian area detection module 106 detects the pedestrian area in the image, marks it as a rectangular frame, and transmits it to the pedestrian area center calculation module 107 to calculate the center point position of the pedestrian area. The obstacle avoidance module 108 receives the pedestrian center point position information, determines whether the forklift needs to avoid pedestrians, formulates corresponding obstacle avoidance strategies, and controls the forklift main body module 101 to perform obstacle avoidance operations. In this way, it solves the technical problem that when the intelligent unmanned forklift system in the prior art uses a single sensor for navigation, its performance will be limited in complex environments, such as in the case of light changes and the presence of obstacles, resulting in a decrease in positioning accuracy and even possible safety accidents.

[0054] Secondly, the data collection module 111 is connected to the binocular vision module 102. The denoising module 112 is connected to both the data collection module 111 and the contrast enhancement module 113. The edge detection module 114 is connected to the contrast enhancement module 113. The image output module 115 is connected to the edge detection module 114.

[0055] The data collection module 111, the denoising module 112, the contrast enhancement module 113, the edge detection module 114, and the image output module 115 inside the image preprocessing unit 103 form a tightly connected processing chain. Through the integration of efficient algorithms among the modules, rapid transfer and precise processing of image data are achieved. The denoising algorithm effectively filters out image noise, the contrast enhancement algorithm significantly improves the visualization of image details, and the edge detection algorithm accurately captures the key edge information in the image. The efficient integration of these algorithms not only improves the real-time performance of image processing but also ensures the accuracy of the processing results.

[0056] In the present invention, the denoising module 112, the contrast enhancement module 113, and the edge detection module 114 all adopt adaptive algorithm designs. The denoising algorithm can be intelligently adjusted according to the characteristics of image noise to effectively remove noise without losing image details. The contrast enhancement algorithm can adaptively adjust images under different lighting conditions to ensure that clear details can be presented in various environments. The edge detection algorithm also has adaptability and can accurately identify the edge features in the image, maintaining a high detection accuracy even in complex or blurred images.

[0057] Meanwhile, the environment perception module 109 is also installed on the forklift main body module 101;

[0058] The environment perception module 109 uses environment perception sensors, such as lidar or ultrasonic sensors, to work in cooperation with the binocular vision module 102 to further improve the perception ability of complex environments.

[0059] In addition, the path planning module 110 is connected to the obstacle avoidance module 108. The path planning module 110 is responsible for planning an optimal or sub-optimal driving path for the forklift according to the current environment information and task requirements by combining the obstacle avoidance routes of the obstacle avoidance module 108.

[0060] The pedestrian area center calculation module 107 is based on the improved Yolov8 pedestrian detection algorithm of GhostNet and combines object detection and segmentation methods to calculate the center of the pedestrian area.

[0061] GhostNet is a lightweight neural network structure. By introducing lightweight convolution operations, it reduces the computational complexity of the model while maintaining the detection accuracy. GhostNet is integrated into the Yolov8 model and used for pedestrian detection and segmentation.

[0062] Subsequently, by precisely segmenting the detected pedestrian area, the central position of the pedestrian is calculated; this can improve the accuracy and real-time performance of dynamic obstacle avoidance, enabling the system to better understand the position and behavior of pedestrians in the environment, and thus making more reasonable obstacle avoidance decisions.

[0063] When using an intelligent unmanned forklift system based on visual navigation according to this embodiment, in specific use, the present invention first uses the binocular vision module 102 installed on the forklift main body module 101 to perform stereoscopic vision perception on the surrounding environment and capture image information; the image preprocessing unit 103 receives the image information captured by the binocular vision module 102 and performs preprocessing to improve the image quality; the binocular ranging module 104 uses the binocular vision principle and combines the preprocessed image information to calculate the precise distance of objects in the environment; the trajectory prediction module 105 predicts the movement trajectory of objects in the environment based on the distance information calculated by the binocular ranging module 104. At the same time, the pedestrian area detection module 106 detects the pedestrian area in the image, marks it as a rectangular frame, and transmits it to the pedestrian area center calculation module 107 to calculate the center point position of the pedestrian area; the obstacle avoidance module 108 receives the pedestrian center point position information, determines whether the forklift needs to avoid pedestrians, and formulates corresponding obstacle avoidance strategies to control the forklift main body module 101 to perform obstacle avoidance operations. In this way, it solves the technical problem that when the intelligent unmanned forklift system in the prior art uses a single sensor for navigation, in complex environments such as light changes and the presence of obstacles, its performance will be limited, resulting in a decrease in positioning accuracy and even possible safety accidents.

[0064] The second embodiment of this application is as follows:

[0065] Based on the first embodiment, please refer to Figure 2 , Figure 2 which is the principle block diagram of the second embodiment of the present invention.

[0066] The present invention provides an intelligent unmanned forklift system based on visual navigation, which further includes an interaction module 201, an operation terminal 202, an adaptive optimization module 203, a safety monitoring module 204, a collaborative operation module 205, an energy management module 206, and a login verification module 207.

[0067] For this specific embodiment, the interaction module 201 is connected to the obstacle avoidance module 108, the operation terminal 202 is connected to the interaction module 201, and the operation terminal 202 is for management personnel to use. Through the interaction module 201, the obstacle avoidance module 108 can be managed.

[0068] Among them, the adaptive optimization module 203 is connected to the path planning module 110.

[0069] The adaptive optimization module 203 adopts an improved adaptive genetic algorithm (AGA), which is specifically used for path planning and scheduling of driverless forklifts in complex environments. Through multiple technological innovation points, it solves the limitations of traditional genetic algorithms in practical applications, enabling it to operate efficiently in a dynamic environment with multiple objectives and multiple constraints.

[0070] Secondly, the safety monitoring module 204 is also installed on the forklift body module 101;

[0071] The monitoring module is used to monitor the operating status of the forklift and the surrounding environment in real time. Once an abnormal situation is detected, it immediately issues an alarm and takes corresponding safety measures.

[0072] Thirdly, the collaborative operation module 205 is connected to the forklift body module 101. The collaborative control module is used to connect with other forklifts to achieve collaborative operation, further improving obstacle avoidance and path planning.

[0073] Finally, the energy management module 206 is connected to the forklift body module 101. The verification and login module is implanted in the operation terminal 202. By setting the energy management module 206, the energy consumption of the forklift body module 101 can be managed, and by the verification and login module, the identity of the management personnel logging in to the operation terminal 202 can be verified.

[0074] When using the intelligent driverless forklift system based on visual navigation of this embodiment, the operation terminal 202 is for management personnel to use. Through the interaction module 201, the obstacle avoidance module 108 can be managed. When specifically used, the adaptive optimization module 203 adopts an improved adaptive genetic algorithm (AGA), which is specifically used for path planning and scheduling of driverless forklifts in complex environments. Through multiple technological innovation points, it solves the limitations of traditional genetic algorithms in practical applications, enabling it to operate efficiently in a dynamic environment with multiple objectives and multiple constraints. By setting the energy management module 206, the energy consumption of the forklift body module 101 can be managed, and by the verification and login module, the identity of the management personnel logging in to the operation terminal 202 can be verified.

[0075] The present invention also provides a method for using an intelligent driverless forklift based on visual navigation, which is applied to the intelligent driverless forklift system as described above.

[0076] It includes the following steps:

[0077] Utilize the binocular vision module 102 installed on the forklift body module 101 to perform stereo vision perception on the surrounding environment and capture image information;

[0078] The image preprocessing unit 103 receives the image information captured by the binocular vision module 102 and performs preprocessing to improve the image quality;

[0079] The binocular ranging module 104 uses the binocular vision principle and combines the preprocessed image information to calculate the precise distance of the objects in the environment;

[0080] The trajectory prediction module 105 predicts the motion trajectories of the objects in the environment based on the distance information calculated by the binocular ranging module 104. At the same time, the pedestrian area detection module 106 detects the pedestrian area in the image, marks it as a rectangular box, and transmits it to the pedestrian area center calculation module 107 to calculate the center point position of the pedestrian area;

[0081] The obstacle avoidance module 108 receives the pedestrian center point position information, determines whether the forklift needs to avoid pedestrians, formulates corresponding obstacle avoidance strategies, and controls the forklift main body module 101 to perform obstacle avoidance operations.

[0082] By introducing the binocular vision module 102 and combining the image preprocessing unit 103 and the binocular ranging module 104, the present invention can achieve three-dimensional perception of the surrounding environment. Compared with a single sensor, the binocular vision system can provide richer depth information and more accurate distance measurement, effectively cope with complex environments such as light changes and the presence of obstacles, and significantly improve the navigation accuracy and stability of the forklift in complex scenarios.

[0083] The addition of the pedestrian area detection module 106 and the pedestrian area center calculation module 107 enables the system to detect and track the pedestrian position in real time and accurately calculate the center position of the pedestrian area. This function is crucial for ensuring pedestrian safety, especially in crowded or dynamically changing scenarios. Combined with the obstacle avoidance module 108, the system can plan the obstacle avoidance path in advance, effectively avoid potential collisions with pedestrians, and greatly improve safety.

[0084] The trajectory prediction module 105 provides the forklift with a farther field of view and more accurate decision-making basis by predicting the motion trajectories of pedestrians and obstacles. This helps the forklift make more reasonable path planning and speed control in complex environments, reduce sudden stops or detours caused by unexpected situations, and improve operation efficiency.

[0085] Through the collaborative work of multiple modules, adaptive adjustment to environmental changes is achieved. Whether it is the light intensity, occlusion situation, or the uncertainty of pedestrian behavior, the system can maintain a high navigation accuracy and obstacle avoidance ability through the mutual cooperation of each module, enhancing the robustness and adaptability of the system.

[0086] The above-disclosed is only a preferred embodiment of the present invention. Of course, it cannot be used to limit the scope of the rights of the present invention. Those of ordinary skill in the art can understand all or part of the processes of implementing the above embodiments, and the equivalent changes made according to the claims of the present invention still fall within the scope covered by the invention.

Claims

1. An intelligent unmanned forklift system based on visual navigation, characterized in that: It includes a forklift body module, a binocular vision module, an image preprocessing unit, a binocular ranging module, a trajectory prediction module, a pedestrian area detection module, a pedestrian area center calculation module and an obstacle avoidance module, wherein the binocular vision module is installed on the forklift body module, the image preprocessing unit is connected to the binocular vision module, the binocular ranging module is connected to the image preprocessing unit, the trajectory prediction module is connected to the binocular ranging module, the pedestrian rectangular area detection module is connected to the trajectory prediction module, the pedestrian center calculation module is connected to the pedestrian rectangular area detection module, and the obstacle avoidance module is connected to the pedestrian area center calculation module; The forklift body module is used as a carrier to perform physical tasks such as handling and stacking; The binocular vision module is used to provide stereoscopic visual perception and provide basic data for subsequent image processing and ranging; The image preprocessing unit receives the image information captured by the binocular vision module and performs preprocessing to improve the image quality; The binocular distance measurement module uses the binocular vision principle and combines the pre-processed image information to calculate the precise distance of objects in the environment; The trajectory prediction module predicts the motion trajectory of objects in the environment based on the distance information calculated by the binocular ranging module, and provides data support for the obstacle avoidance decision of the forklift main body module; The pedestrian area detection module is used to detect the pedestrian area in the image, mark it as a rectangular frame, and transmit it to the pedestrian area center calculation module to calculate the center point position of the pedestrian area; The obstacle avoidance module receives the pedestrian center point position information, determines whether the forklift needs to avoid the pedestrian, and formulates a corresponding obstacle avoidance strategy.

2. The intelligent unmanned forklift system based on visual navigation as claimed in claim 1, characterized in that: The image preprocessing unit includes a data collection module, a denoising module, a contrast enhancement module, an edge detection module and an image output module. The data collection module is connected to the binocular vision module, the denoising module is connected to the data collection module and the contrast enhancement module, the edge detection module is connected to the contrast enhancement module, and the image output module is connected to the edge detection module.

3. The intelligent unmanned forklift system based on visual navigation as claimed in claim 2, characterized in that: The intelligent unmanned forklift system based on visual navigation also includes an environment perception module, which is also installed on the forklift main body module; The environment perception module uses an environment perception sensor, such as a laser radar or an ultrasonic sensor, which works in conjunction with the binocular vision module to further improve the perception capability of complex environments.

4. The intelligent unmanned forklift system based on visual navigation as claimed in claim 3, characterized in that: The intelligent unmanned forklift system based on visual navigation also includes a path planning module, and the path planning module is connected to the obstacle avoidance module.

5. The intelligent unmanned forklift system based on visual navigation as claimed in claim 4, characterized in that: The pedestrian region center calculation module is based on the Yolov8 pedestrian detection algorithm improved by GhostNet, and combines target detection and segmentation methods to calculate the pedestrian region center.

6. The intelligent unmanned forklift system based on visual navigation as claimed in claim 5, characterized in that: The intelligent unmanned forklift system based on visual navigation also includes an interaction module and an operation terminal, the interaction module is connected to the obstacle avoidance module, and the operation terminal is connected to the interaction module.

7. The intelligent unmanned forklift system based on visual navigation as claimed in claim 6, characterized in that: The intelligent unmanned forklift system based on visual navigation also includes an adaptive optimization module, which is connected to the path planning module.

8. The intelligent unmanned forklift system based on visual navigation as claimed in claim 7, characterized in that: The visual navigation-based intelligent unmanned forklift system further includes a safety monitoring module, which is also installed on the forklift main body module; The monitoring module is used to monitor the operating status and surrounding environment of the forklift in real time. Once an abnormal situation is detected, an alarm is immediately issued and corresponding safety measures are taken.

9. The intelligent unmanned forklift system based on visual navigation as claimed in claim 8, characterized in that: The intelligent unmanned forklift system based on visual navigation also includes a collaborative operation module, and the collaborative operation module is connected to the forklift main body module.

10. A method for using an intelligent unmanned forklift based on vision navigation, applied to the intelligent unmanned forklift system based on vision navigation as claimed in claim 9, characterized in that: The steps include: Using the binocular vision module installed on the forklift body module, the surrounding environment is perceived in stereoscopic vision to capture image information; The image preprocessing unit receives the image information captured by the binocular vision module and performs preprocessing to improve the image quality; The binocular distance measurement module uses the binocular vision principle and combines the pre-processed image information to calculate the precise distance of objects in the environment; The trajectory prediction module predicts the motion trajectory of objects in the environment based on the distance information calculated by the binocular ranging module. At the same time, the pedestrian area detection module detects the pedestrian area in the image and marks it as a rectangular frame, and transmits it to the pedestrian area center calculation module to calculate the center point position of the pedestrian area. The obstacle avoidance module receives the pedestrian center point position information, determines whether the forklift needs to avoid the pedestrian, formulates a corresponding obstacle avoidance strategy, and controls the forklift main body module to perform the obstacle avoidance operation.

Images