Automatic loading and unloading device for road freight
By using multi-sensor fusion perception and adaptive path planning algorithms, the problem of insufficient autonomy of infrared forklifts in complex environments has been solved, realizing efficient and safe operation of automated loading and unloading of road freight, and possessing self-learning and fault-tolerant capabilities.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-15
- Publication Date
- 2026-03-27
AI Technical Summary
Existing infrared forklifts lack sufficient automation in road freight transportation, pose safety hazards, and have poor environmental adaptability, making it difficult to achieve autonomous operation in complex environments.
By employing multi-sensor fusion sensing technology, combined with adaptive path planning algorithms and dynamic attitude closed-loop control, high-precision real-time modeling of the operating environment and cargo status is achieved. Furthermore, the system's autonomy and safety are enhanced through adaptive learning modules and fault diagnosis mechanisms.
It enables efficient, safe, and autonomous loading and unloading operations in complex environments, improving the success rate and equipment reliability. It has self-learning and fault-tolerance capabilities, ensuring efficient operation of unattended work.
Smart Images
Figure CN121735176A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of road freight, in particular to a road freight automatic loading and unloading device. BACKGROUND
[0002] Road freight refers to a logistics mode that uses freight vehicles (such as trucks, vans, etc.) to transport goods through a highway network. It is a crucial link in the modern logistics system and is widely used in the circulation of raw materials, semi-finished products, finished products, and daily necessities. It connects production enterprises, warehouse centers, distribution points, and end consumers, and supports the operation of regional economies and even global trade. Road freight needs to be loaded and unloaded because the transportation process of goods from the starting point to the destination must go through two key steps: loading and unloading. Loading usually occurs at the origin of the goods, such as a factory, warehouse, or logistics distribution center. Goods are safely and reasonably loaded onto the transport vehicle by manual, mechanical (such as forklifts, cranes), or automated equipment. This process needs to ensure that the goods are fixed and stable, and the distribution is balanced to avoid damage to the goods or accidents caused by vehicle imbalance during transportation due to bumps and collisions. At the same time, appropriate protective measures should be taken according to the nature of the goods (such as fragile, flammable, refrigerated, etc.). Unloading is the process of removing goods from the vehicle and delivering them to the consignee after the goods arrive at the destination. During unloading, the goods information needs to be checked, the appearance integrity needs to be inspected, and the goods need to be properly moved to the designated location, such as a warehouse, a market, or a production line, thereby completing the closed loop of the entire transportation chain.
[0003] In the process of loading and unloading road freight, the traditional method usually relies on manual operation, which not only requires a large amount of labor, resulting in high labor costs, but also easily causes low efficiency and damage to goods due to fatigue, operational errors, or physical limitations. At the same time, automated technologies such as infrared forklifts have been introduced to optimize this process. It realizes partial automation functions such as goods recognition and path tracking through infrared sensing, which can reduce the dependence on manpower and improve the operation speed. However, the existing infrared automated forklifts still have obvious limitations. The automation degree has not reached the level of complete autonomy. For example, in complex environments (such as changes in light, obstacles interference, or irregular shapes of goods), the perception and decision-making ability is insufficient, and manual intervention or supervision is still needed, which limits the further improvement of efficiency. More importantly, safety problems are more prominent. Infrared technology may misjudge due to environmental factors (such as strong light or reflection), resulting in collisions, goods falling, or equipment failure. In addition, the system integration is not high and the emergency mechanism is not perfect, which cannot completely avoid the risks in operation, thereby posing potential threats to goods, equipment, and personnel safety. SUMMARY
[0004] The road freight automatic loading and unloading device solves the problems of insufficient automation and potential safety hazards of existing infrared forklifts.
[0005] To achieve the above object, the application is implemented by the following technical scheme: a road freight automatic loading and unloading device, comprising a forklift host, a control panel is arranged at the front end of the forklift host, a fixed frame body is fixedly arranged at the rear end of the forklift host, a movable fork frame is arranged between the fixed frame body and the forklift host through a chain wheel and chain lifting mechanism, and a multi-sensor fusion mechanism is fixed to the upper end of the fixed frame body.
[0006] Preferably, an electric wheel is arranged at each of the four corners of the lower end surface of the forklift host.
[0007] A road freight automatic loading and unloading control method, the method is based on the environment and cargo data collected by the multi-sensor fusion mechanism, and the following steps are executed by a central controller: Step S1: Real-time collection of environment point cloud data, cargo image data, distance data and inertial measurement data in the loading and unloading area by the multi-sensor fusion mechanism; Step S2: Fusion processing of the collected data, construction of a dynamic environment model, and identification of cargo position, attitude and obstacle distribution; Step S3: Based on the dynamic environment model, an optimal motion path from the current position to the target loading and unloading point is generated through an adaptive path planning algorithm; Step S4: Control the forklift host to move along the planned path, and adjust the height and attitude of the movable fork frame through the chain wheel and chain lifting mechanism to realize automatic forking or placing of the cargo; Step S5: Real-time monitoring of cargo state and environmental changes during loading and unloading, and adjustment of motion parameters through feedback control until the loading and unloading task is completed.
[0008] Preferably, the adaptive path planning algorithm in step S3 is a multi-modal fusion path planning algorithm, and the path evaluation function is as follows:
[0009] Wherein: represents a candidate path; is the path length; is the path safety score, calculated according to obstacle distance, ground flatness and stability; is the estimated energy consumption; is the estimated time; is the maximum normalized reference value of the corresponding parameter, respectively; is an adjustable weight coefficient, and satisfies , and is dynamically adjusted according to the task type.
[0010] Preferably, the attitude of the active fork is controlled in step S4 by a dynamic attitude adjustment algorithm based on the real-time adjustment of the cargo center of gravity offset and the fork inclination feedback, and the adjustment formula is:
[0011] Wherein: is the adjusted target inclination; is the real-time center of gravity offset error; are the proportional, integral, and derivative coefficients, respectively, which are optimized online by an adaptive learning module.
[0012] Preferably, step S5 includes real-time obstacle avoidance and collision avoidance strategies, which detect dynamic obstacles in real time through a multi-sensor fusion mechanism, and use a local path re-planning strategy based on the speed obstacle method, and the safety distance model is as follows:
[0013] Wherein: is the minimum safety distance; is the relative speed of the forklift and the obstacle; is the system response time; is the maximum deceleration; is the buffer distance constant.
[0014] Preferably, it also includes an adaptive learning and optimization module, which updates the path planning and attitude control parameters based on historical loading and unloading task data through a reinforcement learning algorithm, and the optimization objective function is:
[0015] Wherein: is the control strategy parameter; is the state action obtained immediate reward; is the discount factor; is the total number of task time steps.
[0016] Preferably, it also includes a fault diagnosis and self-recovery mechanism, which automatically switches to a degraded mode when a sensor anomaly or actuator failure is detected, and responds according to the following decision tree: If it is a local sensor failure, enable redundant sensors and re-fuse data; If it is an actuator stuck, try reverse driving and vibration relief; If it is a communication interruption, perform a slow return based on historical path and map data; If it is a multi-system failure, stop immediately and issue an audible and visual alarm.
[0017] The application provides a road freight automatic loading and unloading device. The application provides a road freight automatic loading and unloading device, and the technical advantage of the application is that through fusion sensing data, an adaptive multi-target path planning algorithm and a dynamic posture closed-loop control algorithm are adopted to realize full-process autonomous loading and unloading operation from intelligent decision, accurate execution to real-time optimization, improve operation success rate and comprehensive efficiency, and the device has high safety in operation, can actively avoid static and dynamic obstacles, and intelligently compensate for cargo deviation in the operation process, effectively preventing collision, overturning and cargo damage risk.
[0018] The application provides a road freight automatic loading and unloading device, and the technical advantage of the application is that through fusion sensing data, an adaptive multi-target path planning algorithm and a dynamic posture closed-loop control algorithm are adopted to realize full-process autonomous loading and unloading operation from intelligent decision, accurate execution to real-time optimization, improve operation success rate and comprehensive efficiency, and the device has high safety in operation, can actively avoid static and dynamic obstacles, and intelligently compensate for cargo deviation in the operation process, effectively preventing collision, overturning and cargo damage risk. BRIEF DESCRIPTION OF DRAWINGS
[0019] Figure 1 It is a front view schematic diagram of the application. Figure 2 It is a rear view schematic diagram of the application. Figure 3 It is a control method flow schematic diagram of the application. Figure 4 It is a system composition and data flow schematic diagram of the application.
[0020] Among them, 1, forklift host; 2, control panel; 3, electric wheel; 4, fixed frame body; 5, movable fork frame; 6, multi-sensor fusion mechanism. DETAILED DESCRIPTION
[0021] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.
[0022] As Figures 1-4As shown, the embodiment of the present application provides a kind of road freight automatic loading and unloading device, including forklift host 1: the front end of forklift host 1 is provided with control panel 2, forklift host 1 rear end is fixedly provided with fixed frame body 4, fixed frame body 4 and forklift host 1 between by sprocket chain lifting mechanism setting movable fork frame 5, the upper end of fixed frame body 4 is fixed with multi-sensor fusion mechanism 6.
[0023] The lower end surface of forklift host 1 is provided with electric wheel 3 at four corners.
[0024] A kind of road freight automatic loading and unloading control method, method based on the environment and cargo data collected by multi-sensor fusion mechanism 6, the following steps are executed by central controller: Step S1: the environment point cloud data, cargo image data, distance data and inertial measurement data of loading and unloading area are collected in real time by multi-sensor fusion mechanism 6; Step S2: the collected data are fused, and dynamic environment model is constructed, and cargo position, attitude and obstacle distribution are identified; Step S3: based on dynamic environment model, the optimal motion path from current position to target loading and unloading point is generated by adaptive path planning algorithm; Step S4: control forklift host 1 moves along the planned path, and the height and attitude of movable fork frame 5 are adjusted by sprocket chain lifting mechanism, to realize the automatic fork of cargo or placement; Step S5: cargo state and environmental change are monitored in real time during loading and unloading, and motion parameters are adjusted by feedback control until loading and unloading task is completed.
[0025] The adaptive path planning algorithm in step S3 is multi-modal fusion path planning algorithm, and the path evaluation function is as follows:
[0026] Wherein: Indicates candidate path; It is path length; It is path safety score, is calculated according to obstacle distance, ground flatness and stability; It is estimated energy consumption; It is estimated time; It is the maximum normalized reference value of corresponding parameter respectively; It is adjustable weight coefficient, and satisfies According to task type dynamic adjustment.
[0027] The attitude of movable fork frame 5 is controlled by dynamic attitude adjustment algorithm in step S4, which is based on cargo gravity center offset and fork inclination feedback for real-time adjustment, and the adjustment formula is as follows:
[0028] wherein: is the adjusted target inclination angle; is the real-time center of gravity offset error; are the proportional, integral, and derivative coefficients, respectively, which are optimized online by the adaptive learning module.
[0029] Real-time obstacle avoidance and collision avoidance strategies are included in step S5. Dynamic obstacles are detected in real time by the multi-sensor fusion mechanism 6, and a local path replanning strategy based on the speed obstacle method is used. The safety distance model is as follows:
[0030] wherein: is the minimum safety distance; is the relative speed between the forklift and the obstacle; is the system response time; is the maximum deceleration; is the buffer distance constant.
[0031] An adaptive learning and optimization module is also included. Based on historical loading and unloading task data, the module updates the path planning and posture control parameters through a reinforcement learning algorithm. The optimization objective function is as follows:
[0032] wherein: is the control strategy parameter; is the state is the action is the immediate reward obtained; is the discount factor; is the total number of task time steps.
[0033] A fault diagnosis and self-recovery mechanism is also included. When sensor abnormalities or actuator failures are detected, the system automatically switches to a degraded mode and responds according to the following decision tree: If it is a local sensor failure, redundant sensors are enabled and data is re-fused; If it is an actuator stuck, reverse driving and vibration relief are attempted; If it is a communication interruption, a slow return to the home position is performed based on historical path and map data; If it is a multi-system failure, it immediately stops and sends an audible and visual alarm.
[0034] The present technology first passes through the comprehensive perception stage of the environment and the goods by the multi-sensor fusion mechanism 6. When the device starts, the multi-sensor fusion mechanism 6 synchronously collects the laser point cloud data, visual image information, ultrasonic distance signal and inertial measurement unit data of the loading and unloading area. These raw data are transmitted to the central controller in real time, processed by a set of special data fusion algorithm, first registered and segmented with the point cloud and image, the goods contour, accurate position, stacking posture and the distribution and motion trend of the surrounding obstacles are identified, and the attitude and micro-motion of the forklift itself are perceived in combination with the inertial data. On this basis, the system constructs a high-precision dynamic three-dimensional environment model, which not only contains static elements, but also can update the information of dynamic targets such as moving personnel and other equipment in real time. The outstanding advantage of this stage is that it completely overcomes the blind area and misjudgment problem caused by the dependence of traditional infrared forklifts on single sensing mode, and through multi-source information complementation, it ensures the comprehensiveness and accuracy of environmental understanding, lays a reliable perception cornerstone for subsequent fully autonomous decision-making, and improves the adaptability of the system in complex and variable working scenarios.
[0035] After completing the environment modeling, the system enters the intelligent planning and decision-making stage. The central controller calls the adaptive path planning algorithm based on the dynamic environment model and the task target (such as picking up the goods at the specified location). The algorithm does not simply calculate the shortest path, but comprehensively evaluates the path length, safety margin, expected energy consumption and execution efficiency, etc. in multiple dimensions, and generates an optimal motion trajectory from the current position to the target point in a rapidly changing environment. At the same time, the system will calculate the best approach angle, lifting height and fork opening width required for the active fork 5 according to the identified goods size, center of gravity estimation and stacking condition, forming a complete sequence of action instructions. The advantage of this stage lies in the intelligence and global optimization characteristics of its decision-making, the path planning takes into account safety and efficiency, and the action planning ensures the success rate of the operation, avoiding the repeated adjustment, path conflict and operation failure of traditional automated equipment due to rigid planning, thereby greatly improving the overall process efficiency and economy of loading and unloading operations.
[0036] Next is the precise execution and real-time closed-loop control phase. The planned path and action sequence are issued to the underlying execution mechanism. The forklift host 1 drives the electric wheel 3 to start moving along the planned path, and the chain wheel and chain lifting mechanism controls the lifting of the movable fork 5 according to the instructions. During the entire movement process, the multi-sensor fusion mechanism 6 continuously monitors at a high frequency, and the real-time micro-motion deviation of the goods, the actual attitude of the fork, and the new sudden obstacles in the environment are fed back to the central controller as feedback signals. The controller adjusts the execution instructions in real time through dynamic attitude adjustment algorithms and local obstacle avoidance re-planning strategies, such as automatically compensating for the moment of inertia caused by unbalanced goods during forklifting, or smoothly bypassing suddenly appearing obstacles during movement. This phase forms a real-time closed loop of perception, decision-making, execution, and feedback, which has the advantages of improving automation to a high level of autonomy and precision, achieving flexible operation and active safety protection without human intervention, effectively preventing risks such as goods falling and equipment collision, and ensuring the high reliability and stability of the operation process.
[0037] Finally, the system self-learning optimization and safety fault-tolerant phase. Each complete loading and unloading task data, including sensor logs, control instruction sequences, and final execution effects, are recorded by the system and sent to the adaptive learning module for analysis. The module analyzes historical data to continuously optimize weight parameters in path planning and adjustment coefficients in attitude control, enabling the device to gradually adapt to the layout characteristics of a specific warehouse or the physical properties of commonly used goods, becoming more intelligent. More importantly, the system has a multi-level fault diagnosis and self-recovery mechanism. Once an abnormal sensor data, communication delay, or execution mechanism response deviation is detected, the system will immediately start the diagnosis tree to determine the fault level and automatically switch to a redundant sensing mode, execute a preset safety action, or enter a slow emergency stop state, while sending an alarm to the remote monitoring terminal. The highlight of this phase is to give the device long-term evolution ability and strong fault tolerance, not only improving long-term operation performance through continuous learning, but also avoiding the possibility of system paralysis caused by a single fault point through perfect fault tolerance design, ensuring high operation reliability and safety in unattended working conditions.
[0038] Although embodiments of the present application have been shown and described, it will be understood by those having ordinary skill in the art that various changes, modifications, replacements and variations can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.
Claims
1. An automated loading and unloading device for road freight transport, comprising a forklift host (1), characterized in that: The front end of the forklift host (1) is provided with a control panel (2), and the rear end of the forklift host (1) is fixedly provided with a fixed frame (4). The fixed frame (4) and the forklift host (1) are connected by a sprocket and chain lifting mechanism to provide a movable fork (5). The upper end of the fixed frame (4) is fixed with a multi-sensor fusion mechanism (6).
2. The automatic loading and unloading device for road freight transport according to claim 1, characterized in that: Electric wheels (3) are provided at the four corners of the lower end face of the forklift host (1).
3. An automatic loading and unloading device for road freight transport according to claim 1 or 2, characterized in that, It also includes an automatic loading and unloading control method for road freight, which, based on environmental and cargo data collected by the multi-sensor fusion mechanism (6), executes the following steps through a central controller: Step S1: Real-time acquisition of environmental point cloud data, cargo image data, distance data and inertial measurement data of the loading and unloading area through the multi-sensor fusion mechanism (6); Step S2: The collected data is fused and processed to construct a dynamic environment model and identify the cargo position, posture and obstacle distribution; Step S3: Based on the dynamic environment model, generate the optimal movement path from the current position to the target loading / unloading point through an adaptive path planning algorithm; Step S4: Control the forklift host (1) to move along the planned path, and adjust the height and posture of the movable fork carriage (5) through the sprocket and chain lifting mechanism to realize the automatic picking or placing of goods; Step S5: Monitor the cargo status and environmental changes in real time during loading and unloading, and adjust motion parameters through feedback control until the loading and unloading task is completed.
4. The automatic loading and unloading control method for road freight transport according to claim 3, characterized in that: The adaptive path planning algorithm in step S3 is a multimodal fusion path planning algorithm, and its path evaluation function is as follows: .
5. The automatic loading and unloading control method for road freight transportation according to claim 3, characterized in that: In step S4, the attitude of the movable fork (5) is controlled by a dynamic attitude adjustment algorithm. This algorithm makes real-time adjustments based on the offset of the cargo center of gravity and the fork tilt angle feedback. The adjustment formula is as follows: 。 6. The automatic loading and unloading control method for road freight transport according to claim 3, characterized in that: Step S5 includes a real-time obstacle avoidance and collision avoidance strategy. Dynamic obstacles are detected in real time through a multi-sensor fusion mechanism (6), and a local path replanning strategy based on the speed obstacle method is adopted. The safe distance model is as follows: .
7. The automatic loading and unloading control method for road freight transport according to claim 3, characterized in that: It also includes an adaptive learning and optimization module, which updates path planning and attitude control parameters based on historical loading and unloading task data using a reinforcement learning algorithm. Its optimization objective function is: 。 8. The automatic loading and unloading control method for road freight transport according to claim 3, characterized in that: It also includes fault diagnosis and self-recovery mechanisms. When a sensor malfunction or actuator failure is detected, the system automatically switches to degraded mode and responds according to the following decision tree: If a local sensor fails, a redundant sensor is activated and the data is re-fused. If the actuator is stuck, try reverse drive and vibration release; If communication is interrupted, a slow return to the original location will be performed based on historical paths and map data; If there is a multi-system failure, the system will stop immediately and issue an audible and visual alarm.