Lateral balance control method and system for a heavy telescopic forklift

Abstract

The present invention discloses a lateral balance control method and system for a heavy telescopic forklift, which relates to the field of hoisting, lifting, traction or pushing not included in other categories. It obtains three-dimensional data of the ground environment and the goods in the cargo box through drone scanning, constructs an accurate three-dimensional map and a three-dimensional model of the goods, divides path nodes according to the terrain features and the vehicle body side inclination angle, predicts the dynamic changes of the goods based on the goods type, and then optimizes the navigation path and driving speed of the forklift between the path nodes in real time according to the dynamic changes of the goods to maintain the lateral balance of the vehicle body.

Description

Technical Field

[0001] The present invention relates to the field of hoisting, lifting, towing or pushing not included in other categories, and specifically to a lateral balance control method and system for a heavy telescopic forklift. Background Art

[0002] A heavy telescopic forklift is a multi-purpose forklift with off-road performance and a telescopic boom, and is usually used in the construction industry and certain industrial fields.

[0003] When transporting goods in complex terrains, a heavy telescopic forklift faces a major technical challenge: how to maintain the balance of the vehicle body and safely and efficiently complete the goods transportation task in an uneven and diverse ground environment. The core of this problem lies in that traditional forklifts are prone to rollover when dealing with complex terrains, seriously threatening the safety of operators and the integrity of goods. Specifically, when the forklift is driving on a terrain with uneven road surfaces, the inclination angle of the vehicle body will constantly change, and the goods in the cargo box will also change dynamically accordingly. If this change cannot be timely and accurately perceived and addressed, it is easy for the vehicle body to lose balance. What's more complex is that different types of goods may produce different degrees of displacement or deformation during transportation, further increasing the difficulty of maintaining balance. In addition, due to the complexity of the environment, the pre-planned transportation path may not be adaptable to the actual situation and needs to be adjusted in real time during transportation. These factors combined form a complex technical problem involving multiple aspects and variables, posing extremely high requirements for the intelligent control and path planning of the forklift. Summary of the Invention

[0004] The purpose of the present invention is to provide a lateral balance control method and system for a heavy telescopic forklift, aiming to overcome the above problems existing in the prior art.

[0005] To achieve the purpose, the present invention provides the following technical solutions:

[0006] A lateral balance control method for a heavy telescopic forklift, with a drone as the leader and the forklift as the follower, to control the lateral balance of the forklift body, which specifically includes the following steps:

[0007] Step S1: The drone scans the ground environment; the central control module obtains the drone scan data and constructs a three-dimensional map data structure;

[0008] Step S2: The central control module gives the navigation path and driving speed from the starting point to the ending point according to the above three-dimensional map data structure;

[0009] Step S3: The drone scans the goods in the cargo box in real time and establishes three-dimensional model data of the goods;

[0010] Step S4: Determine the type of goods;

[0011] Step S5: Using the three-dimensional model data of the goods at the current node as the initial value, according to the type of goods, as well as the navigation path, driving speed, and three-dimensional map data structure between the current node and the next node, predict the dynamic changes of the goods, calculate the three-dimensional model data of the goods at the next node, and then calculate the difference in goods between the left half and the right half of the cargo box at the next node;

[0012] Step S6: If the difference in goods at the next node exceeds the second preset threshold, optimize the navigation path and driving speed of the forklift from the current node to the next node.

[0013] Furthermore, in the above Step S1, the central control module obtains the laser point cloud data collected by the drone scanning, extracts the spatial feature points according to the above laser point cloud data, and generates the spatial distribution resolution data of the environment; obtains the spectral data collected by the drone scanning, and performs fusion processing on the above spectral data and the point cloud data filtered by noise to construct a three-dimensional map data structure.

[0014] Furthermore, in the above Step S3, the drone scans the goods in the cargo box in real time and establishes the three-dimensional model data of the goods. Specifically:

[0015] The terahertz imaging device carried by the drone performs omnidirectional and multi-angle transmission scanning on the cargo box to obtain the cargo transmission depth data and the material reflection coefficient; analyzes the resolution and acquisition frequency of the above transmission depth data, determines the cargo boundary clarity threshold, and generates a terahertz image in combination with the dynamic range of the material reflection coefficient; the central control module uses the multi-angle terahertz images collected as projection data and constructs the three-dimensional model data of the goods by using the computer tomography algorithm.

[0016] Furthermore, pre-construct a database of the material reflection coefficients and three-dimensional model data of known goods types; pre-construct a convolutional neural network (CNN), and use the data in the database as prior knowledge to train and optimize the convolutional neural network (CNN);

[0017] In the above Step S4, to determine the type of goods, specifically: use the material reflection coefficient and three-dimensional model data obtained by the drone scanning as input parameters and input them into the trained convolutional neural network, and the convolutional neural network identifies and outputs the type of goods. Furthermore, in the above Step S6, if the difference in goods exceeds the second preset threshold, optimize the navigation path and driving speed of the forklift from the current node to the next node. Specifically:

[0018] Combining navigation path optimization and driving speed adjustment, establish a comprehensive optimization model. The goal of this comprehensive optimization model is to minimize the risk of the forklift body rollover and maximize the driving efficiency of the forklift.

[0019] Furthermore, a multi-objective optimization method is adopted to optimize the roll risk and driving efficiency as two independent objective functions:

[0020] Minimize the roll risk of the forklift body:

[0021] min f 1( x)=ω1×θ + ω2×F 轮胎 +ω3×Δm;

[0022] where x is the decision variable, including the navigation path parameter and the driving speed parameter; θ is the roll angle of the vehicle body; F 轮胎 is the tire force; Δm is the cargo difference; ω1, ω2, ω3 are the weight coefficients of each factor, and ω1 + ω2 + ω3 = 1;

[0023] Maximize the driving efficiency of the forklift:

[0024] max f2(x)=(L 优化 / L 初始 )×(T 初始 / T 优化 );

[0025] where L 初始 is the path length before optimization; L 优化 is the path length after optimization; T 初始 is the driving time before optimization; T 优化 is the driving time after optimization.

[0026] Furthermore, through the weighted summation method, the two objective functions are transformed into a comprehensive objective function:

[0027] F(x)=α×min f1(x)+β×max f2(x);

[0028] where α and β are the weight coefficients, and α + β = 1, which are used to balance the influence of the roll risk and the driving efficiency.

[0029] The present invention also discloses a lateral balance control system for a heavy telescopic forklift. This system is used to execute the above control method and includes a drone, a central control module, and a forklift. The central control module is communicatively connected to the drone and the forklift; the above forklift is equipped with a lateral acceleration sensor, a speed sensor, and a biaxial inclination sensor; the above drone is equipped with a lidar, a Herz imaging device, and a hyperspectral camera.

[0030] Furthermore, the above central control module is mounted on the forklift.

[0031] The present invention has the following beneficial effects compared with the prior art:

[0032] First, the present invention discloses a lateral balance control method for a heavy telescopic forklift. It obtains three-dimensional data of the ground environment and the goods in the cargo box through drone scanning, constructs an accurate three-dimensional map and a three-dimensional model of the goods, divides path nodes according to the terrain features and the vehicle body roll angle, predicts the dynamic changes of the goods based on the type of goods, and then optimizes the navigation path and driving speed of the forklift between the path nodes in real time according to the dynamic changes of the goods to maintain the lateral balance of the vehicle body.

[0033] Second, the present invention realizes stable and efficient transportation of the forklift in complex terrains by continuously obtaining feedback data for path optimization. This method effectively solves the problem that traditional forklifts are prone to rollover during transportation on uneven terrains, and significantly improves the safety and efficiency of goods transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flowchart of the lateral balance control method in the present invention.

[0035] Figure 2 It is a flowchart of the lateral balance control system in the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0036] The following describes the specific embodiments of the present invention with reference to the drawings. To fully understand the present invention, many details are described below. However, for those skilled in the art, the present invention can be implemented without these details.

[0037] As Figure 1 and Figure 2 shown, a lateral balance control method for a heavy telescopic forklift uses a drone as the leader and the forklift as the follower to control the lateral balance of the forklift body, which specifically includes the following steps:

[0038] Step S1: The drone scans the ground environment; the central control module obtains the drone scanning data and constructs a three-dimensional map data structure.

[0039] As Figure 1 and Figure 2 shown, in a specific embodiment, step S1 specifically includes the following contents:

[0040] Step S101: The drone is equipped with a lidar and a hyperspectral camera. When the drone flies, it scans the ground environment, collects lidar point cloud data through the lidar, and collects spectral data through the hyperspectral camera.

[0041] Step S102: The central control module obtains the lidar point cloud data collected by the drone scanning, extracts spatial feature points according to the lidar point cloud data, generates spatial distribution resolution data of the environment; and obtains the spectral data collected by the drone scanning.

[0042] Step S103: The central control module fuses the spectral data with the point cloud data that has been noise-filtered to construct a 3D map data structure.

[0043] Step S2: The central control module gives the navigation path and driving speed from the starting point (i.e., the position where the forklift loads the cargo box) to the ending point (i.e., the position where the forklift unloads the cargo box) according to the 3D map data structure; according to the 3D map data structure, taking the positions where the body roll angle exceeds the first preset threshold as nodes, the navigation path is divided into several segments. It should be noted that the 3D map data structure includes the starting point, the ending point, and the navigation path.

[0044] As Figure 1 and Figure 2 shown, in a specific embodiment, Step S2 specifically includes the following content:

[0045] Step S201: The central control module gives the navigation path and driving speed from the starting point to the ending point according to the 3D map data structure. Among them, the navigation path can at least avoid obstacles, and the driving speed shall not exceed the maximum speed limit. In addition, giving the navigation path and driving speed according to the starting point and the ending point belongs to the prior art and will not be elaborated here.

[0046] Step S202: The central control module takes the positions where the body roll angle exceeds the first preset threshold as nodes and divides the navigation path into several segments. It should be noted that: when the forklift is driving straight at a certain preset driving speed (for example, 3 km / h) under no-load conditions, if the road surface tilts left and right to a certain critical value, the forklift will roll over. This critical value is the first preset threshold.

[0047] Step S3: The drone scans the goods in the cargo box in real time to establish 3D model data of the goods;

[0048] As Figure 1 and Figure 2 shown, in a specific embodiment, Step S3 includes the following content:

[0049] Step S301: The terahertz imaging device carried by the drone performs omnidirectional and multi-angle transmission scanning on the cargo box to obtain the cargo transmission depth data and the material reflection coefficient; analyzes the resolution and acquisition frequency of the transmission depth data, determines the threshold of the cargo boundary clarity, and generates a terahertz image in combination with the dynamic range of the material reflection coefficient;

[0050] Step S302: The central control module takes the multi-angle terahertz images collected as projection data and constructs 3D model data of the goods using the computed tomography algorithm.

[0051] Step S4: Determine the type of goods;

[0052] AsFigure 1 and Figure 2 As shown in and

[0053] , in a specific embodiment, in step S4, the type of goods is determined through the prior knowledge of the goods, which specifically includes the following:

[0053] Step S401: Pre-build a database of material reflection coefficients, 3D model data, etc. of known goods types. This database is established and stored in the central control module.

[0054] Step S402: Build a convolutional neural network (CNN). This convolutional neural network (CNN) is established and stored in the central control module;

[0055] Step S403: Use the data in the database as prior knowledge to train and optimize the convolutional neural network (CNN).

[0056] Step S404: Use the material reflection coefficient and 3D model data obtained in step S301 as input parameters and input them into the trained convolutional neural network (CNN), and the convolutional neural network (CNN) identifies and outputs the goods type.

[0057] For example: The goods type can be classified according to one or more of the following methods:

[0058] (1) Classification by material. By analyzing the reflection coefficient, the material of the goods can be identified, so it can be classified by material.

[0059] For example, metal products: have a relatively high reflection coefficient, and a relatively high density and hardness; wood products: have a relatively low reflection coefficient and a low density; plastic products: have a low reflection coefficient and density; glass products: have a relatively high reflection coefficient and are fragile, etc.

[0060] (2) Classification by shape and size. By analyzing the 3D model data, the shape and size of the goods can be identified, so it can be classified by shape and size.

[0061] For example, goods with regular shapes: such as cubes, cuboids, cylinders, etc., the 3D model data of these goods is relatively simple and easy to identify; goods with irregular shapes: such as furniture, mechanical parts, etc., the 3D model data of these goods is relatively complex and requires more detailed scanning and analysis; large goods: such as mechanical equipment, etc., the weight and center of gravity position of these goods have a greater impact on the stability of the vehicle; small goods: such as electronic products, packaging boxes, etc., the weight and center of gravity position of these goods have a smaller impact on the stability of the vehicle.

[0062] (3) Classification by fixing method. By analyzing the 3D model data, the fixing method of the goods can be identified, so it can be classified by fixing method.

[0063] For example, goods are fixed with straps: For goods fixed with straps, special attention should be paid to the tightness of the straps and the fixing points during transportation; goods are fixed with brackets: For goods fixed with brackets, these goods usually have a relatively high center of gravity, and special attention should be paid to the stability of the vehicle; goods are placed freely: For bulk goods or packaged goods placed freely, these goods are prone to movement during transportation, and special attention should be paid to fixing and distribution.

[0064] Correspondingly, the database can further include the following data: (1) The weight and center of gravity position of the goods: which helps to predict the stability of the goods during vehicle driving. (2) The size and shape of the goods: so as to accurately represent the physical characteristics of the goods in prior knowledge or models. (3) The way of fixing the goods: determining how the goods are fixed on the vehicle, such as using straps, brackets or other fixing devices. These data help to identify the type of goods, help to evaluate the stability of the goods during driving, and predict the dynamic changes of the goods.

[0065] Step S5: Taking the three-dimensional model data of the goods at the current node as the initial value, according to the type of goods, as well as the navigation path, driving speed and three-dimensional map data structure between the current node and the next node, predict the dynamic changes of the goods, calculate the three-dimensional model data of the goods at the next node, and then calculate the difference in goods between the left half and the right half of the cargo box at the next node.

[0066] Such as Figure 1 and Figure 2 As shown, in a specific embodiment, in step S5, it specifically includes the following contents:

[0067] Step S501: The central control module takes the three-dimensional model data of the goods at the current node as the initial value, and according to the type of goods, as well as the navigation path, driving speed and three-dimensional map data structure between the current node and the next node, predicts the dynamic changes of the goods and calculates the three-dimensional model data of the goods at the next node.

[0068] Step S502: Using the first vertical plane as the dividing plane, divide the goods into the left half and the right half. Among them, the first vertical plane is a plane parallel to the front-back direction of the forklift body and passing through the axis of symmetry of the forklift forks.

[0069] Step S503: According to the three-dimensional model data of the goods at the next node, calculate the difference in goods between the left half and the right half of the cargo box at the next node. Preferably, the difference in goods is the difference in the weight of the goods between the left half and the right half.

[0070] Step S6: If the difference in goods at the next node exceeds the second preset threshold, optimize the navigation path and driving speed of the forklift from the current node to the next node.

[0071] Such asFigure 1 and Figure 2 As shown in Figure 2 , in a specific embodiment, step S6 specifically includes the following content:

[0072] Step S601: Pre-combine navigation path optimization and driving speed adjustment to establish a comprehensive optimization model. The goal of this comprehensive optimization model is to minimize the risk of forklift body roll and maximize the driving efficiency of the forklift. This comprehensive optimization model is established and stored in the central control module.

[0073] More specifically, step S601 includes but is not limited to the following content:

[0074] Step S6011: Adopt a multi-objective optimization method to optimize the roll risk and driving efficiency as two independent objective functions:

[0075] Minimize the risk of forklift body roll: min f1(x) = ω1×θ + ω2×F 轮胎 + ω3×Δm;

[0076] Where x is the decision variable, including navigation path parameters and driving speed parameters; θ is the body roll angle (unit: radian); F 轮胎 is the tire force (unit: Newton); Δm is the cargo difference (unit: kilogram); ω1, ω2, ω3 are the weight coefficients of each factor, and ω1 + ω2 + ω3 = 1;

[0077] Maximize the driving efficiency of the forklift: max f2(x) = (L 优化 / L 初始 )×(T 初始 / T 优化 );

[0078] Where L 初始 is the path length from the current node to the next node before optimization; L 优化 is the path length from the current node to the next node after optimization; T 初始 is the driving time from the current node to the next node before optimization; T 优化 is the driving time from the current node to the next node after optimization. Both driving times can be calculated by dividing the path length by the driving speed.

[0079] Step S6012: Through the weighted summation method, transform the two objective functions into a comprehensive objective function:

[0080] F(x) = α×min f1(x) + β×max f2(x);

[0081] Where α and β are weight coefficients, and α + β = 1, which are used to balance the influence of roll risk and driving efficiency.

[0082] The above comprehensive objective function is the core of the comprehensive optimization model, which is used to balance the influence of roll risk and driving efficiency. It can, from the perspective of global optimization, directly guide the method of the comprehensive optimization model. By adjusting the weight coefficients α and β, the priorities of roll risk and driving efficiency can be flexibly controlled.

[0083] Step S602: The central control module determines whether the cargo difference exceeds the second preset threshold; if it does not exceed the second preset threshold, step S603 is executed; if it exceeds the second preset threshold, step S604 is executed.

[0084] Step S603: The forklift moves from the current node to the next node according to the navigation path and driving speed given in step S201; after the forklift reaches the next node, it jumps to step S5 until it reaches the end point.

[0085] Step S604: Optimize the navigation path and driving speed of the forklift from the current node to the next node; the forklift moves from the current node to the next node according to the optimized navigation path and driving speed; after the forklift reaches the next node, it jumps to step S5 until it reaches the end point.

[0086] As Figure 1 and Figure 2 shown, in a specific implementation, in step S604, the comprehensive optimization model takes the path length, the number of inflection points, and the lateral acceleration as weight factors to give the optimal navigation path; takes the body roll angle, the tire force, and the cargo difference at the current node as weight factors to dynamically adjust the driving speed to ensure the lateral balance of the forklift body during driving. The specific contents are as follows:

[0087] Step S6041: The weight calculation formula for navigation path optimization is:

[0088] Path optimization score = a1 × path length + a2 × number of inflection points + a3 × lateral acceleration;

[0089] Among them, the path length refers to the path length from the current node to the next node after optimization; the number of inflection points refers to the number of points where the direction changes significantly in the path of the forklift from the current node to the next node after optimization, and the lateral acceleration refers to the maximum acceleration generated when the forklift turns at each inflection point after optimization; a1, a2, and a3 are the corresponding weight coefficients of the path length, the number of inflection points, and the lateral acceleration respectively, and a1 + a2 + a3 = 1.

[0090] The comprehensive optimization model improves the driving efficiency of the forklift by optimizing the path length, reducing the number of inflection points, and reducing the lateral acceleration. Therefore, the path optimization score directly affects the driving efficiency max f2(x), and thus indirectly affects the comprehensive objective function F(x).

[0091] Step S6042. The weight calculation formula for the driving speed adjustment is as follows:

[0092] Speed adjustment score = ω1×θ + ω2×F 轮胎 + ω3×Δm;

[0093] where θ is the body roll angle of the forklift when it reaches the next node after optimization; F 轮胎 is the maximum tire force of all wheels of the forklift when it reaches the next node after optimization; Δm is the cargo difference of the forklift when it reaches the next node after optimization; ω1, ω2, and ω3 are the corresponding weight coefficients of the roll angle, tire force, and cargo difference respectively, and ω1 + ω2 + ω3 = 1.

[0094] The comprehensive optimization model reduces the risk of body roll by dynamically adjusting the driving speed. Therefore, the speed adjustment score directly affects the body roll risk min f1(x), and thus indirectly affects the comprehensive objective function F(x).

[0095] Step S6043. Calculate the comprehensive score through the following formula:

[0096] Comprehensive score = γ×Path optimization score + δ×Speed adjustment score;

[0097] where γ and δ are the weight coefficients of path optimization and speed adjustment respectively, and γ + δ = 1.

[0098] The comprehensive score is a specific quantitative index in the actual optimization process of the comprehensive optimization model, which is used to guide the final decision of navigation path optimization and driving speed adjustment. The comprehensive score is a specific implementation form of the comprehensive objective function F(x). By optimizing the comprehensive score, the optimization of the comprehensive objective function can be indirectly achieved.

[0099] Step S6044. If the comprehensive score reaches the third preset threshold, output the optimized navigation path and driving speed; if the comprehensive score does not reach the third preset threshold, repeat steps S6041 - S6044; when the number of repetitions exceeds N times, pause the forklift and send out a warning signal indicating that the forklift has a risk of tipping over, waiting for manual intervention.

[0100] Such as Figure 2As shown in the figure, the present invention also discloses a lateral balance control system for a heavy telescopic forklift, which is used to execute the above control method and includes an unmanned aerial vehicle, a central control module, and a forklift. The central control module is communicatively connected to the unmanned aerial vehicle and the forklift; the forklift is equipped with a lateral acceleration sensor, a speed sensor, and a biaxial inclination sensor. Among them, the biaxial inclination sensor is used to measure the roll angle of the forklift body; the unmanned aerial vehicle is equipped with a lidar, a Hertz imaging device, and a hyperspectral camera. Preferably, the central control module is carried on the central control system of the forklift. Of course, the central control module can also be set in a computer in the control room and connected to the forklift and the unmanned aerial vehicle through wireless communication to achieve remote communication and control.

[0101] The above is only the specific implementation manner of the present invention, but the design concept of the present invention is not limited thereto. Any non-substantial modification of the present invention using this concept shall fall within the scope of infringement of the protection of the present invention.

Claims

1. A lateral balance control method for a heavy telescopic forklift, characterized in that: With a drone as the leader and a forklift as the follower, the lateral balance of the forklift body is controlled, specifically including the following steps: Step S1: The drone scans the ground environment; the central control module obtains the drone scan data and constructs a three-dimensional map data structure; Step S2: The central control module gives the navigation path and driving speed from the starting point to the end point according to the three-dimensional map data structure; according to the three-dimensional map data structure, the navigation path is divided into several segments with the positions where the vehicle body roll angle exceeds the first preset threshold as nodes; Step S3: The drone scans the goods in the cargo box in real time and establishes three-dimensional model data of the goods, specifically including: the terahertz imaging device carried by the drone performs omnidirectional and multi-angle transmission scans on the cargo box to obtain the goods transmission depth data and material reflection coefficient; analyze the resolution and acquisition frequency of the transmission depth data, determine the goods boundary clarity threshold, and generate a terahertz image in combination with the dynamic range of the material reflection coefficient; the central control module uses the multi-angle terahertz images collected as projection data and constructs three-dimensional model data of the goods using a computed tomography algorithm; Step S4: Determine the type of goods; A database of material reflection coefficients and three-dimensional model data of known goods types is pre-constructed; a convolutional neural network is pre-constructed, and the data in the database is used as prior knowledge to train and optimize the convolutional neural network; in step S4, determining the type of goods is specifically: using the material reflection coefficient and three-dimensional model data obtained by the drone scan as input parameters, inputting them into the trained convolutional neural network, and the convolutional neural network identifies and outputs the type of goods; Step S5: Using the three-dimensional model data of the goods at the current node as the initial value, according to the type of goods, as well as the navigation path, driving speed, and three-dimensional map data structure between the current node and the next node, predict the dynamic changes of the goods, calculate the three-dimensional model data of the goods at the next node, and then calculate the goods difference between the left half and the right half of the cargo box at the next node; Step S6: If the goods difference at the next node exceeds the second preset threshold, optimize the navigation path and driving speed of the forklift from the current node to the next node, specifically including: combining navigation path optimization and driving speed adjustment, establishing a comprehensive optimization model, and the goal of this comprehensive optimization model is to minimize the vehicle body roll risk of the forklift and maximize the driving efficiency of the forklift.

2. The lateral balance control method of a heavy telescopic forklift according to claim 1, characterized in that: In step S1, the central control module obtains the laser point cloud data collected by the drone scan, extracts spatial feature points according to the laser point cloud data, and generates spatial distribution resolution data of the environment; obtains the spectral data collected by the drone scan, and performs fusion processing on the spectral data and the point cloud data after noise filtering to construct a three-dimensional map data structure.

3. A method for controlling the lateral balance of a heavy telescopic forklift according to claim 1, characterized in that: Adopt a multi-objective optimization method to optimize the roll risk and driving efficiency as two independent objective functions: Minimize the vehicle body roll risk of the forklift: min f1(x)=ω1×θ+ω2×F 轮胎 +ω3×Δm; where x is a decision variable, including navigation path parameters and driving speed parameters; θ is the vehicle body roll angle; F 轮胎 is the tire force; Δm is the cargo difference; ω1, ω2, ω3 are the weight coefficients of various factors, and ω1 + ω2 + ω3 = 1; Maximize the driving efficiency of the forklift: max f2(x)=(L 优化 / L 初始 )×(T 初始 / T 优化 ); Among them, L 初始 is the path length before optimization; L 优化 is the path length after optimization; T 初始 is the travel time before optimization; T 优化 is the travel time after optimization.

4. A lateral balance control method for a heavy telescopic forklift according to claim 3, characterized in that: By the weighted summation method, the two objective functions are transformed into a comprehensive objective function: F(x) = α × min f1(x) + β × max f2(x); Wherein, α and β are weighting coefficients, and α + β = 1, which are used to balance the influence of roll risk and driving efficiency.

5. A lateral balance control system for a heavy telescopic forklift, characterized in that: This system is used to execute a lateral balance control method for a heavy telescopic forklift as described in any one of claims 1-4, including a drone, a central control module and a forklift, and the central control module is communicatively connected to the drone and the forklift; the forklift is equipped with a lateral acceleration sensor, a speed sensor and a biaxial inclination sensor; the drone is equipped with a lidar, a Herz imaging device and a hyperspectral camera.

6. The lateral balance control system of a heavy telescopic forklift according to claim 5, characterized in that: The central control module is mounted on the forklift.

Images