A carbon block grouping and placing method based on multi-modal perception and dynamic safety evaluation

Abstract

The present application relates to the field of industrial automation and hoisting and transportation technology, and particularly relates to a carbon block grouping and placing method based on multi-modal perception and dynamic safety evaluation, comprising the following steps: establishing a furnace chamber three-dimensional model; collecting carbon block grouping position information, controlling the movement and clamping of a clamping mechanism; obtaining the contour information and clamping force information of the carbon block grouping, establishing a carbon block grouping three-dimensional model, and calculating the safety envelope of the carbon block grouping under the current posture; calculating the dynamic safety margin of placing the carbon block grouping into a target furnace chamber under the current clamping state; constructing a dynamic virtual channel with adaptive cross-sectional size; planning a collision-free motion trajectory for the clamping mechanism based on the dynamic virtual channel constraint; and controlling the clamping mechanism to move according to the planned collision-free motion trajectory, and placing the carbon block grouping at a target position in the furnace chamber. The present application solves the problem of bumping caused by irregular shape and uncertainty of posture when the carbon block grouping is placed into the furnace chamber, and greatly reduces the damage such as edge collapse and corner drop of the carbon block.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of industrial automation and hoisting and transportation technology, in particular to a carbon block grouping and placing method based on multi-modal perception and dynamic safety evaluation. BACKGROUND

[0002] In the baking production of carbon products, the green body needs to go through the high-temperature heat treatment process of "loading furnace-temperature rising-holding cooling-out of the furnace" to realize the densification of its structure and the stability of its performance. The loading operation is the starting and key link of the whole baking process, and its core is to accurately lift the carbon block grouping completed by the grouping machine into the baking furnace chamber, which is crucial to the final product quality, because the position, spacing and attitude of the carbon block in the furnace chamber directly determine the uniformity of its heating in the subsequent high-temperature treatment process, and then affect the consistency of the physical and chemical indicators of the product.

[0003] At present, this link mainly relies on manual remote control or semi-automatic scheme with preset program. The typical process is as follows: the operator roughly positions the carbon block grouping by visual observation or fixed two-dimensional visual sensor, and then moves the clamping mechanism of the multifunctional crane to the predetermined coordinates for grabbing. Finally, the equipment moves and places the whole carbon block grouping into the baking furnace chamber according to a fixed spatial trajectory previously demonstrated.

[0004] However, the above-mentioned prior art scheme has the following technical defects in the actual application of carbon baking workshop:

[0005] (1) Although the carbon block grouping itself is a regular whole, the single carbon block green body has size tolerance, and after being clamped, the horizontal rotation angle (yaw) and slight inclination attitude of the heavy grouping in the lifting appliance are uncertain. The internal space of the baking furnace chamber is compact, and the running clearance between the carbon block grouping and the internal fixed parts of the furnace chamber is extremely small (usually 30-80 mm). The fixed clamping point and the rigid motion trajectory cannot adapt to the fluctuation of such attitude, which brings great challenges to accurate and efficient alignment.

[0006] (2) When the carbon block grouping is placed into the furnace chamber, any yaw caused by uncertain attitude is extremely easy to cause the high-value carbon block grouping to collide with the inner wall of the baking furnace chamber or the internal heating system. Since the green body has low strength before baking, such collision may cause the carbon block green body to crack, corner drop and other damage, or even become a waste product, causing serious waste of raw materials and energy; the collision may also scratch or break the lining refractory bricks in the furnace chamber, which has extremely high maintenance cost and may cause the baking furnace to be out of production for a long time, seriously affecting the production plan.

[0007] (3) The existing scheme lacks real-time perception and closed-loop feedback of the real pose of the heavy carbon block after clamping, and is essentially an open-loop or semi-blind operation. In the face of the above uncertainties, the system cannot independently correct and decide online, often requiring a speed reduction or relying on repeated fine-tuning by the operator, which not only significantly reduces the efficiency of the furnace charging operation and prolongs the kiln turnover cycle, but also is more likely to cause operation accidents due to human fatigue or misjudgment.

[0008] Therefore, the carbon baking industry urgently needs an advanced furnace charging and placing method that can intelligently perceive the real-time pose of heavy carbon block grouping, dynamically evaluate its placement risk in the narrow furnace chamber space, and autonomously plan a collision-free path, thereby ensuring furnace charging quality, protecting equipment safety, and comprehensively improving production automation level and efficiency. SUMMARY

[0009] In order to solve the problems existing in the prior art, the present application provides a carbon block grouping and placing method based on multi-modal perception and dynamic safety evaluation, comprising the following steps:

[0010] Collecting dense three-dimensional point cloud data inside the target furnace chamber and uploading to the central processing platform to establish a three-dimensional model of the furnace chamber;

[0011] Collecting the position information of the carbon block grouping to be transported, controlling the clamping mechanism to move above the carbon block grouping and clamping;

[0012] Obtaining the contour information and clamping force information of the carbon block grouping, establishing a three-dimensional model of the carbon block grouping, and calculating the safety envelope of the current pose of the carbon block grouping;

[0013] Based on the safety envelope, the three-dimensional model of the furnace chamber, and the clamping force information, the dynamic safety margin of placing the carbon block grouping into the target furnace chamber under the current clamping state is calculated;

[0014] Based on the dynamic safety margin, a dynamic virtual channel with adaptive cross-sectional size is constructed;

[0015] Based on the dynamic virtual channel constraint, a collision-free motion trajectory is planned for the clamping mechanism;

[0016] Controlling the clamping mechanism to move according to the planned collision-free motion trajectory to place the carbon block grouping at the target position in the furnace chamber.

[0017] Further, the step of establishing a three-dimensional model of the furnace chamber comprises:

[0018] Obtaining three-dimensional point cloud data inside the target furnace chamber;

[0019] Identifying the planar point set representing each surface in the furnace chamber based on the three-dimensional point cloud data using a plane segmentation algorithm;

[0020] calculate size, position and pose parameters of the furnace chamber based on the planar point set;

[0021] generate a three-dimensional geometric model of the furnace chamber based on the size, position and pose parameters of the target furnace chamber.

[0022] Further, the safety envelope is the smallest circumscribed cuboid tightly wrapping the three-dimensional model of the carbon block grouping in the clamped state, and the step of establishing the safety envelope comprises:

[0023] a structured light sensor and a tactile sensor are arranged on the clamping jaw, the structured light sensor is used to collect three-dimensional profile point cloud of the side surface of the carbon block grouping, and the tactile sensor is used to collect the distribution map of the clamping force;

[0024] based on the three-dimensional profile point cloud and the distribution map of the clamping force, a three-dimensional point cloud model of the carbon block grouping in the clamped state is established;

[0025] a coordinate system is established with the end flange plate of the clamping mechanism as the origin and the force arm of the clamping mechanism and the support column as the coordinate axes, and the Z-axis of the coordinate system is parallel to the downward direction of the clamping jaw;

[0026] the point cloud data of the carbon block grouping is input into a principal component analysis algorithm, and the safety envelope of the carbon block grouping is calculated.

[0027] Further, the dynamic safety margin is calculated according to the following formula:

[0028]

[0029] wherein represents a preset reference safety gap, represents pose uncertainty variance of the projection profile of the safety envelope on the horizontal plane, represents the current actual clamping force, represents the maximum allowable clamping force set to prevent damage to the carbon block grouping, and represents a weight coefficient greater than zero.

[0030] Further, the dynamic virtual channel takes the motion path of the safety envelope as the reference, has the same cross-sectional shape as the safety envelope, and has a cross-sectional size equal to the sum of the cross-sectional size of the safety envelope and the dynamic safety margin.

[0031] Further, a pose adjustment strategy is adopted to plan a collision-free motion trajectory, and the pose adjustment strategy comprises:

[0032] for each candidate pose in the preset candidate adjustment poses, a pose adjustment priority score is calculated.

[0033] Adjusting the pose according to the priority score In descending order, each candidate pose is tried for trajectory planning in sequence until planning succeeds or all candidate poses are tried.

[0034] Further, the pose adjustment priority score The formula for calculating is:

[0035]

[0036] Wherein represents the absolute value of the angle of rotation required to adjust from the current pose to the candidate pose represents the horizontal projection area of the safety envelope at the candidate pose represents the pose compatibility coefficient based on historical successful placement data at the candidate pose , and represent weight coefficients greater than zero and satisfy . Compared with the prior art, the present application has the following beneficial effects:

[0037] Compared with the prior art, the present application has the following beneficial effects:

[0038] (1) The method provided by the present application can accurately grasp the real pose of the carbon block grouping in space, and plan a safe and reliable placement trajectory based on dynamic safety evaluation. This fundamentally avoids the damage and scrap of the carbon block green body, and more avoids the structural damage to the expensive furnace body, directly saving a large amount of material cost and equipment maintenance cost, and providing technical support for continuous and stable production of carbon baking workshop.

[0039] (2) The method provided by the present application can realize collision-free placement of carbon block grouping, laying a solid foundation for subsequent uniform heating and stable sintering, and significantly improving product qualification rate. At the same time, the self-adaptive adjustment and high-efficiency planning capability of the system greatly shortens the furnace charging operation cycle, speeds up the kiln turnover, and directly improves the overall production capacity.

[0040] (3) The method provided by the present application enables the hoisting and transportation equipment to completely autonomously complete the whole operation from perception, decision-making to execution, and the operator does not need to perform manual remote control or intervention in the harsh environment of high temperature and high dust, greatly reducing the labor intensity and the risk of human operation errors. BRIEF DESCRIPTION OF DRAWINGS

[0041] ​​In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0042] Figure 1 Flow chart of the carbon block grouping and placing method based on multi-modal perception and dynamic safety evaluation of the present application;

[0043] Figure 2 Trajectory planning logic diagram of the carbon block grouping and placing method based on multi-modal perception and dynamic safety evaluation of the present application. DETAILED DESCRIPTION

[0044] In order to make the person skilled in the art better understand the present application, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0045] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily have to include only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to the process, method, product or device.

[0046] As Figure 1 shown, the present application provides the following technical solutions:

[0047] Collect dense three-dimensional point cloud data inside the target furnace chamber and upload to the central processing platform to establish a three-dimensional model of the furnace chamber;

[0048] Collect the grouping position information of the carbon blocks to be transported, control the clamping mechanism to move above the carbon block grouping and clamp;

[0049] Obtain the profile information and clamping force information of the carbon block group, establish a three-dimensional model of the carbon block group, and calculate a safety envelope of the carbon block group in a current posture;

[0050] Based on the safety envelope, the three-dimensional model of the furnace chamber, and the clamping force information, calculate a dynamic safety margin of placing the carbon block group into a target furnace chamber in a current clamping state;

[0051] Based on the dynamic safety margin, construct a dynamic virtual channel with adaptive cross-sectional size;

[0052] Based on the dynamic virtual channel constraint, plan a collision-free motion trajectory for the clamping mechanism;

[0053] Control the clamping mechanism to move according to the planned collision-free motion trajectory and place the carbon block group at a target position in the furnace chamber.

[0054] In one embodiment, the step of establishing a three-dimensional model of the furnace chamber includes:

[0055] Obtain three-dimensional point cloud data inside the target furnace chamber;

[0056] Based on the three-dimensional point cloud data, identify a planar point set representing each surface in the furnace chamber using a plane segmentation algorithm;

[0057] Based on the planar point set, calculate the size, position, and posture parameters of the furnace chamber;

[0058] Based on the size, position, and posture parameters of the target furnace chamber, generate a three-dimensional geometric model of the furnace chamber.

[0059] In one embodiment, the safety envelope is a minimum circumscribed cuboid tightly wrapping the three-dimensional model of the carbon block group in a clamping state, and the step of establishing the safety envelope includes:

[0060] A structured light sensor and a tactile sensor are arranged on the clamping jaw, the structured light sensor is used to collect three-dimensional profile point cloud of the carbon block group, and the tactile sensor is used to collect the distribution map of the clamping force;

[0061] Based on the three-dimensional profile point cloud and the distribution map of the clamping force, a three-dimensional point cloud model of the carbon block group in the clamping state is established;

[0062] A coordinate system is established with the end flange plate of the clamping mechanism as the origin and the force arm and support column of the clamping mechanism as the coordinate axes, and the Z-axis of the coordinate system is parallel to the downward direction of the clamping jaw;

[0063] The point cloud data of the carbon block group is input into a principal component analysis algorithm to calculate the safety envelope of the carbon block group.

[0064] In one embodiment, the dynamic safety margin The calculation formula is:

[0065]

[0066] in This indicates the preset baseline safety clearance. This represents the variance of the attitude uncertainty of the projected profile of the safety envelope on the horizontal plane. This indicates the current actual clamping force. This indicates the maximum permissible clamping force set to prevent damage to the carbon block assembly. and This represents a weighting coefficient that is greater than zero.

[0067] In one embodiment, the dynamic virtual channel is based on the motion path of the safety envelope, and its channel cross-sectional shape is the same as that of the safety envelope, and the channel cross-sectional size is equal to the sum of the safety envelope cross-sectional size and the dynamic safety margin.

[0068] In one embodiment, an attitude adjustment strategy is employed to plan a collision-free motion trajectory, the attitude adjustment strategy including:

[0069] In the preset candidate adjustment poses, for each candidate pose Calculate attitude adjustment priority score ;

[0070] Scoring based on posture adjustment priority In descending order, trajectory planning is attempted for each candidate pose in turn until the planning is successful or all candidate poses have been tried.

[0071] In one embodiment, the attitude adjustment priority score The calculation formula is:

[0072]

[0073] in This indicates a shift from the current pose to a candidate pose. The absolute value of the required rotation angle, Indicates the candidate posture The horizontal projected area of ​​the safety envelope described below Indicates the candidate posture The following is a posture compatibility coefficient derived from historical successful placement data. , and This represents a weight coefficient that is greater than zero and satisfies... .

[0074] Example

[0075] The implementation process and effects of the application are described by taking a carbon roasting workshop as an example. The data are based on industry measurement standards and engineering research, and can be reproduced and support beneficial effects.

[0076] S1: three-dimensional model of furnace chamber is established

[0077] S11: in the system initialization stage, the operator controls the multifunctional unit, temporarily fixes a high-precision 3D laser scanner on the hoist of the multifunctional unit, controls the multifunctional unit to drive the scanner to scan the interior of the non-operating roasting furnace chamber, and obtains dense three-dimensional point cloud data in the interior.

[0078] S12: in the central processing platform, the point cloud is preprocessed to remove noise, and then a random sample consensus algorithm (RANSAC) is used to segment five plane point sets representing four inner walls and the bottom of the furnace chamber from the point cloud. By calculating the normal vectors, intersection lines and corner points of these planes, the accurate length, width and depth dimensions of the furnace chamber, and the position and attitude (i.e. the direction of the furnace mouth) in the world coordinate system are analyzed.

[0079] S13: based on the size, position and attitude parameters of the target furnace chamber, a cuboid furnace chamber three-dimensional model corresponding to the physical furnace chamber 1:1 is generated and stored.

[0080] S2: initial clamping

[0081] The carbon block grouping (already grouped carbon block) to be put into the roasting furnace chamber is transported to the waiting station. The global visual sensor fixed above the workshop takes a picture of the carbon block grouping and identifies it, determines its two-dimensional position (X, Y) and horizontal rotation angle through image processing. The pose information is sent to the control system of the multifunctional unit, the multifunctional unit is moved to the top of the carbon block grouping, and the clamping jaw is driven to perform the clamping action.

[0082] S3: build a safety envelope

[0083] S31: after the clamping jaw is closed to clamp the carbon block grouping, the multi-line structured light sensor integrated on the inside of the clamping jaw is immediately started, projects light stripes to the exposed side surface of the carbon block grouping, and quickly obtains high-precision three-dimensional profile point cloud of the side surface. At the same time, the tactile sensor array attached to the contact surface of the clamping jaw starts to work, reads the distribution map of the clamping force, and confirms that the clamping center is stable and there is no risk of slipping.

[0084] S32: the processing unit fuses the visual and tactile data to generate a complete real-time three-dimensional point cloud model of the carbon block grouping in the clamping state.

[0085] S33: Based on the three-dimensional point cloud model, the principal direction of the point cloud is calculated in the gripper coordinate system using the principal component analysis (PCA) algorithm, and a minimum circumscribed cuboid that tightly wraps all points, i.e., a safety envelope, is determined. The pose of the safety envelope is completely synchronized with the real pose of the carbon block grouping.

[0086] S4: Calculate dynamic safety margin

[0087] The system calls the pre-stored furnace three-dimensional model and the safety envelope just generated, and according to the formula:

[0088]

[0089] Calculate the dynamic safety margin of this placement . Among them, the reference safety gap is set to 15 mm, the attitude uncertainty calculated according to the real-time point cloud model is 2 mm, the current clamping force is 90% of the rated clamping force . After substituting the weight coefficients , , the of this time is calculated = 15 + 2 + 0.5 = 17.5 mm.

[0090] S5: Generate virtual channel

[0091] The system generates a dynamic virtual channel based on the planned motion path of the safety envelope. The cross-sectional shape of the channel is the same as the horizontal cross-section of the safety envelope, but its size is expanded in both length and width by (i.e., 17.5 mm). Therefore, the channel is a virtual motion constraint space that closely follows the furnace inner wall model but leaves a 17.5 mm safety space.

[0092] S6: Collision-free motion trajectory planning

[0093] Figure 2 As shown in , the path planning algorithm plans a collision-free motion trajectory based on the safety envelope dynamic virtual channel, and plans a collision-free motion trajectory for the multifunctional unit from the current position to the target point at the bottom of the furnace.

[0094] Planning success: directly execute S7.

[0095] Planning failure: due to poor initial attitude, the safety envelope cannot pass through the channel. The system then starts the attitude adjustment strategy, and for each attitude in the pre-set candidate attitude library, calculates the attitude adjustment priority score :

[0096]

[0097] in This indicates a shift from the current pose to a candidate pose. The absolute value of the required rotation angle, Indicates the candidate posture The horizontal projected area of ​​the safety envelope described below Indicates the candidate posture The following is a posture compatibility coefficient derived from historical successful placement data. , and This represents a weight coefficient that is greater than zero and satisfies... .

[0098] The system selects candidate poses by priority. Try descending order and adjust to the next candidate pose. Based on the new attitude, S3-S6 are repeated to update the safety envelope and virtual channel, re-plan the trajectory and determine whether the planning is successful, until all attitudes are tried. If planning still fails after trying all attitudes, an alarm is triggered and manual intervention is awaited.

[0099] S7: Collision-free motion trajectory execution

[0100] The multi-functional unit moves strictly according to the planned collision-free trajectory, driving the carbon blocks to pass safely and smoothly through the furnace opening, and finally accurately place them in the target position inside the baking oven, without any bumps or knocks throughout the process.

[0101] This invention proposes a placement method based on multimodal perception and dynamic safety assessment, which completely eliminates the risk of collision during the grouping and placement of carbon blocks into the baking oven, fundamentally ensuring production safety and equipment integrity. Simultaneously, the non-destructive and precise placement improves the product qualification rate of carbon blocks and achieves fully automated and efficient operation.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for grouping and placing carbon blocks based on multimodal sensing and dynamic security assessment, characterized in that, Includes the following steps: Collect dense 3D point cloud data of the target furnace chamber and upload it to the central processing platform to build a 3D model of the furnace chamber; Collect the location information of the carbon blocks to be transported into a group, and control the clamping mechanism to move above the carbon block group and clamp it. Obtain the contour information and clamping force information of the carbon block assembly, establish a three-dimensional model of the carbon block assembly, and calculate the safety envelope of the carbon block assembly in the current posture. Based on the safety envelope, the three-dimensional model of the furnace chamber, and the clamping force information, the dynamic safety margin for placing the carbon blocks into the target furnace chamber under the current clamping state is calculated. Based on the aforementioned dynamic safety margin, a dynamic virtual channel with adaptive cross-sectional dimensions is constructed; Based on the dynamic virtual channel constraints, a collision-free motion trajectory is planned for the clamping mechanism; The clamping mechanism is controlled to move along a planned collision-free trajectory to group and place the carbon blocks into the target position inside the furnace chamber. The dynamic safety margin The calculation formula is: in This indicates the preset baseline safety clearance. This represents the variance of the attitude uncertainty of the projected profile of the safety envelope on the horizontal plane. This indicates the current actual clamping force. This indicates the maximum permissible clamping force set to prevent damage to the carbon block assembly. and This represents a weighting coefficient that is greater than zero.

2. The carbon block grouping and placement method based on multimodal sensing and dynamic security assessment according to claim 1, characterized in that, The steps for creating a 3D model of the furnace chamber include: Acquire 3D point cloud data of the target furnace chamber; Based on 3D point cloud data, a planar segmentation algorithm is used to identify the set of planar points representing each surface inside the furnace chamber; The dimensions, position, and orientation parameters of the furnace chamber are calculated based on the planar point set; A three-dimensional geometric model of the furnace chamber is generated based on the size, location, and orientation parameters of the target furnace chamber.

3. The carbon block grouping and placement method based on multimodal sensing and dynamic security assessment according to claim 1, characterized in that, The safety envelope is the smallest circumscribed cuboid of the three-dimensional model of the carbon block assembly under tight clamping conditions. The steps for establishing the safety envelope include: A structured light sensor and a tactile sensor are installed on the gripper. The structured light sensor is used to collect the three-dimensional contour point cloud of the side of the carbon block assembly, and the tactile sensor is used to collect the distribution map of the gripper's clamping force. Based on the three-dimensional contour point cloud and the distribution map of clamping force, a three-dimensional point cloud model of carbon block grouping under clamping state is established. A coordinate system is established with the end flange of the clamping mechanism as the origin and the clamping mechanism lever arm and support column as the coordinate axes. The Z-axis of the coordinate system is parallel to the downward direction of the jaws. The point cloud data of the carbon block group is input into the principal component analysis algorithm to calculate the safe envelope of the carbon block group.

4. The carbon block grouping and placement method based on multimodal sensing and dynamic security assessment according to claim 1, characterized in that, The dynamic virtual channel is based on the movement path of the safety envelope, and its cross-sectional shape is the same as that of the safety envelope. The cross-sectional size of the channel is equal to the sum of the cross-sectional size of the safety envelope and the dynamic safety margin.

5. The carbon block grouping and placement method based on multimodal sensing and dynamic security assessment according to claim 1, characterized in that, A collision-free motion trajectory is planned using an attitude adjustment strategy, which includes: In the preset candidate adjustment poses, for each candidate pose Calculate attitude adjustment priority score ; Scoring based on posture adjustment priority In descending order, trajectory planning is attempted for each candidate pose in turn until the planning is successful or all candidate poses have been tried.

6. The carbon block grouping and placement method based on multimodal sensing and dynamic security assessment according to claim 5, characterized in that, The attitude adjustment priority score The calculation formula is: in This indicates a shift from the current pose to a candidate pose. The absolute value of the required rotation angle, Indicates the candidate posture The horizontal projected area of ​​the safety envelope described below Indicates the candidate posture The following is a posture compatibility coefficient derived from historical successful placement data. , and This represents a weight coefficient that is greater than zero and satisfies... .

Images