Method and Device for Acceleration Compensation Control of Coffee Machine Robotic Arm Based on Tilting Prevention and Spill Prevention

Abstract

This invention provides a method and device for acceleration compensation control of a coffee machine robotic arm based on tipping prevention, belonging to the field of robotic arm control technology. After receiving a real-time order, the coffee machine calls a real-time acceleration sequence from an acceleration control library. During the movement of the liquid-pumping robotic arm along a collision-free trajectory while holding the coffee cup, a six-axis force sensor integrated into the gripper measures the liquid's center of gravity offset in real time. Based on this offset, the hysteresis of the real-time acceleration sequence is compensated. This solves the technical problem in existing coffee machine robotic arms where insufficient displacement control stability easily leads to tipping and spillage when transporting coffee beverages containing heterogeneous liquids. It effectively prevents spillage during coffee liquid transportation, significantly improves the transport stability of complex coffee beverages containing ice, foam, etc., ensures safe and efficient multi-robotic arm collaborative operation, and greatly reduces the risk of delays or spills caused by liquid sloshing.

Description

Technical Field

[0001] This invention relates to the field of robotic arm control technology, and in particular to an acceleration compensation control method and device for a coffee machine robotic arm based on tipping prevention and spillage prevention. Background Technology

[0002] Most existing robotic arms for coffee machines are designed for transporting homogeneous liquids (such as pure coffee liquid), and their displacement control strategies typically employ fixed acceleration curves or simple PID feedback regulation.

[0003] However, when transporting heterogeneous liquids such as ice and milk foam, the dynamic inertial abrupt change caused by the density difference inside the liquid can cause violent shaking. Traditional control methods are difficult to match the coupling changes between the liquid centroid shift and the robotic arm's motion state in real time. Especially in multi-robotic arm collaborative operation scenarios, the need for trajectory obstacle avoidance further amplifies the discontinuity of motion and exacerbates the liquid inertial impact.

[0004] Existing technologies attempt to improve stability by reducing the pipetting speed or increasing the stiffness of the mechanical structure, but this significantly sacrifices manufacturing efficiency and fails to resolve the fundamental contradiction in the dynamic response of heterogeneous liquid multiphase flow.

[0005] In addition, feedback control based on a single sensor (such as vision or inertial unit) suffers from response delay and insufficient anti-interference capability, making it difficult to achieve high-precision anti-spillage control under complex working conditions. Summary of the Invention

[0006] This invention provides an acceleration compensation control method and device for a coffee machine robotic arm based on tipping prevention, which addresses the technical problem in the prior art where the displacement control stability of the coffee machine robotic arm is insufficient and prone to tipping and spilling when transporting coffee beverages containing heterogeneous liquids.

[0007] In view of the above problems, the present invention provides a method and device for acceleration compensation control of a coffee machine robotic arm based on tipping prevention and spill prevention.

[0008] The first aspect of the present invention provides an acceleration compensation control method for a coffee machine robotic arm based on spill prevention. The method includes: constructing an acceleration control library by performing a spill prevention test along a non-collision-free dispensing trajectory, wherein the acceleration control library stores N standard coffee drinks and N benchmark acceleration sequences using coded association; after receiving a real-time order, the coffee machine retrieves a real-time acceleration sequence from the acceleration control library by parsing the drink type code of the real-time order; after the real-time order is completed at the coffee machine's dispensing port, the real-time acceleration sequence is used as a feedforward control input to activate the coffee machine's dispensing robotic arm; during the process of the dispensing robotic arm moving along the non-collision-free dispensing trajectory while holding the coffee cup of the order, the liquid center of gravity offset is measured in real time by a six-axis force sensor integrated into the gripper, and the hysteresis swaying of the real-time acceleration sequence is canceled according to the real-time center of gravity offset.

[0009] In one implementation, an acceleration control library is built by performing a tipping and spill prevention test along a collision-free pipetting trajectory. Prior to this, the following processing is also performed:

[0010] The coordinates of the liquid inlet and the capping area are extracted from the coffee machine's workspace; by fitting the obstacle avoidance trajectory of the multi-robotic arm between the coordinates of the liquid inlet and the capping area, the collision-free liquid transfer trajectory of the liquid transfer robotic arm is output.

[0011] In one implementation, an acceleration control library is built by performing a tipping and spill prevention test along a collision-free pipetting trajectory, and the following processing is also performed:

[0012] Through the HMI interface of the coffee machine, information on the N liquid components of the N standard coffee beverages is entered; N sample liquid models are constructed based on the structural parameters of the coffee cup and the information on the N liquid components; using the non-collision trajectory of liquid transfer as a constraint, the N sample liquid models are subjected to iterative tests to prevent spillage in a CFD simulation environment, and N benchmark acceleration sequences are output; the N standard coffee beverages and the N benchmark acceleration sequences are associated and stored to obtain the acceleration control library.

[0013] In one implementation, using the described collision-free pipetting trajectory as a constraint, iterative tests to prevent spillage are performed on the N sample liquid models in a CFD simulation environment, outputting N benchmark acceleration sequences, and the following processing is also performed:

[0014] Discretize the collision-free pipetting trajectory to obtain a time-series acceleration control point; predefine two anti-spillage criteria, which consist of a longitudinal offset threshold for the liquid surface and a lateral offset threshold for the centroid; in a CFD simulation environment, run the pipetting robot along the collision-free pipetting trajectory with an initial acceleration to perform a spill prevention test on a first sample liquid model, obtaining a first robot arm end-effector pose sequence, a first dynamic offset sequence, and a first end-effector acceleration sequence corresponding to the time-series acceleration control point; traverse the first dynamic offset sequence using the anti-spillage criteria to map and locate multiple first end-effector deviation accelerations in the first end-effector acceleration sequence; starting from the multiple first end-effector deviation accelerations, perform an iterative spill prevention test on the first sample liquid model until a first reference acceleration sequence that satisfies the anti-spillage criteria is output.

[0015] In one implementation, starting with the plurality of first end deviation accelerations, the first sample liquid model is subjected to an iterative test to prevent spillage until a first benchmark acceleration sequence that satisfies the two spillage prevention criteria is output. The following processing is then performed:

[0016] When W dynamic offsets in the first candidate acceleration sequence output by the tipping and spill prevention iterative test stably deviate from the anti-spillage dual criteria, W first robotic arm pose deviation points corresponding to the W dynamic offsets are called from the first robotic arm end pose sequence; robotic arm pose adaptive compensation is performed at the W first robotic arm pose deviation points to output a first reference pose sequence; the first candidate acceleration sequence and the first reference pose sequence are spatially fused to obtain the first reference acceleration sequence.

[0017] In one implementation, as the pipetting robot arm moves along the collision-free pipetting trajectory while holding the ordered coffee cup, it measures the liquid center of gravity offset in real time using a six-axis force sensor integrated into the gripper. Based on the real-time center of gravity offset, it performs hysteresis cancellation of the real-time acceleration sequence. Prior to this, the following processing is also performed:

[0018] A predefined liquid center of gravity offset feature is defined. During the process of using the first reference acceleration sequence to drive the pipetting robot arm to perform spill prevention verification on the first sample liquid model, the liquid center of gravity offset feature is added as a perturbation term at the time-series acceleration control point for perturbation testing, and a first perturbation offset sequence is output. The first reference pose sequence is compensated based on the first perturbation offset sequence, and a first compensated pose sequence is output. Similarly, N compensated pose sequences of the N reference acceleration sequences are constructed. The N standard coffee drinks and the N compensated pose sequences are stored using encoding association to obtain a pose compensation library.

[0019] In one implementation, as the pipetting robot arm moves along the collision-free pipetting trajectory while holding the ordered coffee cup, it measures the liquid center of gravity offset in real time using a six-axis force sensor integrated into the gripper, and performs hysteresis cancellation of the real-time acceleration sequence based on the real-time center of gravity offset, and also performs the following processing:

[0020] Based on the beverage type code of the real-time order, a real-time compensation sequence is retrieved from the pose compensation library; during the process of the pipetting robot arm moving along the collision-free pipetting trajectory while holding the coffee cup of the order, the liquid center of gravity offset is measured in real time by a six-axis force sensor integrated into the gripper, and the real-time pose compensation feature is retrieved from the real-time compensation sequence based on the acceleration control point where the real-time center of gravity offset is located; the real-time pose compensation feature is used to drive the pipetting robot arm to cancel the hysteresis of the real-time acceleration sequence.

[0021] A second aspect of the present invention provides an acceleration compensation control device for a coffee machine robotic arm based on spill prevention. The device includes: a data library building unit for constructing an acceleration control library by performing a spill prevention test along a non-collision-free dispensing trajectory, wherein the acceleration control library stores N standard coffee drinks and N benchmark acceleration sequences using coded association; a data retrieval unit for retrieving a real-time acceleration sequence from the acceleration control library after the coffee machine receives a real-time order by parsing the drink type code of the real-time order; a device activation unit for activating the coffee machine's dispensing robotic arm by using the real-time acceleration sequence as a feedforward control input after the real-time order is completed at the coffee machine's dispensing port; and an offset compensation unit for measuring the liquid center of gravity offset in real-time using a six-axis force sensor integrated in the gripper during the movement of the dispensing robotic arm holding the coffee cup along the non-collision-free dispensing trajectory, and compensating for the hysteresis of the real-time acceleration sequence based on the real-time center of gravity offset.

[0022] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0023] The method provided in this invention constructs an acceleration control library by performing a spill prevention test along a collision-free liquid transfer trajectory. This library stores N standard coffee beverages and N benchmark acceleration sequences using a coded association. After receiving a real-time order, the coffee machine parses the beverage type code of the order and retrieves the real-time acceleration sequence from the acceleration control library. After the real-time order is processed at the coffee machine's dispensing port, the real-time acceleration sequence is used as a feedforward control input to activate the coffee machine's liquid transfer robotic arm. During the movement of the liquid transfer robotic arm along the collision-free trajectory while holding the coffee cup, a six-axis force sensor integrated into the gripper measures the liquid's center of gravity offset in real time and uses this offset to compensate for the hysteresis caused by the real-time acceleration sequence. This method effectively prevents spillage during coffee liquid transportation, significantly improves the transport stability of complex coffee beverages containing ice or foam, ensures the safe and efficient operation of multiple robotic arms, and greatly reduces the risk of delays or spills caused by liquid sloshing. Attached Figure Description

[0024] Figure 1 This invention provides a schematic flowchart of the acceleration compensation control method for a coffee machine robotic arm based on tipping prevention.

[0025] Figure 2 A schematic diagram of the acceleration compensation control device for a coffee machine robotic arm based on tipping prevention provided by the present invention is shown.

[0026] Explanation of reference numerals in the attached diagram: Data database creation unit 1, data retrieval unit 2, device activation unit 3, offset compensation unit 4. Detailed Implementation

[0027] This invention provides an acceleration compensation control method and device for a coffee machine robotic arm based on tipping prevention, which addresses the technical problem in the prior art where the displacement control stability of the coffee machine robotic arm is insufficient and prone to tipping and spilling when transporting coffee beverages containing heterogeneous liquids.

[0028] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0029] Example 1: A flowchart of the acceleration compensation control method for a coffee machine robotic arm based on tipping prevention provided in this embodiment of the invention. (See attached diagram) Figure 1 The method includes:

[0030] Step S100: By performing a spill prevention test along the non-collision trajectory of liquid transfer, an acceleration control library is constructed, wherein the acceleration control library stores N standard coffee drinks and N benchmark acceleration sequences in an coded association.

[0031] In one implementation, an acceleration control library is constructed by performing a spill prevention test along a collision-free pipetting trajectory. Prior to this, step S100 also includes:

[0032] Step S100-1: Extract the coordinates of the liquid inlet and the lid-closing area from the coffee machine's working space.

[0033] Step S100-2: By fitting the obstacle avoidance trajectory of the multi-robotic arm between the coordinates of the liquid receiving port and the coordinates of the capping area, the collision-free liquid transfer trajectory of the liquid transfer robotic arm is output.

[0034] It should be understood that the spillage of coffee from the cup occurs during the process of the liquid transfer robot arm moving the coffee cup from the liquid inlet to the lid-closing area by gripping the cup with its jaws. Therefore, in this embodiment, the spatial coordinates of the liquid inlet and the lid-closing area are accurately located within the three-dimensional workspace of the coffee machine.

[0035] The coordinates of the liquid inlet are determined by a vision sensor to locate the cup placement area or by preset mechanical structure parameters. These coordinates are used to ensure the accuracy of the starting point position for the robotic arm to grasp the coffee cup.

[0036] The coordinates of the capping area must avoid the movement range of other mechanical components (such as steam nozzles and cup cap storage racks) and take into account the space margin required for the cup capping operation. Obtaining the coordinates of the liquid inlet and the capping area provides an accurate physical spatial reference for subsequent path planning.

[0037] Based on the coordinates of the liquid inlet and the capping area, a collision-free motion trajectory for the liquid handling robot is generated using a multi-robot collaborative obstacle avoidance algorithm. It should be understood that this trajectory fitting requires real-time calculation of the dynamic working range of other robots (such as the capping robot and the ice-retrieving robot) and the construction of a 3D obstacle point cloud model to ensure that there is no physical interference throughout the liquid handling path.

[0038] The final liquid transfer collision-free trajectory, representing the displacement spatial path of the liquid transfer robot in space, serves as the kinematic basis for subsequent acceleration optimization.

[0039] In one implementation, an acceleration control library is constructed by performing a spill prevention test along a non-collision transfer trajectory. The method step S100 provided by this invention includes:

[0040] Step S110: Enter the information of the N liquid components of the N standard coffee drinks through the HMI interface of the coffee machine.

[0041] Step S120: Construct N sample liquid models based on the structural parameters of the coffee cup and the information of the N liquid components.

[0042] Step S130: Using the described non-collision pipetting trajectory as a constraint, perform iterative tests on the N sample liquid models in a CFD simulation environment to prevent spillage, and output N benchmark acceleration sequences.

[0043] Step S140: Associate and store the N standard coffee drinks and N benchmark acceleration sequences to obtain the acceleration control library.

[0044] Specifically, through the human-computer interaction interface (such as a touch screen or mobile application) of the coffee machine, the information of N liquid components of N standard coffee drinks is pre-input. The liquid component information includes the ratio of basic liquids such as water, coffee concentrate, and milk, as well as special attributes such as the amount of ice added and the thickness of the foam layer. This constitutes the core dataset of the beverage feature database, providing a raw material characteristic benchmark for subsequent liquid modeling.

[0045] By combining the preset physical parameters of the coffee cup (height, diameter, material stiffness) with the entered information of the N liquid components, a unique liquid dynamic behavior model is established for N standard coffee drinks, resulting in the N sample liquid models.

[0046] The sample liquid model quantifies the viscosity change, inertial distribution, and free surface fluctuation characteristics of the corresponding coffee beverage in the coffee cup during motion, such as the heterogeneous mass distribution effect of ice-containing beverages or the viscoelastic damping characteristics of the milk foam layer, which constitute the physical basis for simulation testing.

[0047] Using a pre-generated collision-free liquid transfer trajectory as a fixed path constraint, the spill-proof performance of N sample liquid models is tested in a computational fluid dynamics simulation environment. By iteratively adjusting the acceleration values ​​of each trajectory point of the robotic arm (e.g., reducing the peak acceleration during sharp turns), the changes in liquid sloshing amplitude are observed until the liquid surface fluctuation and center of mass offset at all test points are below the safety threshold. This generates the optimal acceleration control sequence for the corresponding beverage, and outputs N benchmark acceleration sequences. The technical implementation of the spill-proof iterative test is described in detail in the subsequent specification.

[0048] A bidirectional index relationship is established between the optimized N standard coffee beverages and N benchmark acceleration sequences (such as "Americano" and "Iced Latte"), and stored in the non-volatile memory of the coffee machine control system as the acceleration control library. The acceleration control library supports fast retrieval of matching benchmark acceleration sequences (acceleration curves) through beverage codes, providing a feedforward control benchmark for real-time liquid handling operations, while also allowing dynamic expansion with subsequent additions of new beverage types.

[0049] In one implementation, using the non-collision-free pipetting trajectory as a constraint, iterative tests to prevent spillage are performed on the N sample liquid models in a CFD simulation environment, outputting N benchmark acceleration sequences. Step S130 of the method provided by this invention includes:

[0050] Step S131: Discretize the collision-free pipetting trajectory to obtain the time-series acceleration control point.

[0051] Step S132: Predefine two anti-spillage criteria, wherein the two anti-spillage criteria consist of a longitudinal displacement threshold of the liquid surface and a lateral displacement threshold of the centroid.

[0052] Step S133: In the CFD simulation environment, the pipetting robot arm is run along the pipetting collision-free trajectory with initial acceleration to perform a tipping and spill prevention test on the first sample liquid model, and the first robot arm end pose sequence, first dynamic offset sequence and first end acceleration sequence corresponding to the time-series acceleration control point are obtained.

[0053] Step S134: By using the anti-spillage dual criteria to traverse the first dynamic offset sequence, multiple first end deviation accelerations are mapped and located in the first end acceleration sequence.

[0054] Step S135: Starting from the plurality of first end deviation accelerations, perform iterative tests on the first sample liquid model to prevent spillage until the first benchmark acceleration sequence that satisfies the two anti-spillage criteria is output.

[0055] In one implementation, starting with the plurality of first end deviation accelerations, an iterative test to prevent spillage is performed on the first sample liquid model until a first benchmark acceleration sequence that satisfies the two spillage prevention criteria is output. Step S135 of the method provided by this invention includes:

[0056] Step S1351: When W dynamic offsets in the first alternative acceleration sequence output by the tipping and spill prevention iterative test stably deviate from the anti-spillage dual criteria, call the W first robotic arm pose deviation points corresponding to the W dynamic offsets from the first robotic arm end pose sequence.

[0057] Step S1352: Perform adaptive compensation for the robot arm pose at the W first robot arm pose deviation points and output the first reference pose sequence.

[0058] Step S1353: Spatial fusion of the first candidate acceleration sequence and the first reference pose sequence to obtain the first reference acceleration sequence.

[0059] Specifically, in this embodiment, the continuous collision-free movement path is divided into multiple discrete acceleration control points. Each point corresponds to the position of the robotic arm at different times, which is used to determine the acceleration control parameters for the corresponding position.

[0060] For example, a path that takes 5 seconds can be discretized into 500 time points, each spaced 10 milliseconds apart, forming a refined set of timing instructions. Discretization allows simulation tests to analyze the dynamic response of the liquid frame by frame, accurately matching the robotic arm's motion with the liquid's inertial delay.

[0061] A dual spill prevention criterion is established, consisting of a longitudinal offset threshold and a lateral offset threshold. For example, the longitudinal offset threshold is set to ensure that the liquid level does not exceed 15% below the cup rim (to prevent overflow), and the lateral offset threshold is set to ensure that the lateral offset of the center of gravity does not exceed 25% of the cup rim radius (to prevent tipping).

[0062] These two thresholds combine the safety boundary of liquid volume with the requirements for motion stability; exceeding either threshold triggers the acceleration control optimization process at the trajectory discrete points. Furthermore, the criterion value is dynamically adjusted based on the cup's size; for example, the lateral offset threshold for a cup with an 8cm diameter is set at 2cm.

[0063] In the first CFD simulation, a first sample liquid model of a first standard coffee beverage (such as a standard Americano) is run, and the robotic arm moves along the said liquid transfer collision-free trajectory at the initial theoretical acceleration.

[0064] The simulation records the actual pose (e.g., end gripper tilted 3°), liquid centroid offset (e.g., left offset 5 mm), and real acceleration fluctuation data (e.g., peak overshoot 20%) of multiple discrete acceleration control points in the time-series acceleration control points, and outputs the first robotic arm end pose sequence, the first dynamic offset sequence, and the first end acceleration sequence.

[0065] The first robotic arm end-effector pose sequence, the first dynamic offset sequence, and the first end-effector acceleration sequence expose potential defects in the initial trajectory, such as rapid acceleration causing the frontal surface of the liquid to overshoot and touch the criterion limit.

[0066] By traversing the first dynamic offset sequence using the aforementioned anti-spillage dual criteria, multiple first end-deviation accelerations are mapped and located in the first end-acceleration sequence. For example, at the Mth discrete acceleration control point on the path, the longitudinal fluctuation of the liquid level reaches 18% of the cup height, exceeding the 15% threshold, and the acceleration value at this point is marked as needing optimization. At the Wth discrete acceleration control point, the centroid lateral deviation is found to be excessive (28% offset from the cup rim radius), and the corresponding acceleration command is located as the target to be corrected.

[0067] Finally, multiple first-end deviation accelerations are obtained corresponding to multiple deviation acceleration points that require acceleration optimization.

[0068] At the plurality of deviation acceleration points, using the plurality of first terminal deviation accelerations as the starting point for parameter tuning, the acceleration parameter values ​​at these multiple deviation acceleration points are iteratively adjusted using a gradient descent algorithm. For example, at a certain deviation acceleration point, the peak acceleration value is adjusted from 2.5 m / s². 2 Reduced to 2.0 m / s 2 A smooth transition section was inserted. After 5 iterations of testing, the liquid level fluctuation stabilized within 12%, and the centroid offset converged to 22%, generating the first benchmark acceleration sequence that finally satisfies the aforementioned anti-spillage dual criteria.

[0069] Meanwhile, there is an extreme case: if adjusting the acceleration at the deviation point alone still fails to achieve the first reference acceleration sequence that satisfies the two anti-spillage criteria, this embodiment considers adjusting the gripping posture of the gripper at the end of the pipetting robot arm to compensate for the anti-spillage that the acceleration adjustment failed to achieve.

[0070] Specifically, when the liquid offset of some acceleration control points still fails to meet the anti-spillage criterion after multiple iterations of adjusting the acceleration parameters, it is necessary to locate the corresponding robotic arm spatial pose points of these acceleration control points.

[0071] For example, excessive liquid surface fluctuations at W acceleration control points indicate that acceleration adjustment alone cannot completely suppress the swaying. In this case, it is necessary to backtrack the simulation data and extract W first robotic arm pose deviation points corresponding to the W acceleration control points. These robotic arm pose deviation points include the position and attitude parameters of the robotic arm end effector (such as gripper height and tilt angle), which are used as intervention targets for pose compensation. The W first robotic arm pose deviation points are used for coupled analysis based on liquid dynamic response and robotic arm kinematic parameters, providing a spatial correction benchmark for subsequent compensation.

[0072] At the W first robotic arm pose deviation points, adaptive compensation design for robotic arm pose is implemented. For example, at the pose deviation point where there is a violent forward surge in the liquid surface, the gripper lifting height is increased by 2-3 mm to reduce the inertial impact of the liquid. At the pose deviation point where the center of mass is severely lateralized, the gripper tilt angle is adjusted to reduce the centrifugal effect.

[0073] The compensated pose sequence needs to be re-verified for its effect on suppressing liquid displacement. By fine-tuning the parameters, the corrected pose data can offset the residual shaking that was not resolved by the acceleration adjustment, and a set of reference pose instructions adapted to the characteristics of the current beverage is generated, which constitutes the first reference pose sequence. The first reference pose sequence includes W reference pose adjustment instructions for the W first robotic arm pose deviation points.

[0074] Based on the W first robotic arm pose deviation points, the first reference pose sequence is spatially fused into the first candidate acceleration sequence to obtain the first reference acceleration sequence.

[0075] The pose-acceleration fusion adjustment in this embodiment breaks through the limitations of single parameter adjustment. By co-optimizing the dynamics and kinematic behavior of the robotic arm, it ultimately outputs the first reference acceleration sequence of the first standard coffee beverage that meets the spill prevention requirements, ensuring the stability and efficiency of the first standard coffee beverage during transportation.

[0076] Similarly, using the same method, the N standard coffee drinks and N benchmark acceleration sequences are constructed, and the data configuration of the acceleration control library is completed by storing the N standard coffee drinks and N benchmark acceleration sequences in association.

[0077] This embodiment constructs a multi-beverage acceleration control library through offline simulation, deeply coupling the dynamic characteristics of liquids with the kinematics of the robotic arm to achieve precise spill prevention control across beverage types, improve the stability and efficiency of liquid transfer, and at the same time reduce the real-time computing load, ensuring the technical effect of versatility and reliability under complex working conditions.

[0078] Step S200: After receiving a real-time order, the coffee machine retrieves the real-time acceleration sequence from the acceleration control library by parsing the beverage type code of the real-time order.

[0079] Step S300: After the real-time order is completed at the coffee machine's liquid inlet, the real-time acceleration sequence is used as a feedforward control input to activate the coffee machine's liquid transfer robotic arm.

[0080] Specifically, after a user places an order, the coffee machine parses the beverage type identifier code (such as "Latte01") embedded in the order and quickly retrieves a matching reference acceleration sequence from a pre-built acceleration control library as the real-time acceleration sequence.

[0081] The matching process utilizes an encoding mapping mechanism to accurately associate beverage characteristics with motion control parameters, ensuring that the pipetting robot can instantly call motion commands that are adapted to the physical properties of the current beverage, thus avoiding spill prevention failures caused by differences in liquid properties.

[0082] After the beverage is prepared at the receiving port (such as coffee extraction or milk frothing), the retrieved real-time acceleration sequence is used as a feedforward control signal and loaded into the control system of the pipetting robot.

[0083] By directly injecting offline optimized motion parameters into the real-time control link, the robotic arm is driven to start running according to the preset anti-spillage acceleration curve, bypassing the delay problem of not having time for online calculation, and ensuring high-precision trajectory tracking and liquid stability in the initial stage of transportation.

[0084] Step S400: During the process of the liquid pipetting robot arm moving along the liquid pipetting collision-free trajectory while holding the order coffee cup, the liquid center of gravity offset is measured in real time by a six-axis force sensor integrated in the gripper, and the hysteresis sway is canceled by the real-time acceleration sequence based on the real-time center of gravity offset.

[0085] In one implementation, as the pipetting robot arm moves along the non-collision pipetting trajectory while holding the ordered coffee cup, it measures the liquid center of gravity offset in real time using a six-axis force sensor integrated into the gripper, and cancels the hysteresis of the real-time acceleration sequence based on the real-time center of gravity offset. Previously, the method step S400 provided by this invention also included:

[0086] Step S400-1: Predefine liquid center of gravity offset features.

[0087] Step S400-2: During the process of using the first reference acceleration sequence to drive the pipetting robot arm to perform the spill prevention verification of the first sample liquid model, the liquid center of gravity offset feature is added as a disturbance term at the time-series acceleration control point for disturbance testing, and the first disturbance offset sequence is output.

[0088] Step S400-3: Compensate the first reference pose sequence according to the first perturbation offset sequence, and output the first compensated pose sequence.

[0089] Step S400-4: By analogy, construct N compensated pose sequences for the N reference acceleration sequences.

[0090] Step S400-5: Use encoding association to store the N standard coffee drinks and N compensated pose sequences to obtain a pose compensation library.

[0091] In one implementation, as the pipetting robot arm moves along the non-collision pipetting trajectory while holding an order coffee cup, it measures the liquid center of gravity offset in real time using a six-axis force sensor integrated into the gripper, and cancels the hysteresis of the real-time acceleration sequence based on the real-time center of gravity offset. Step S400 of the method provided by this invention includes:

[0092] Step S410: Based on the beverage type code of the real-time order, retrieve the real-time compensation sequence from the pose compensation library.

[0093] Step S420: During the process of the liquid pipetting robot arm moving along the liquid pipetting collision-free trajectory while holding the order coffee cup, the liquid center of gravity offset is measured in real time by a six-axis force sensor integrated in the gripper, and the real-time pose compensation feature is called from the real-time compensation sequence according to the acceleration control point where the real-time center of gravity offset is located.

[0094] Step S430: The real-time pose compensation feature is used to drive the pipetting robot arm to cancel the hysteresis of the real-time acceleration sequence.

[0095] Specifically, typical center of gravity shift patterns that may occur during liquid transportation are predefined, such as liquid surface hysteresis, lateral centrifugal shift, or random sloshing characteristics caused by cup shaking. These predefined shift characteristics are derived from fluid dynamics principles and historical experimental data, providing a standardized set of disturbance scenarios for subsequent perturbation tests, covering common instability factors.

[0096] During the process of using the first reference acceleration sequence to drive the pipetting robot to perform spill prevention verification on the first sample liquid model, the liquid center of gravity offset feature is added as a disturbance term at the time-series acceleration control point for disturbance testing, and the first disturbance offset sequence is output. The first disturbance offset sequence is the dynamic response data (such as the maximum center of gravity offset and recovery time) of the liquid at multiple (all) time-series acceleration control points under the disturbance.

[0097] Based on the first disturbance offset sequence exposed in the disturbance test, a dynamic pose compensation design is performed on the first reference pose sequence for each acceleration control point to obtain multiple compensated pose parameters for each acceleration control point under various center of gravity offset modes, which constitute the first compensated pose sequence.

[0098] Similarly, N compensated pose sequences are constructed from the N reference acceleration sequences. The N standard coffee drinks and the N compensated pose sequences are stored using an coded association to obtain a pose compensation library.

[0099] Based on this, according to the beverage type code of the real-time order, a real-time compensation sequence is called from the pose compensation library. The real-time compensation sequence includes multiple compensation pose parameters for each of the time-series acceleration control points in various center of gravity offset modes.

[0100] During the process of the liquid-pipette robotic arm moving along the liquid-pipette collision-free trajectory while holding the coffee cup, the liquid center of gravity offset is measured in real time by a six-axis force sensor integrated into the gripper. When the center of gravity offset does not meet the anti-spillage dual criteria mentioned above, the system locates and calls multiple compensation pose parameters of the point in various center of gravity offset modes in the real-time compensation sequence based on the acceleration control point where the real-time center of gravity offset is located.

[0101] The real-time center of gravity offset is compared to see which of the various center of gravity offset modes it falls into, so as to extract the corresponding compensation pose parameters as the real-time pose compensation features, and the real-time pose compensation features are used to drive the pipetting robot to cancel the hysteresis of the real-time acceleration sequence.

[0102] This embodiment achieves the technical effect of effectively preventing spillage during the transportation of liquid coffee, significantly improving the transportation stability of complex coffee beverages containing ice, foam, etc., ensuring the safe and efficient collaborative operation of multiple robotic arms, and greatly reducing the risk of delays or spillage caused by liquid sloshing.

[0103] Example 2 is based on the same inventive concept as the acceleration compensation control method for the tilt-prevention coffee machine robotic arm in the previous examples, such as... Figure 2 As shown, the present invention provides an acceleration compensation control device for a coffee machine robotic arm based on tipping prevention, wherein the device includes:

[0104] Data library building unit 1 is used to build an acceleration control library by performing a spill prevention test along a non-collision trajectory of liquid transfer. The acceleration control library uses coded association to store N standard coffee drinks and N benchmark acceleration sequences.

[0105] Data retrieval unit 2 is used by the coffee machine to retrieve the real-time acceleration sequence from the acceleration control library after receiving a real-time order by parsing the beverage type code of the real-time order;

[0106] The device activation unit 3 is used to activate the liquid transfer robot arm of the coffee machine by using the real-time acceleration sequence as a feedforward control input after the real-time order is completed at the liquid inlet of the coffee machine.

[0107] The offset compensation unit 4 is used to measure the liquid center of gravity offset in real time by a six-axis force sensor integrated in the gripper during the process of the liquid pipetting robot arm moving along the liquid pipetting collision-free trajectory while holding the order coffee cup, and to cancel the hysteresis of the real-time acceleration sequence based on the real-time center of gravity offset.

[0108] In one implementation, the database building unit 1 is further used for:

[0109] The coordinates of the liquid inlet and the capping area are extracted from the coffee machine's workspace; by fitting the obstacle avoidance trajectory of the multi-robotic arm between the coordinates of the liquid inlet and the capping area, the collision-free liquid transfer trajectory of the liquid transfer robotic arm is output.

[0110] In one implementation, the database building unit 1 is further used for:

[0111] Through the HMI interface of the coffee machine, information on the N liquid components of the N standard coffee beverages is entered; N sample liquid models are constructed based on the structural parameters of the coffee cup and the information on the N liquid components; using the non-collision trajectory of liquid transfer as a constraint, the N sample liquid models are subjected to iterative tests to prevent spillage in a CFD simulation environment, and N benchmark acceleration sequences are output; the N standard coffee beverages and the N benchmark acceleration sequences are associated and stored to obtain the acceleration control library.

[0112] In one implementation, the database building unit 1 is further used for:

[0113] Discretize the collision-free pipetting trajectory to obtain a time-series acceleration control point; predefine two anti-spillage criteria, which consist of a longitudinal offset threshold for the liquid surface and a lateral offset threshold for the centroid; in a CFD simulation environment, run the pipetting robot along the collision-free pipetting trajectory with an initial acceleration to perform a spill prevention test on a first sample liquid model, obtaining a first robot arm end-effector pose sequence, a first dynamic offset sequence, and a first end-effector acceleration sequence corresponding to the time-series acceleration control point; traverse the first dynamic offset sequence using the anti-spillage criteria to map and locate multiple first end-effector deviation accelerations in the first end-effector acceleration sequence; starting from the multiple first end-effector deviation accelerations, perform an iterative spill prevention test on the first sample liquid model until a first reference acceleration sequence that satisfies the anti-spillage criteria is output.

[0114] In one implementation, the database building unit 1 is further used for:

[0115] When W dynamic offsets in the first candidate acceleration sequence output by the tipping and spill prevention iterative test stably deviate from the anti-spillage dual criteria, W first robotic arm pose deviation points corresponding to the W dynamic offsets are called from the first robotic arm end pose sequence; robotic arm pose adaptive compensation is performed at the W first robotic arm pose deviation points to output a first reference pose sequence; the first candidate acceleration sequence and the first reference pose sequence are spatially fused to obtain the first reference acceleration sequence.

[0116] In one implementation, the offset compensation unit 4 is further configured to:

[0117] A predefined liquid center of gravity offset feature is defined. During the process of using the first reference acceleration sequence to drive the pipetting robot arm to perform spill prevention verification on the first sample liquid model, the liquid center of gravity offset feature is added as a perturbation term at the time-series acceleration control point for perturbation testing, and a first perturbation offset sequence is output. The first reference pose sequence is compensated based on the first perturbation offset sequence, and a first compensated pose sequence is output. Similarly, N compensated pose sequences of the N reference acceleration sequences are constructed. The N standard coffee drinks and the N compensated pose sequences are stored using encoding association to obtain a pose compensation library.

[0118] In one implementation, the offset compensation unit 4 is further configured to:

[0119] Based on the beverage type code of the real-time order, a real-time compensation sequence is retrieved from the pose compensation library; during the process of the pipetting robot arm moving along the collision-free pipetting trajectory while holding the coffee cup of the order, the liquid center of gravity offset is measured in real time by a six-axis force sensor integrated into the gripper, and the real-time pose compensation feature is retrieved from the real-time compensation sequence based on the acceleration control point where the real-time center of gravity offset is located; the real-time pose compensation feature is used to drive the pipetting robot arm to cancel the hysteresis of the real-time acceleration sequence.

[0120] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An acceleration compensation control method for a coffee machine robotic arm based on tipping prevention, characterized in that, include: An acceleration control library was constructed by performing a spill prevention test along a non-collision pipetting trajectory. The acceleration control library stores N standard coffee drinks and N benchmark acceleration sequences in an coded association. After receiving a real-time order, the coffee machine retrieves a real-time acceleration sequence from the acceleration control library by parsing the beverage type code of the real-time order. After the real-time order is completed at the coffee machine's liquid inlet, the real-time acceleration sequence is used as a feedforward control input to activate the coffee machine's liquid-pumping robotic arm. As the liquid-pipette robotic arm moves along the non-collision liquid-pipette trajectory while holding an order coffee cup, it measures the liquid center of gravity offset in real time using a six-axis force sensor integrated into the gripper, and cancels the hysteresis of the real-time acceleration sequence based on the real-time center of gravity offset.

2. The acceleration compensation control method for a coffee machine robotic arm based on tipping prevention as described in claim 1, characterized in that, An acceleration control library was built by performing a tipping and spill prevention test along a collision-free pipetting trajectory. Previously, this also included: Extract the coordinates of the liquid inlet and the lid-closing area from the coffee machine's workspace; By fitting the obstacle avoidance trajectory of the multi-robot arm between the coordinates of the liquid inlet and the coordinates of the capping area, the collision-free liquid transfer trajectory of the liquid transfer robot arm is output.

3. The acceleration compensation control method for a coffee machine robotic arm based on tipping prevention as described in claim 2, characterized in that, An acceleration control library was constructed by performing a tipping and spill prevention test along a collision-free pipetting trajectory, including: The N liquid component information of the N standard coffee drinks is entered through the HMI interface of the coffee machine. Based on the structural parameters of the coffee cup and the information of the N liquid components, construct N sample liquid models; Using the described non-collision pipetting trajectory as a constraint, the N sample liquid models are subjected to iterative tests to prevent spillage in a CFD simulation environment, and N benchmark acceleration sequences are output. The N standard coffee drinks and N benchmark acceleration sequences are associated and stored to obtain the acceleration control library.

4. The acceleration compensation control method for a coffee machine robotic arm based on tipping prevention as described in claim 3, characterized in that, Using the described collision-free pipetting trajectory as a constraint, iterative tests to prevent spillage were conducted on the N sample liquid models in a CFD simulation environment, outputting N benchmark acceleration sequences, including: Discretize the collision-free pipetting trajectory to obtain the time-series acceleration control points; A predefined anti-spillage criterion is provided, wherein the anti-spillage criterion is composed of a longitudinal displacement threshold of the liquid surface and a lateral displacement threshold of the centroid; In a CFD simulation environment, the manipulator is run with initial acceleration along the non-collision trajectory of the manipulator to perform a tipping and spill prevention test on the first sample liquid model, and the first manipulator end pose sequence, first dynamic offset sequence and first end acceleration sequence corresponding to the time-series acceleration control point are obtained. By using the anti-spray dual criteria to traverse the first dynamic offset sequence, multiple first end deviation accelerations are mapped and located in the first end acceleration sequence. Starting from the plurality of first end deviation accelerations, the first sample liquid model is subjected to iterative tests to prevent spillage until the first benchmark acceleration sequence that satisfies the two anti-spillage criteria is output.

5. The acceleration compensation control method for a coffee machine robotic arm based on tipping prevention as described in claim 4, characterized in that, Starting with the plurality of first end deviation accelerations, the first sample liquid model is subjected to iterative tests to prevent spillage until a first benchmark acceleration sequence that satisfies the two spillage prevention criteria is output, including: When W dynamic offsets in the first alternative acceleration sequence output by the tipping and spill prevention iterative test stably deviate from the anti-spillage dual criteria, the W first robotic arm pose deviation points corresponding to the W dynamic offsets are called from the first robotic arm end pose sequence. Perform adaptive compensation of robot arm pose at the W first robot arm pose deviation points and output the first reference pose sequence; The first candidate acceleration sequence and the first reference pose sequence are spatially fused to obtain the first reference acceleration sequence.

6. The acceleration compensation control method for a coffee machine robotic arm based on tilt prevention as described in claim 5, characterized in that, As the robotic arm moves along the collision-free dispensing trajectory while holding the ordered coffee cup, it measures the liquid center of gravity offset in real time using a six-axis force sensor integrated into the gripper. Based on this real-time offset, it compensates for the hysteresis of the real-time acceleration sequence. Prior to this, the procedure also includes: Predefined liquid center of gravity offset features; During the process of using the first reference acceleration sequence to drive the pipetting robot arm to perform the spill prevention verification of the first sample liquid model, the liquid center of gravity offset feature is added as a disturbance term at the time-series acceleration control point for disturbance testing, and the first disturbance offset sequence is output. The first reference pose sequence is compensated based on the first perturbation offset sequence, and the first compensated pose sequence is output. Similarly, N compensated pose sequences are constructed from the N reference acceleration sequences; The N standard coffee drinks and N compensated pose sequences are stored using an coded association method to obtain a pose compensation library.

7. The acceleration compensation control method for a coffee machine robotic arm based on tipping prevention as described in claim 6, characterized in that, As the robotic arm moves along the collision-free dispensing trajectory while holding an ordered coffee cup, it measures the liquid's center of gravity offset in real time using a six-axis force sensor integrated into the gripper. Based on this real-time offset, it compensates for the hysteresis of the real-time acceleration sequence, including: Based on the beverage type code of the real-time order, the real-time compensation sequence is retrieved from the pose compensation library; During the process of the liquid-pipette robotic arm moving along the liquid-pipette collision-free trajectory while holding the coffee cup of the order, the liquid center of gravity offset is measured in real time by a six-axis force sensor integrated in the gripper, and the real-time pose compensation feature is called from the real-time compensation sequence based on the acceleration control point where the real-time center of gravity offset is located. The real-time pose compensation feature is used to drive the pipetting robot to cancel the hysteresis of the real-time acceleration sequence.

8. An acceleration compensation control device for a coffee machine robotic arm based on tipping prevention and spill prevention, characterized in that, The steps for implementing the method according to any one of claims 1 to 7 include: The data building unit is used to construct an acceleration control library by performing a spill prevention test along a non-collision-free pipetting trajectory. The acceleration control library stores N standard coffee drinks and N benchmark acceleration sequences using coded association. The data retrieval unit is used by the coffee machine to retrieve the real-time acceleration sequence from the acceleration control library after receiving a real-time order by parsing the beverage type code of the real-time order. The equipment activation unit is used to activate the liquid transfer robot arm of the coffee machine by using the real-time acceleration sequence as a feedforward control input after the real-time order is completed at the liquid inlet of the coffee machine. An offset compensation unit is used to measure the liquid center of gravity offset in real time by a six-axis force sensor integrated in the gripper during the movement of the liquid-pickup coffee cup along the non-collision liquid-pickup trajectory, and to cancel the hysteresis of the real-time acceleration sequence based on the real-time center of gravity offset.

Images