Control method and system of intelligent coffee selling intelligent robot

Abstract

The invention discloses a control method and system of an intelligent coffee selling robot, and particularly relates to the technical field of robot control. Motion state information of a beverage cup clamped by an end effector of a mechanical arm is collected in real time, amplitude and frequency characteristics of real-time shaking of liquid in the cup are extracted, dominant frequency causing liquid shaking is recognized, a critical threshold value causing liquid splashing is determined, a high-splashing-risk area on a path is determined, and the high-splashing-risk area is determined. The liquid shaking track of the mechanical arm in the next time period is predicted, self-adaptive motion track optimization data are obtained through a multi-constraint optimization algorithm, and the acceleration, posture and path of the motion track of the mechanical arm are adjusted in real time, so that accurate control over shaking and splashing risks of the liquid in the cup body is achieved, and coffee quality and equipment operation safety are guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot control, and more particularly, to a control method and system of an intelligent coffee vending robot. BACKGROUND

[0002] The intelligent coffee vending robot in the prior art usually uses a mechanical arm end effector to hold a beverage cup for making and delivering. The mechanical arm end effector needs to frequently change position, speed and direction during movement, resulting in continuous shaking of the liquid in the cup and even spilling.

[0003] The mechanical arm end effector of the intelligent coffee vending robot in the prior art lacks real-time monitoring of the liquid shaking characteristics in the cup and the ability of trajectory dynamic optimization, so that the motion trajectory of the mechanical arm is difficult to adapt to the changes in the liquid shaking characteristics under different operating conditions, thereby causing the problems of difficult control of the liquid shaking in the cup and difficult guarantee of the beverage quality.

[0004] To solve the above problems, a technical solution is provided. SUMMARY

[0005] To overcome the above-mentioned defects of the prior art, embodiments of the present application provide a control method and system of an intelligent coffee vending robot to solve the problems raised in the background art.

[0006] To achieve the above-mentioned purposes, the present application provides the following technical solutions:

[0007] A control method of an intelligent coffee vending robot, comprising the following steps:

[0008] S1: Real-time acquisition of motion state information of a beverage cup held by a mechanical arm end effector, and extraction of amplitude and frequency characteristics of real-time shaking of the liquid inside the cup to generate cup motion-liquid shaking initial state data;

[0009] S2: Based on the cup motion-liquid shaking initial state data, identification of the dominant frequency causing the liquid shaking, and determination of the acceleration / deceleration critical threshold causing the liquid spilling, output of frequency domain liquid shaking characteristic data;

[0010] S3: Based on the cup motion-liquid shaking initial state data, determination of the path high spilling risk area of the mechanical arm when performing the current action, and output of spatial domain liquid shaking risk data;

[0011] S4: Based on the cup motion-liquid shaking initial state data, determination of the amplitude prediction trajectory of the liquid shaking in the next period of motion of the mechanical arm, and output of time domain liquid shaking prediction data;

[0012] S5: generating adaptive motion trajectory optimization data of the end effector of the robot arm based on the frequency domain liquid sloshing characteristic data, the spatial domain liquid sloshing risk data and the time domain liquid sloshing prediction data through a multi-constraint optimization algorithm;

[0013] S6: adjusting the acceleration, posture and path of the motion trajectory of the end effector of the robot arm in real time according to the adaptive motion trajectory optimization data.

[0014] In a preferred embodiment, S1, specifically:

[0015] Real-time acquisition of motion state information of the cup body gripped by the end effector of the intelligent coffee vending robot arm, obtaining cup body pose, cup body speed and cup body acceleration;

[0016] Gravity compensation and filtering are performed on the cup body pose, cup body speed and cup body acceleration to obtain a cup body vertical acceleration sequence;

[0017] Acquisition of cup body mass change sequence;

[0018] According to the cup body vertical acceleration sequence and the mass change sequence, the amplitude and frequency characteristics of the real-time liquid sloshing in the cup body are calculated to generate cup body motion-liquid sloshing initial state data.

[0019] In a preferred embodiment, S2, specifically:

[0020] Based on the dominant frequency of the real-time liquid sloshing in the cup body motion-liquid sloshing initial state data, the sloshing period is determined;

[0021] The peak and valley values of the cup body vertical acceleration sequence are counted and the peak-valley difference is calculated within the sloshing period;

[0022] The peak-valley difference is compared with a preset spilling judgment threshold to determine the acceleration-deceleration critical threshold that causes liquid spilling, and frequency domain liquid sloshing characteristic data is output.

[0023] In a preferred embodiment, S3, specifically:

[0024] Based on the cup body motion-liquid sloshing initial state data, the cup body pose sequence and the cup body speed change sequence corresponding to the planned trajectory of the end effector of the robot arm are obtained;

[0025] According to the cup body pose sequence, the trajectory curvature change amount is calculated and the curvature mutation position is located;

[0026] The cup body speed change sequence is converted into trajectory acceleration-deceleration change amount and compared with the acceleration-deceleration critical threshold that causes liquid spilling;

[0027] The curvature mutation position is spatially mapped with the acceleration / deceleration super-threshold position to form a path high splashing risk area, and spatial domain liquid splashing risk data is output.

[0028] In a preferred embodiment, S4, specifically:

[0029] Based on the cup motion-liquid splashing initial state data, a cup vertical acceleration sequence and a cup pose sequence are read, and a liquid splashing phase sequence is constructed according to a dominant frequency and a splashing period;

[0030] The liquid splashing phase sequence is convoluted and added with the trajectory acceleration / deceleration change amount to obtain a future period splashing driving sequence;

[0031] The future period splashing driving sequence is recursively integrated to obtain a liquid splashing amplitude prediction trajectory, and time domain liquid splashing prediction data is output.

[0032] In a preferred embodiment, S5, specifically:

[0033] Based on the frequency domain liquid splashing feature data, the spatial domain liquid splashing risk data, and the time domain liquid splashing prediction data, a risk cost sequence with a unified time reference is constructed;

[0034] With the minimum risk cost sequence as the target, a constraint condition that the trajectory acceleration / deceleration change amount does not exceed a critical threshold of liquid splashing and avoids the path high splashing risk area is set;

[0035] According to the risk cost sequence and the constraint condition, adaptive motion trajectory optimization data of the robot arm end effector is solved.

[0036] In a preferred embodiment, S6, specifically:

[0037] The adaptive motion trajectory optimization data of the robot arm end effector is read and analyzed to obtain a target cup pose sequence, a target cup velocity sequence, and a target cup acceleration sequence;

[0038] Based on the target cup pose sequence, the target cup velocity sequence, and the target cup acceleration sequence, deviation calculations are performed on the cup pose, the cup velocity, and the cup acceleration to obtain a pose deviation sequence, a velocity deviation sequence, and an acceleration deviation sequence;

[0039] According to the pose deviation sequence, the velocity deviation sequence, and the acceleration deviation sequence, control instructions are updated and output to the robot arm drive to perform acceleration, attitude, and path adjustment of the robot arm end effector motion trajectory.

[0040] On the other hand, the present application provides a control system of an intelligent coffee vending intelligent robot, comprising:

[0041] Motion acquisition module: real-time acquisition of the motion state information of the beverage cup body held by the end effector of the mechanical arm, and extraction of the amplitude and frequency characteristics of the real-time shaking of the liquid inside the cup body, to generate cup body motion-liquid shaking initial state data;

[0042] Frequency domain threshold module: based on the cup body motion-liquid shaking initial state data, identifying the dominant frequency that causes liquid shaking, and determining the acceleration and deceleration critical threshold that causes liquid spilling, to output frequency domain liquid shaking feature data;

[0043] Risk area module: based on the cup body motion-liquid shaking initial state data, determining the path high-spilling risk area of the mechanical arm when performing the current action, to output spatial domain liquid shaking risk data;

[0044] Time domain prediction module: based on the cup body motion-liquid shaking initial state data, determining the amplitude prediction trajectory of liquid shaking in the next period of motion of the mechanical arm, to output time domain liquid shaking prediction data;

[0045] Trajectory optimization module: based on the frequency domain liquid shaking feature data, spatial domain liquid shaking risk data, and time domain liquid shaking prediction data, generating adaptive motion trajectory optimization data of the end effector of the mechanical arm through a multi-constraint optimization algorithm;

[0046] Trajectory adjustment module: according to the adaptive motion trajectory optimization data, real-time adjustment of the acceleration, attitude, and path of the motion trajectory of the end effector of the mechanical arm.

[0047] The technical effects and advantages of the control method and system of the intelligent coffee vending intelligent robot of the present application are:

[0048] By acquiring the motion state information of the beverage cup body held by the end effector of the mechanical arm and extracting the real-time shaking amplitude and frequency characteristics of the liquid inside the cup body, cup body motion-liquid shaking initial state data is formed, so that the cup body motion state and the liquid shaking state are correlated; by determining the dominant frequency of liquid shaking and giving the acceleration and deceleration critical threshold that causes liquid spilling, the motion constraint of the end effector of the mechanical arm has a quantifiable boundary; by identifying the path high-spilling risk area, the motion path of the end effector of the mechanical arm has an avoidable spatial area; the output of the liquid shaking amplitude prediction trajectory makes the motion control of the end effector of the mechanical arm have a time domain reference for the next period; by generating adaptive motion trajectory optimization data under the unified constraint of frequency domain, spatial domain, and time domain information, the trajectory planning is changed from a fixed strategy to an optimization strategy facing the characteristics of liquid shaking; according to the deviation calculation to update the control instruction and real-time adjustment of the acceleration, attitude, and path, the motion process of the end effector of the mechanical arm is kept consistent with the optimized trajectory, thereby reducing the liquid shaking and spilling probability inside the cup body and improving the stability of the beverage making process. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 A control method of the intelligent coffee vending intelligent robot;

[0050] Figure 2 A structural schematic diagram of the control system of the intelligent coffee vending intelligent robot. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0052] Embodiment 1

[0053] Figure 1 A control method of the intelligent coffee vending intelligent robot is given, which comprises the following steps:

[0054] S1: Real-time acquisition of motion state information of a beverage cup body clamped by an end effector of a mechanical arm, and extraction of amplitude and frequency characteristics of real-time shaking of a liquid inside the cup body, to generate cup body motion-liquid shaking initial state data;

[0055] S2: Based on the cup body motion-liquid shaking initial state data, identification of a dominant frequency causing liquid shaking, and determination of an acceleration / deceleration critical threshold value leading to liquid spilling, to output frequency domain liquid shaking characteristic data;

[0056] S3: Based on the cup body motion-liquid shaking initial state data, determination of a path high-spilling risk region of the mechanical arm when performing a current action, to output spatial domain liquid shaking risk data;

[0057] S4: Based on the cup body motion-liquid shaking initial state data, determination of a magnitude prediction trajectory of liquid shaking in a next period of motion of the mechanical arm, to output time domain liquid shaking prediction data;

[0058] S5: Based on the frequency domain liquid shaking characteristic data, the spatial domain liquid shaking risk data, and the time domain liquid shaking prediction data, generation of adaptive motion trajectory optimization data of the end effector of the mechanical arm through a multi-constraint optimization algorithm;

[0059] S6: Real-time adjustment of acceleration, posture, and path of a motion trajectory of the end effector of the mechanical arm according to the adaptive motion trajectory optimization data.

[0060] S1: Real-time acquisition of the motion state information of the beverage cup body clamped by the end effector of the robot arm, and extraction of the amplitude and frequency characteristics of the real-time shaking of the liquid inside the cup body, to generate cup body motion-liquid shaking initial state data, including:

[0061] Real-time acquisition of the motion state information of the cup body clamped by the end effector of the intelligent coffee vending robot arm, obtaining the cup body pose, cup body speed, and cup body acceleration;

[0062] The real-time acquisition of the cup body pose is as follows: a plurality of optical positioning marker points are pre-set on the outer surface of the cup body clamped by the end effector of the robot arm, each optical positioning marker point has three or more than three, forming a stable three-dimensional positioning structure; a three-dimensional optical positioning sensor installed outside the robot is used to track the three-dimensional spatial position coordinates of each optical positioning marker point in real time; the three-dimensional spatial position coordinates of all optical positioning marker points are converted into real-time pose information of the cup body through a coordinate transformation algorithm, and the pose information includes the three-dimensional spatial position and three-dimensional spatial attitude angle of the cup body.

[0063] The cup body speed is determined based on the continuous time sequence change of the cup body pose information, specifically: at each sampling time, the cup body pose information at the current sampling time is acquired in real time, and compared with the cup body pose information acquired at the previous sampling time; the cup body speed at the current time is calculated according to the ratio of the difference value of the three-dimensional spatial position coordinates of the cup body between the previous and subsequent sampling times to the interval time of the sampling times, and the real-time cup body speed information is obtained.

[0064] The cup body acceleration is determined based on the continuous time sequence change of the cup body speed information, specifically: at each sampling time, the cup body speed information at the current time is acquired in real time, and compared with the cup body speed information acquired at the previous sampling time; the cup body acceleration at the current time is calculated according to the ratio of the difference value of the cup body speed between the previous and subsequent sampling times to the interval time of the sampling times, and the real-time cup body acceleration information is obtained.

[0065] Gravity compensation and filtering are performed on the cup body pose, cup body speed, and cup body acceleration to obtain a cup body vertical acceleration sequence.

[0066] The three-dimensional spatial attitude angle of the cup body in the real-time pose information of the cup body is determined, the included angle between the current direction of the cup body and the direction of gravity is determined according to the three-dimensional spatial attitude angle, and the gravity components in the cup body speed and cup body acceleration information are decomposed and removed from the cup body speed and cup body acceleration information according to the included angle to obtain the cup body speed and cup body acceleration information after gravity compensation. The filtering process adopts a low-pass filtering algorithm, specifically: the cup body speed and cup body acceleration information after gravity compensation are input into a low-pass filter to remove high-frequency noise interference, and the cutoff frequency of the low-pass filter is determined according to the measurement noise distribution under the actual motion state of the cup body, for example, it is set to be within the range of 0.5 Hz to 2 Hz to ensure the filtering effect.

[0067] According to the cup acceleration information after gravity compensation and filtering processing, the cup vertical acceleration sequence is extracted, specifically: the current vertical direction unit vector of the cup is determined based on the cup pose information; the real-time cup acceleration information is projected onto the current vertical direction unit vector of the cup to obtain the cup vertical acceleration at each sampling time, and the cup vertical accelerations at all times constitute the cup vertical acceleration sequence.

[0068] The cup mass change sequence is collected;

[0069] A high-precision weighing sensor is arranged at the bottom of the cup gripped by the end effector of the mechanical arm to collect the current mass of the cup in real time, the collection frequency is consistent with the collection frequency of the cup motion state information, for example, it can be set to 50 to 100 times per second, and a cup mass sequence corresponding to the cup vertical acceleration sequence is generated, which is called a cup mass change sequence.

[0070] According to the cup vertical acceleration sequence and the mass change sequence, the amplitude and frequency characteristics of the real-time liquid sloshing in the cup are calculated to generate the cup motion-liquid sloshing initial state data;

[0071] The real-time mass of the liquid in the cup is determined based on the cup mass change sequence; the real-time liquid level height of the liquid in the cup is determined according to the real-time mass of the liquid in the cup and the inner diameter size of the cup container; based on the real-time liquid level height and the cup vertical acceleration sequence, the real-time sloshing amplitude is calculated, and the calculation method is: a liquid free sloshing model is established at the liquid level height position of the cup, the liquid free sloshing model includes but is not limited to a single degree of freedom swing model, the single degree of freedom swing model is excited and responded according to the real-time cup vertical acceleration to obtain the real-time sloshing amplitude of the liquid at each sampling time; then the frequency spectrum amplitude distribution is obtained by performing frequency spectrum amplitude calculation on the cup vertical acceleration sequence by using the fast Fourier transform method; the frequency corresponding to the peak amplitude in the frequency spectrum amplitude distribution is determined as the dominant frequency, and the frequencies corresponding to the secondary peak amplitudes in the frequency spectrum amplitude distribution except the dominant frequency are determined as the secondary frequencies.

[0072] Finally, the obtained cup motion state information and the amplitude and frequency characteristics of the real-time liquid sloshing are fused to form the cup motion-liquid sloshing initial state data. The cup motion-liquid sloshing initial state data includes the real-time pose, real-time speed, real-time acceleration, real-time vertical acceleration, real-time mass of the cup, real-time liquid level height of the liquid in the cup, real-time sloshing amplitude of the liquid, and real-time sloshing dominant frequency of the liquid at each sampling time.

[0073] S2: Based on the cup motion-liquid sloshing initial state data, the dominant frequency causing the liquid sloshing is identified, and the acceleration and deceleration critical threshold causing the liquid spatter is determined, and the frequency domain liquid sloshing characteristic data is output, including:

[0074] The sloshing period is determined based on the dominant frequency of real-time liquid sloshing in the initial state data of cup motion and liquid sloshing.

[0075] The sloshing period is the reciprocal of the dominant frequency; that is, the liquid sloshing period is obtained by taking the reciprocal of the dominant frequency. For example, when the dominant frequency is 0.8 Hz, the sloshing period is 1 / 0.8 s, or 1.25 s. After determining the sloshing period, based on the sampling frequency of the vertical acceleration sequence of the cup, the vertical acceleration sequence of the cup is divided into several periodic data segments with the sloshing period as the length. The data length of each periodic data segment is the product of the sloshing period and the sampling frequency. For example, when the sloshing period is 1.25 s and the sampling frequency is 50 Hz, the length of each periodic data segment is 62.5 data points. 62 or 63 data points can be selected to ensure the integrity of the periodic data segment.

[0076] Within the shaking period, the peak and trough values ​​of the vertical acceleration sequence of the cup are statistically analyzed, and the peak-trough difference is calculated.

[0077] Within each periodic data segment, the peak and trough values ​​of the cup's vertical acceleration sequence are statistically analyzed. The peak value is defined as the maximum value of the cup's vertical acceleration sequence within each periodic data segment, and the trough value is defined as the minimum value. Determining the peak and trough values ​​involves a point-by-point comparison method, where the values ​​of each data point within the periodic data segment are compared sequentially to find the maximum and minimum values, which are then recorded as the peak and trough values ​​for the current period, respectively. The peak and trough values ​​within each periodic data segment are then subtracted from each other to obtain the peak-to-trough difference for each period.

[0078] The peak-to-valley difference is compared with the preset splash judgment threshold to determine the critical threshold for acceleration and deceleration that causes liquid splashing, and the frequency domain liquid sloshing characteristic data is output.

[0079] The method for determining the splash detection threshold is as follows: In the actual application environment of the intelligent coffee vending robot, the cup held by the end effector of the robotic arm moves under acceleration of different amplitudes. By gradually adjusting the amplitude and speed of the movement, the actual situation of liquid splashing is observed. The peak-to-valley difference that just causes liquid splashing is recorded as the splash detection threshold. The splash detection threshold is determined by averaging multiple experiments. For example, more than 10 consecutive experiments are conducted, and the average of the peak-to-valley differences that cause liquid splashing is taken as the splash detection threshold.

[0080] The peak-to-valley difference of each periodic data segment is compared with the spilling determination threshold one by one, and when the peak-to-valley difference in a certain periodic data segment exceeds the spilling determination threshold, it is judged that the acceleration and deceleration amplitude of the current cup movement exceeds the critical state, and the cup acceleration and deceleration at this time is recorded as the acceleration and deceleration critical threshold that causes liquid spilling; if multiple periodic data segments exceed the spilling determination threshold, the minimum value of all recorded cup acceleration and deceleration is taken as the acceleration and deceleration critical threshold that causes liquid spilling.

[0081] The dominant frequency, the sloshing period and the acceleration and deceleration critical threshold that causes liquid spilling are integrated to form the frequency domain liquid sloshing feature data. The frequency domain liquid sloshing feature data includes the dominant frequency of the cup vertical acceleration sequence, the sloshing period of the liquid sloshing, and the acceleration and deceleration critical threshold that causes liquid spilling.

[0082] S3: Based on the cup motion-liquid sloshing initial state data, determine the path high spilling risk area of the mechanical arm when performing the current action, and output the spatial domain liquid sloshing risk data, including:

[0083] Based on the cup motion-liquid sloshing initial state data, obtain the cup pose sequence and the cup speed change sequence corresponding to the end effector planning trajectory of the mechanical arm;

[0084] According to the real-time pose of the cup recorded at each sampling time in the cup motion-liquid sloshing initial state data, the real-time pose of the cup at all sampling times is arranged in sequence according to the complete motion trajectory planned by the end effector of the mechanical arm, to form a complete cup pose sequence. The cup pose sequence includes the three-dimensional space position coordinates and the three-dimensional space attitude angles of the cup at each sampling time, wherein the three-dimensional space position coordinates are based on the reference coordinate system of the robot workspace, and the three-dimensional space attitude angles are represented in the form of rotation matrix or Euler angle. The length and number of the cup pose sequence are determined by the total duration of the motion trajectory planned by the end effector of the mechanical arm and the sampling frequency of the cup pose data, for example, when the duration of the motion trajectory planned by the end effector of the mechanical arm is 5 seconds and the sampling frequency of the cup pose data is 50 Hz, the cup pose sequence contains 250 cup pose data points in total.

[0085] According to the real-time speed of the cup recorded at each sampling time in the cup motion-liquid sloshing initial state data, a complete cup speed change sequence is obtained. The method for obtaining the cup speed change sequence is: taking the real-time speed of the cup at each sampling time in the cup pose sequence as the data points of the speed change sequence, and arranging the data points in sequence to form the cup speed change sequence. The length and sampling points of the cup speed change sequence are consistent with the cup pose sequence. Each data point in the cup speed change sequence is represented in the form of a speed vector, i.e. a three-dimensional speed vector in the reference coordinate system of the robot workspace, and each speed vector is composed of three components along the coordinate axis direction.

[0086] Calculate the change in trajectory curvature based on the cup's pose sequence and locate the location of curvature abrupt changes;

[0087] Extract the three-dimensional spatial coordinate sequence of the cup from the cup's pose sequence. For three consecutive data points in the position coordinate sequence, calculate the chord length and corresponding angle between two adjacent points. Based on the arc determined by the three data points, approximate the curvature of the trajectory at each intermediate data point. The formula is expressed as:

[0088] ;in, Indicates the first The trajectory curvature at each data point, in units of 1 / m; Indicates the first The angle formed between three consecutive data points at the center of a data point, in radians; This represents the average chord length between adjacent data points, in meters; calculate the trajectory curvature at all position data points in the cup pose sequence to obtain a complete sequence of trajectory curvature changes.

[0089] Based on the trajectory curvature change sequence, the curvature abrupt change location is determined: the trajectory curvature change sequence is scanned point by point, and the absolute value of the curvature change between two adjacent data points is compared with a pre-set curvature abrupt change judgment threshold. When the absolute value of the curvature change is greater than the curvature abrupt change judgment threshold, the corresponding data point position is recorded as the curvature abrupt change location. The curvature abrupt change judgment threshold is determined by an experimental calibration method. Under the actual working state of the robot, the influence of the end effector of the robotic arm on liquid splashing under different amplitude bending trajectories is repeatedly tested, and the trajectory curvature change that causes liquid splashing is used as the curvature abrupt change judgment threshold.

[0090] The sequence of cup velocity changes was converted into trajectory acceleration and deceleration changes and compared with the critical acceleration and deceleration thresholds that cause liquid spillage;

[0091] Based on the velocity change sequence of the cup, the difference between the velocity vectors of every two adjacent data points is calculated and divided by the time interval between adjacent sampling moments to obtain the acceleration / deceleration change sequence of the cup trajectory at each moment. The acceleration / deceleration change is represented as a scalar, that is, the magnitude of the absolute value of the velocity vector difference. Its physical meaning is the acceleration / deceleration intensity of the cup's motion trajectory driven by the robotic arm's end effector, with units of m / s². Each data point in the acceleration / deceleration change sequence is compared one by one with the acceleration / deceleration threshold that causes liquid splashing in the frequency domain liquid sloshing feature data. When the acceleration / deceleration change value exceeds the acceleration / deceleration threshold that causes liquid splashing, the corresponding data point position is recorded as the acceleration / deceleration over-threshold position.

[0092] By spatially mapping the locations of curvature abrupt changes with the locations of acceleration and deceleration exceeding the threshold, a high-splash-risk area along the path is formed, and spatial domain liquid sloshing risk data is output.

[0093] The spatial mapping method is as follows: using the robot's workspace reference coordinate system as a unified benchmark, the spatial coordinates of curvature abrupt change locations and acceleration / deceleration over-threshold locations are determined and marked on the planned trajectory path of the robotic arm's end effector. All marked locations are then categorized and aggregated, with the distance between adjacent coordinates used as the clustering criterion. When the distance between adjacent marked coordinates is less than a preset clustering distance threshold, they are merged into the same high-splash-risk area along the same path. The clustering distance threshold is determined through repeated experimental verification; for example, it can be set within the range of 0.05m to 0.15m to ensure the accuracy of the high-splash-risk area locations.

[0094] The spatial domain liquid sloshing risk data consists of the location coordinate ranges of several high-splash-risk areas along a path. Each high-splash-risk area is represented as follows: the average spatial coordinates of all marked points within the high-splash-risk area are used as the center coordinates, and the distance between the farthest marked point and the center coordinates is used as the radius, forming a circular area representing the spatial location range of the high-splash-risk area. The spatial domain liquid sloshing risk data is stored in a data table format, including the center coordinates and radius of each high-splash-risk area.

[0095] S4: Based on the initial state data of the cup motion and liquid sloshing, determine the predicted trajectory of the liquid sloshing amplitude during the next time period of the robotic arm's movement, and output time-domain liquid sloshing prediction data, including:

[0096] Based on the initial state data of cup motion and liquid sloshing, the vertical acceleration sequence and cup pose sequence of the cup are read, and the liquid sloshing phase sequence is constructed according to the dominant frequency and sloshing period;

[0097] From the initial state data of the cup's motion and liquid sloshing, the vertical acceleration sequence of the cup is extracted. This includes the real-time acceleration of the cup held by the robotic arm's end effector along the vertical direction at each sampling moment, as well as the cup's three-dimensional spatial position coordinates and three-dimensional spatial attitude angle corresponding to the cup's pose sequence at the corresponding moment. Based on the dominant frequency of the real-time liquid sloshing and the sloshing period of the liquid, the liquid sloshing phase at each sampling moment is calculated. The liquid sloshing phase is defined as follows: using the sloshing period of the liquid as a reference, the complete period corresponding to the dominant frequency of the real-time liquid sloshing is set as a 360-degree phase period. The liquid sloshing phase angle at the current sampling moment is obtained by multiplying the ratio of the time difference between the current sampling moment and the start time of the nearest sloshing period to the sloshing period of the liquid by 360 degrees. Thus, the liquid sloshing phase is assigned to each data point in the vertical acceleration sequence of the cup, forming a liquid sloshing phase sequence.

[0098] The sloshing phase sequence and the trajectory acceleration / deceleration change are convolved and summed to obtain the sloshing driving sequence for future time periods;

[0099] The liquid sloshing phase sequence can be represented as a function that varies with time, where the liquid sloshing phase angle at each sampling moment corresponds to a phase function value, which can be expressed in sine or cosine function form. For example, it can be defined as:

[0100] ;in, Represents the phase function of liquid sloshing; The numerical value representing the sloshing period of the liquid, in seconds; This indicates the current sampling time, in seconds; This represents the initial phase offset angle, expressed in radians. The initial phase offset angle is determined by fitting the liquid sloshing phase corresponding to the initial data points of the vertical acceleration sequence of the cup, ensuring that the initial moment of the liquid sloshing phase function matches the initial phase of the actual liquid sloshing phase sequence.

[0101] The liquid sloshing phase function is convolved with the trajectory acceleration / deceleration change sequence. The trajectory acceleration / deceleration change sequence represents the acceleration / deceleration changes at each sampling moment of the planned trajectory of the robotic arm's end effector. The convolution calculation is performed using numerical convolution integration. At each sampling moment, with the current moment as the convolution center, the liquid sloshing phase function and the trajectory acceleration / deceleration change sequence are translated along both positive and negative directions, multiplied point by point, and then summed to form the sloshing drive value corresponding to the current moment. This convolution process is repeated for all sampling moments within the future time period of the robotic arm's end effector's planned trajectory, obtaining the sloshing drive value for each sampling moment to form the future time period sloshing drive sequence.

[0102] The predicted trajectory of liquid sloshing amplitude is obtained by recursively integrating the sloshing drive sequence in the future time period, and the time-domain liquid sloshing prediction data is output.

[0103] The first sampling moment in the future time period sloshing drive sequence is taken as the initial point of integration. The liquid sloshing amplitude at the initial point of integration is taken from the current real-time liquid sloshing amplitude corresponding to the cup motion-liquid sloshing initial state data. Numerical integration is performed successively on the future time period sloshing drive sequence, and the formula for calculating the predicted liquid sloshing amplitude at each integration moment is:

[0104] ;in, This represents the predicted amplitude of liquid sloshing at the next sampling time, in meters. This represents the predicted amplitude of liquid sloshing at the current sampling moment, in meters. This indicates the shaking drive value corresponding to the current sampling time; This represents the sampling time interval, in seconds. Through recursive integration time-by-time calculation, the predicted liquid sloshing amplitude corresponding to each sampling time in the future period is successively obtained, forming a predicted trajectory of the liquid sloshing amplitude.

[0105] The predicted trajectory of liquid sloshing amplitude is output as time-domain liquid sloshing prediction data. The predicted trajectory of liquid sloshing amplitude is expressed as a data sequence, including the predicted liquid sloshing amplitude corresponding to each sampling time in the future time period, and is stored in the form of a data table, where each row contains the sampling time timestamp and the corresponding predicted liquid sloshing amplitude.

[0106] S5: Based on frequency domain liquid sloshing feature data, spatial domain liquid sloshing risk data, and time domain liquid sloshing prediction data, adaptive motion trajectory optimization data for the robotic arm end effector is generated through a multi-constraint optimization algorithm, including:

[0107] Based on frequency domain liquid sloshing feature data, spatial domain liquid sloshing risk data, and time domain liquid sloshing prediction data, a risk cost sequence with a unified time benchmark is constructed.

[0108] The method for determining the unified time reference is as follows: Extract the complete time interval of the robotic arm's end effector's planned trajectory from the initial state data of the cup's motion and liquid sloshing. Divide this complete time interval evenly into fixed time steps, using this as the unified time reference for the risk cost sequence. The time step is obtained by dividing the sampling frequency by the reciprocal of the sampling frequency. The sampling frequency is the frequency used when acquiring the initial state data of the cup's motion and liquid sloshing. For example, when the data sampling frequency is 50Hz, the time step is 0.02 seconds.

[0109] Under a unified time reference, frequency domain liquid sloshing feature data, spatial domain liquid sloshing risk data, and time domain liquid sloshing prediction data are matched accordingly. The frequency domain liquid sloshing feature data includes the dominant frequency of real-time liquid sloshing, the sloshing period, and the acceleration / deceleration thresholds leading to liquid splashing. The spatial domain liquid sloshing risk data includes the center coordinates and radius of high-splash-risk areas along the path. The time domain liquid sloshing prediction data includes the predicted liquid sloshing amplitude for each sampling moment in the future period. Interpolation or mapping is performed on the frequency domain liquid sloshing feature data, spatial domain liquid sloshing risk data, and time domain liquid sloshing prediction data at each time step of the unified time reference to ensure a correspondence between the data at each moment.

[0110] At each sampling moment based on a unified time reference, the risk cost is calculated. The risk cost is defined as a comprehensive quantification of the risks associated with liquid sloshing amplitude, trajectory acceleration / deceleration, and spatial splashing:

[0111] The method for quantifying the risk of liquid sloshing amplitude is as follows: The predicted liquid sloshing amplitude is compared with a preset safe sloshing amplitude threshold. The safe sloshing amplitude threshold is determined through experimental calibration, including continuously increasing the amplitude of the cup's movement in an actual operating environment and observing whether liquid spillage occurs. The sloshing amplitude that causes spillage is recorded as the safe sloshing amplitude threshold, for example, set within the range of 0.5 to 2 cm. If the current predicted liquid sloshing amplitude exceeds the safe sloshing amplitude threshold, the risk value is calculated as the difference between the predicted amplitude and the safe threshold; otherwise, the risk value is 0.

[0112] The trajectory acceleration / deceleration risk quantification method compares the change in trajectory acceleration / deceleration with the critical acceleration / deceleration threshold that causes liquid splashing. When the change in trajectory acceleration / deceleration exceeds the critical acceleration / deceleration threshold, the difference between the change in trajectory acceleration / deceleration and the critical acceleration / deceleration threshold is taken as the risk value; otherwise, the risk value is 0.

[0113] Spatial spill risk quantification method: This involves determining whether the cup's pose, corresponding to the planned trajectory of the robotic arm's end effector, is within a high-splash-risk area along the path. This includes calculating the distance between the cup's real-time spatial coordinates and the center coordinates of the high-splash-risk area. If the distance is less than the area radius, the cup is within the risk area; otherwise, it is outside. When the cup is within the risk area, the risk value is calculated as the difference between the radius of the high-splash-risk area and the current distance; otherwise, the risk value is 0.

[0114] Each risk weight coefficient is assigned, and the three risk values ​​are weighted and summed to form the comprehensive risk cost at each unified time reference moment. The risk weight coefficients are obtained by combining expert experience with experimental verification. In the actual operation environment of the robot, multiple verification tests are conducted on the motion trajectory of the robotic arm's end effector with different types of risk weight combinations. The probability and degree of liquid splashing in each test are recorded. The risk weights are adjusted and optimized according to the actual severity of liquid splashing to obtain the optimal risk weight coefficients for liquid sloshing amplitude risk, trajectory acceleration and deceleration risk, and spatial splashing risk. For example, they can be set to 0.4, 0.3, and 0.3, respectively.

[0115] With the goal of minimizing the risk cost sequence, constraints are set such that the trajectory acceleration and deceleration changes do not exceed the critical acceleration and deceleration threshold that would cause liquid splashing, and the path avoids high-splashing risk areas.

[0116] With the objective of minimizing the risk cost sequence, a trajectory optimization model is established, constrained by conditions that the trajectory acceleration / deceleration changes do not exceed the critical acceleration / deceleration threshold leading to liquid splashing and that high-splashing risk areas along the path are avoided. The construction process of the trajectory optimization model is as follows:

[0117] The objective function is the sum of the risk cost sequences corresponding to all sampling moments of the robotic arm's end effector trajectory. The objective function is defined as follows:

[0118] ;in, This represents the objective function, which is the sum of the risk cost sequence. This represents the total number of discrete sampling moments involved in the summation under a unified time reference. Represents the first time under each unified time base The comprehensive risk cost corresponding to each sampling moment;

[0119] The constraints include trajectory acceleration / deceleration constraints and spatial position constraints. Specifically, the trajectory acceleration / deceleration constraint is that the acceleration / deceleration change at each moment in the sequence of acceleration / deceleration changes of the robotic arm end effector trajectory must not exceed the critical acceleration / deceleration threshold that would cause liquid spillage. The spatial position constraint is that the spatial position coordinates of the robotic arm end effector trajectory must not enter the high-splash-risk area of ​​the path.

[0120] The adaptive motion trajectory optimization data of the robotic arm end effector is obtained by solving the risk cost sequence and constraint conditions;

[0121] A sequential quadratic programming method is employed to solve the trajectory optimization model. The current trajectory planned by the robotic arm's end effector is used as the initial trajectory solution. A set of optimization equations is established for the objective function and constraints. In each iteration, the sequential quadratic programming method transforms the nonlinear optimization problem into a quadratic programming subproblem and solves iteratively until a trajectory solution that minimizes the objective function and satisfies the constraints is obtained. The iteration termination criterion is that the change in trajectory solution between two adjacent iterations is less than a preset convergence threshold. The convergence threshold is determined empirically, for example, within the range of 0.01 to 0.001 meters, to ensure stable convergence of the algorithm.

[0122] The adaptive motion trajectory optimization data of the robotic arm end effector is obtained by optimizing the solution algorithm. The adaptive motion trajectory optimization data consists of the trajectory pose and velocity planning values ​​at each moment under a unified time reference, and is stored in the form of a standard data table. Each row in the table includes the sampling time timestamp, trajectory pose planning, and trajectory velocity planning values.

[0123] S6: Based on adaptive motion trajectory optimization data, adjust the acceleration, attitude, and path of the robotic arm's end effector in real time, including:

[0124] Read the adaptive motion trajectory optimization data of the robotic arm end effector and parse it to obtain the target cup pose sequence, target cup velocity sequence, and target cup acceleration sequence;

[0125] From the adaptive motion trajectory optimization data of the robotic arm's end effector, the trajectory pose planning value and trajectory velocity planning value corresponding to each sampling moment under a unified time reference are extracted. The trajectory pose planning value includes the three-dimensional spatial position coordinates and three-dimensional spatial attitude angles of the target cup in the robot's workspace reference coordinate system. The three-dimensional spatial attitude angles are represented in the form of rotation matrices or Euler angles. The trajectory velocity planning value is represented in the form of velocity vectors, and each velocity vector consists of velocity components along the three axes of the robot's workspace reference coordinate system.

[0126] Based on the extracted trajectory pose planning value sequence and trajectory velocity planning value sequence, the target cup acceleration sequence is calculated time-by-time: For each sampling time under a unified time reference, the trajectory velocity planning value of the current sampling time and the next adjacent sampling time are read respectively. The difference vector of the trajectory velocity planning values ​​of adjacent sampling times is calculated and then divided by the sampling time interval to obtain the target cup acceleration at the current sampling time. The sampling time interval is determined by a fixed time step under a unified time reference. For example, when the sampling frequency is 50Hz, the sampling time interval is 0.02 seconds. By repeating the above process time-by-time, the target cup acceleration sequence corresponding to the trajectory pose planning value sequence and trajectory velocity planning value sequence can be obtained.

[0127] Based on the target cup pose sequence, target cup velocity sequence, target cup acceleration sequence and the cup pose, cup velocity, cup acceleration, the deviation is calculated to obtain the pose deviation sequence, velocity deviation sequence and acceleration deviation sequence;

[0128] The cup pose deviation is calculated as follows: For each sampling moment under a unified time reference, the cup position coordinates and cup attitude angles in the target cup pose sequence are extracted, as well as the cup real-time position coordinates and cup real-time attitude angles at the corresponding moment in the cup motion-liquid sloshing initial state data. The difference between the target cup pose and the real-time cup pose is calculated separately. The position coordinate difference is the target position coordinate minus the real-time position coordinate, and the attitude angle difference is the target attitude angle minus the real-time attitude angle. Thus, the cup pose deviation corresponding to each sampling moment is obtained, and the cup pose deviations at all sampling moments constitute the pose deviation sequence.

[0129] The method for calculating the cup velocity deviation is as follows: at each sampling time, the velocity vector in the target cup velocity sequence and the real-time velocity vector in the cup motion-liquid sloshing initial state data at the corresponding time are extracted. The target velocity vector and the real-time velocity vector are subtracted component by component to obtain the velocity deviation vector. The absolute value of the velocity deviation vector is taken as the velocity deviation at the current sampling time. The velocity deviations at all sampling times form a velocity deviation sequence.

[0130] The method for calculating the acceleration deviation of the cup is similar to that for calculating the velocity deviation. At each sampling moment, the acceleration in the target cup acceleration sequence is extracted and the real-time acceleration of the cup at the corresponding moment is extracted from the cup motion-liquid sloshing initial state data. The target acceleration is subtracted from the real-time acceleration and the absolute value is taken to obtain the acceleration deviation at each sampling moment. The acceleration deviations at all sampling moments constitute the acceleration deviation sequence.

[0131] The control commands are updated based on the pose deviation sequence, velocity deviation sequence, and acceleration deviation sequence and output to the robotic arm drive to perform acceleration, attitude, and path adjustment of the robotic arm end effector's motion trajectory;

[0132] The control gain coefficients are multiplied for the pose deviation sequence, velocity deviation sequence, and acceleration deviation sequence respectively, and the deviation compensation control quantity at each sampling time is calculated. The control gain coefficients adopt the proportional, integral, and derivative control method in the PID control algorithm. The setting method is as follows: through repeated experiments in the actual robot environment, the proportional, integral, and derivative control gain coefficients are adjusted according to the sensitivity of the robot arm end effector to the response of pose deviation, velocity deviation, and acceleration deviation, so as to maximize the trajectory response speed of the robot arm end effector without overshoot or oscillation. For example, the proportional control gain coefficient for pose deviation can be set between 1.0 and 5.0, the proportional control gain coefficient for velocity deviation can be set between 0.5 and 2.0, and the proportional control gain coefficient for acceleration deviation can be set between 0.1 and 1.0.

[0133] The calculation formula for the control command update is as follows: At the j-th sampling time of the unified time base, read the pose deviation value, velocity deviation value, and acceleration deviation value corresponding to the j-th sampling time; multiply the pose deviation value with the pose deviation control gain coefficient to obtain the pose deviation compensation component; multiply the velocity deviation value with the velocity deviation control gain coefficient to obtain the velocity deviation compensation component; multiply the acceleration deviation value with the acceleration deviation control gain coefficient to obtain the acceleration deviation compensation component; add the pose deviation compensation component, velocity deviation compensation component, and acceleration deviation compensation component together, and the accumulated result is determined as the control command update value at the j-th sampling time.

[0134] After calculating the updated control command values ​​for the robotic arm end effector, the values ​​are sent to the drive control unit of the end effector via the robot control interface. Upon receiving the updated control command values, the drive control unit immediately adjusts the acceleration, attitude, and path of the current execution trajectory of the robotic arm end effector. The adjustment process involves adjusting the servo motor drive torque and motion angle of each joint of the robotic arm end effector in real time based on the updated control command values. This ensures that the actual motion trajectory of the robotic arm end effector converges quickly with the target trajectory, thereby executing the optimized adaptive motion trajectory.

[0135] During real-time control, the drive control unit of the robotic arm end effector can adopt a closed-loop feedback control method to improve trajectory execution accuracy. The closed-loop feedback control method is as follows: continuously collect and provide real-time feedback on the cup's posture, velocity, and acceleration during the actual movement of the robotic arm end effector. By recalculating the deviation between the real-time feedback information and the target cup posture sequence, target cup velocity sequence, and target cup acceleration sequence, and correcting the updated control command value in real time, the trajectory of the robotic arm end effector is always consistent with the target trajectory, thereby dynamically correcting the trajectory error and ensuring that the real-time sloshing of the liquid inside the cup is kept within a safe threshold.

[0136] Example 2

[0137] The difference between Embodiment 2 and Embodiment 1 is that this embodiment introduces a control system for an intelligent coffee vending robot.

[0138] Figure 2 A schematic diagram of the control system for an intelligent coffee vending robot according to the present invention is provided. The control system for the intelligent coffee vending robot includes:

[0139] Motion acquisition module: Real-time acquisition of motion state information of the beverage cup held by the end effector of the robotic arm, and extraction of the amplitude and frequency characteristics of the real-time sloshing of the liquid inside the cup, generating cup motion-liquid sloshing initial state data;

[0140] Frequency domain threshold module: Based on the cup motion-liquid sloshing initial state data, it identifies the dominant frequency that causes liquid sloshing, determines the acceleration and deceleration critical threshold that causes liquid splashing, and outputs frequency domain liquid sloshing characteristic data;

[0141] Risk Area Module: Based on the cup motion-liquid sloshing initial state data, determine the high spill risk area along the path when the robotic arm performs the current action, and output spatial domain liquid sloshing risk data;

[0142] Time-domain prediction module: Based on the initial state data of cup motion and liquid sloshing, determine the amplitude prediction trajectory of liquid sloshing during the next time period of the robotic arm's movement, and output time-domain liquid sloshing prediction data;

[0143] Trajectory optimization module: Based on frequency domain liquid sloshing feature data, spatial domain liquid sloshing risk data and time domain liquid sloshing prediction data, it generates adaptive motion trajectory optimization data for the robotic arm end effector through a multi-constraint optimization algorithm;

[0144] Trajectory Adjustment Module: Based on adaptive motion trajectory optimization data, adjusts the acceleration, attitude, and path of the robotic arm's end effector in real time.

[0145] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0146] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0147] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0148] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0149] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0150] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0151] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0152] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0153] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A control method of an intelligent coffee vending intelligent robot, characterized in that, The method comprises the following steps: S1: Real-time acquisition of the motion state information of the beverage cup body clamped by the end effector of the mechanical arm, and extraction of the amplitude and frequency characteristics of the real-time shaking of the liquid inside the cup body, to generate cup body motion-liquid shaking initial state data; S2: Based on the cup body motion-liquid shaking initial state data, the dominant frequency causing the liquid shaking is identified, and the acceleration / deceleration critical threshold causing the liquid spilling is determined, to output the frequency domain liquid shaking characteristic data; S3: Based on the cup body motion-liquid shaking initial state data, the path high spilling risk area of the mechanical arm when performing the current action is determined, to output the spatial domain liquid shaking risk data; S4: Based on the cup body motion-liquid shaking initial state data, the amplitude prediction trajectory of the liquid shaking in the next period of motion of the mechanical arm is determined, to output the time domain liquid shaking prediction data; S5: Based on the frequency domain liquid shaking characteristic data, the spatial domain liquid shaking risk data and the time domain liquid shaking prediction data, the adaptive motion trajectory optimization data of the end effector of the mechanical arm is generated through a multi-constraint optimization algorithm; S6: According to the adaptive motion trajectory optimization data, the acceleration, posture and path of the motion trajectory of the end effector of the mechanical arm are adjusted in real time. 2.The control method of the intelligent coffee vending robot according to claim 1, wherein, S1, specifically: Real-time acquisition of the motion state information of the cup body clamped by the end effector of the intelligent coffee vending intelligent robot mechanical arm, to obtain the cup body pose, cup body speed and cup body acceleration; Gravity compensation and filtering are performed on the cup body pose, cup body speed and cup body acceleration to obtain a cup body vertical acceleration sequence; A cup body mass change sequence is collected; According to the cup body vertical acceleration sequence and the mass change sequence, the amplitude and frequency characteristics of the real-time shaking of the liquid inside the cup body are calculated to generate cup body motion-liquid shaking initial state data. 3.The control method of the intelligent coffee vending robot according to claim 2, wherein, S2, specifically: Based on the dominant frequency of the real-time shaking of the liquid in the cup body motion-liquid shaking initial state data, the shaking period is determined; The peak value and valley value of the cup body vertical acceleration sequence are counted and the peak-valley difference is calculated within the shaking period; The peak-valley difference is compared with a preset spilling judgment threshold to determine the acceleration / deceleration critical threshold causing the liquid spilling, and the frequency domain liquid shaking characteristic data is output.

4. The control method of claim 3, wherein the control method further comprises: S3, specifically: Based on the cup body motion-liquid shaking initial state data, the cup body pose sequence and cup body speed change sequence corresponding to the planned trajectory of the end effector of the mechanical arm are obtained; According to the cup body pose sequence, the trajectory curvature change amount is calculated and the curvature mutation position is located; The cup body speed change sequence is converted into trajectory acceleration / deceleration change amount and compared with the acceleration / deceleration critical threshold causing the liquid spilling; The curvature mutation position and the acceleration / deceleration super-threshold position are spatially mapped to form a path high spilling risk area, and the spatial domain liquid shaking risk data is output.

5. The control method of claim 4, wherein, S4, specifically: Based on the cup body motion-liquid shaking initial state data, the cup body vertical acceleration sequence and the cup body pose sequence are read, and a liquid shaking phase sequence is constructed according to the dominant frequency and the shaking period; The liquid shaking phase sequence and the trajectory acceleration / deceleration change amount are convoluted and added to obtain a future period shaking driving sequence; The future period shaking driving sequence is recursively integrated to obtain a liquid shaking amplitude prediction trajectory, and the time domain liquid shaking prediction data is output.

6. The control method of claim 5, wherein, S5, specifically: Based on the frequency domain liquid shaking characteristic data, the spatial domain liquid shaking risk data and the time domain liquid shaking prediction data, a risk cost sequence with unified time reference is constructed; With the minimum risk cost sequence as the target, a constraint condition that the trajectory acceleration and deceleration variation does not exceed the critical threshold of acceleration and deceleration leading to liquid spilling and avoids the high spilling risk area of the path is set; The adaptive motion trajectory optimization data of the end effector of the robot arm is obtained according to the risk cost sequence and the constraint condition.

7. The control method of claim 6, wherein the control method further comprises: S6, specifically: Read the adaptive motion trajectory optimization data of the end effector of the robot arm and parse to obtain the target cup body pose sequence, target cup body velocity sequence and target cup body acceleration sequence; Based on the target cup body pose sequence, target cup body velocity sequence and target cup body acceleration sequence, the pose deviation sequence, velocity deviation sequence and acceleration deviation sequence are calculated by deviation calculation based on the cup body pose, cup body velocity and cup body acceleration; According to the pose deviation sequence, velocity deviation sequence and acceleration deviation sequence, the control instruction is updated and output to the robot arm drive to perform acceleration, attitude and path adjustment of the motion trajectory of the end effector of the robot arm.

8. A control system of an intelligent coffee vending intelligent robot, used for implementing the control method of the intelligent coffee vending intelligent robot according to any one of claims 1-7, characterized in that, It includes: Motion acquisition module: real-time acquisition of the motion state information of the beverage cup body clamped by the end effector of the robot arm, and extraction of the amplitude and frequency characteristics of the real-time shaking of the liquid in the cup body, to generate cup body motion-liquid shaking initial state data; Frequency domain threshold module: based on the cup body motion-liquid shaking initial state data, the dominant frequency causing liquid shaking is identified, and the critical threshold of acceleration and deceleration leading to liquid spilling is determined, and the frequency domain liquid shaking characteristic data is output; Risk area module: based on the cup body motion-liquid shaking initial state data, the high spilling risk area of the path when the robot arm performs the current action is determined, and the spatial domain liquid shaking risk data is output; Time domain prediction module: based on the cup body motion-liquid shaking initial state data, the amplitude prediction trajectory of liquid shaking in the next period of robot arm motion is determined, and the time domain liquid shaking prediction data is output; Trajectory optimization module: based on the frequency domain liquid shaking characteristic data, the spatial domain liquid shaking risk data and the time domain liquid shaking prediction data, the adaptive motion trajectory optimization data of the end effector of the robot arm is generated through a multi-constraint optimization algorithm; Trajectory adjustment module: according to the adaptive motion trajectory optimization data, the acceleration, attitude and path of the motion trajectory of the end effector of the robot arm are adjusted in real time.

Images