A humanoid robot dual-arm cooperative moving operation generation method and device

Abstract

This application provides a method and apparatus for generating collaborative handling motions of a humanoid robot with two arms, including: acquiring the phase deviation angle of the same-side hand and foot, the peak value of the object's angular velocity around the horizontal axis, and the instantaneous joint torque during the humanoid robot's walking cycle; determining an estimated value of the moment of inertia of the object being handled based on the instantaneous joint torque and the peak value of the object's angular velocity around the horizontal axis; determining the inertia level of the object being handled by comparing the estimated value of the moment of inertia with a preset threshold; determining a corresponding phase correction strategy for different inertia levels of the objects being handled; adjusting the phase deviation angle of the same-side hand and foot using the phase correction strategy; and correcting the phase correction strategy based on the monitoring results of the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation, until the phase deviation angle of the same-side hand and foot converges to a preset range.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method and apparatus for generating collaborative handling motions of a humanoid robot with two arms. Background Technology

[0002] Humanoid robot bi-arm cooperative handling is a crucial research area in robotics, directly impacting its ability to safely and efficiently perform complex object handling tasks similar to human movements in industrial production, domestic services, and medical assistance. With increasing demands for robot flexibility and adaptability, bi-arm cooperative handling has become a key aspect of demonstrating a robot's overall coordination capabilities. Current methods for handling bi-arm cooperative gait mostly treat unidirectional gait (where the arm and leg on the same side swing in the same direction simultaneously) as an abnormal movement requiring immediate elimination. This gait causes continuous lateral swaying of the robot's torso in the walking direction, reducing handling stability. Therefore, existing methods typically, upon detecting unidirectional gait, quickly pull the phase difference of the arm swings back from near-zero synchronization to a half-cycle interval of alternating left and right swings, forcibly restoring the normal coordinated rhythm of the right leg stepping forward when the left arm swings forward, and the left leg stepping forward when the right arm swings forward. This approach is generally effective under empty-handed walking or light-load conditions, but the situation changes significantly when the robot is handling objects with large rotational inertia using both arms. These objects with large moments of inertia include long rod-shaped tools, heavy containers, or irregularly shaped loads, which exhibit strong resistance to changes in direction during movement. Rapid phase changes cause the hands to synchronously change direction, but the object being moved, due to its large moment of inertia, cannot immediately follow this change, resulting in strong angular velocity fluctuations. These fluctuations are then transmitted through the arms to the robot's torso, causing more severe pitch oscillations than in the forward-leaning state. The core of this phenomenon lies in the direct coupling between the moment of inertia of the object being moved and the phase adjustment speed of the arms. Objects with larger moments of inertia are more resistant to sudden changes in arm movement. When phase correction is too rapid, the resistance from the object's inertia forces the robot's torso to withstand greater transient torque impacts. Conversely, if forward-leaning is not corrected at all, smaller pitch disturbances accumulate due to the continuous unidirectional swinging of the limbs on the same side. Although the sources of torso oscillation differ in these two states, both ultimately manifest as instability in the object's posture during handling. For example, when handling open containers filled with liquid, the continuous small-amplitude shaking caused by turning in the same direction can easily cause the liquid to slowly tilt and overflow, while a forced and rapid phase correction can cause violent shaking or even large-scale spillage due to inertial impact. When handling precision instruments, the former causes long-term cumulative damage from micro-vibrations, while the latter brings a short-term large impact that directly damages the internal structure. Therefore, how to avoid continuous attitude disturbances caused by turning in the same direction while preventing transient impact oscillations caused by phase change correction when handling objects with large rotational inertia has become a key problem that humanoid robots must address in collaborative handling tasks. Summary of the Invention

[0003] This invention provides a method for generating collaborative handling actions of a humanoid robot with two arms, mainly including: The phase deviation angle of the same side hand and foot, the peak value of the object's angular velocity around the horizontal axis, and the instantaneous joint torque are obtained during the walking cycle of the humanoid robot. The estimated value of the rotational inertia of the transported object is determined based on the instantaneous joint torque and the peak value of the object's angular velocity around the horizontal axis. The inertia level of the transported object is determined by comparing the estimated moment of inertia with a preset threshold. For different inertia levels of the transported objects, determine the corresponding phase correction strategies; After adjusting the phase deviation angle of the same-side hand and foot using a phase correction strategy, the phase correction strategy is modified based on the monitoring results of the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation, until the phase deviation angle of the same-side hand and foot converges to a preset range.

[0004] Furthermore, the phase deviation angle of the same-side hand and foot, the peak value of the object's angular velocity around the horizontal axis, and the instantaneous joint torque are obtained during the humanoid robot's walking cycle, including: The phase angle of the left arm swing and the phase angle of the left leg step are obtained during the walking cycle. The difference between the phase angle of the left arm swing and the phase angle of the left leg step is calculated by the phase difference calculation method to obtain the phase deviation angle of the same side hand and foot. The inertial measurement unit collects the angular velocity time series data of the transported object around the horizontal axis, and the absolute maximum value of the angular velocity time series data is taken as the peak value of the object's angular velocity around the horizontal axis. The instantaneous joint torque is calculated by reading the real-time current value and reduction ratio value of the dual-arm joint drive motor.

[0005] Furthermore, determining the estimated moment of inertia of the transported object based on the instantaneous joint torque and the peak value of the object's angular velocity about the transverse axis includes: The angular acceleration value is obtained by differentiating the peak angular velocity of the object about the horizontal axis in the time domain. The moment of inertia of the transported object is estimated by dividing the instantaneous joint torque by the angular acceleration value.

[0006] Furthermore, the step of determining the inertia level of the transported object by comparing the estimated moment of inertia with a preset threshold includes: The estimated moment of inertia is obtained and compared with a preset large inertia threshold. If the estimated moment of inertia is less than the large inertia threshold, the object being transported is determined to be of the ordinary inertia level. If the estimated moment of inertia is greater than or equal to the large inertia threshold, the object being transported is determined to be of the large inertia sensitivity level.

[0007] Furthermore, determining the corresponding phase correction strategy for different inertia levels of the transported objects includes: for the ordinary inertia level, adjusting the phase deviation angle of the same-side hand and foot to a preset range using a standard correction rate; For the aforementioned high inertia sensitivity level, the evaluation process for the upper limit of the phase correction rate is initiated.

[0008] Furthermore, for the aforementioned high inertia sensitivity level, an evaluation process for the upper limit of the phase correction rate is initiated, including: for the transported object with a high inertia sensitivity level, obtaining the estimated value of the rotational inertia and a preset angular velocity safety threshold. The maximum permissible angular acceleration is obtained by dividing the angular velocity safety threshold by the walking cycle duration; Based on the maximum angular acceleration and the estimated moment of inertia, calculate the maximum driving torque that the two arms can apply in a single phase adjustment; Based on the ratio of the maximum driving torque to the torque output capacity of the shoulder joints of both arms, the maximum allowable correction amplitude for a single phase adjustment is determined, which serves as the upper limit of the phase correction rate.

[0009] Furthermore, the upper limit of the phase correction rate is corrected based on the monitoring results of the peak angular velocity of the object around the transverse axis and the amplitude of the torso pitch oscillation, including: At the end of each walking cycle, the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation are collected; The peak value of the object's angular velocity around the horizontal axis is compared with a preset angular velocity safety threshold, and the pitch oscillation amplitude of the torso is compared with a preset pitch safety threshold. If either the peak value of the object's angular velocity around the horizontal axis or the amplitude of the torso's pitch oscillation exceeds the corresponding threshold, then the upper limit of the phase correction rate is multiplied by a preset attenuation coefficient to obtain the corrected upper limit of the phase correction rate. Based on the revised upper limit of the phase correction rate, the phase deviation angles of the same side hands and feet are redivided equally to obtain the updated intermediate phase target.

[0010] Furthermore, after applying the updated intermediate phase target to the motion trajectory of the shoulder joints of both arms, the difference between the current ipsilateral hand and foot phase deviation angle and the preset range is calculated. If the difference between the current ipsilateral hand and foot phase deviation angle and the preset range is less than the preset convergence threshold, then it is determined that the ipsilateral hand and foot phase deviation angle has converged to the preset range. If the difference between the current ipsilateral hand and foot phase deviation angle and the preset range is greater than or equal to the preset convergence threshold, then the ipsilateral hand and foot phase deviation angle will continue to be adjusted according to the upper limit of the phase correction rate.

[0011] This invention provides a humanoid robot dual-arm cooperative handling motion generation device, mainly comprising: The data acquisition module is used to acquire the phase deviation angle of the same-side hand and foot, the peak value of the object's angular velocity around the horizontal axis, and the instantaneous joint torque during the humanoid robot's walking cycle. The inertia estimation module is used to determine the estimated value of the rotational inertia of the transported object based on the instantaneous joint torque and the peak value of the object's angular velocity about the horizontal axis. The level determination module is used to determine the inertia level of the transported object by comparing the estimated value of rotational inertia with a preset threshold. The phase adjustment module is used to determine the corresponding phase correction strategy for different inertia levels of the transported objects. After adjusting the phase deviation angle of the same-side hand and foot using the phase correction strategy, the phase correction strategy is corrected based on the monitoring results of the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation, until the phase deviation angle of the same-side hand and foot converges to the preset range.

[0012] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a method and apparatus for generating collaborative handling motions of a humanoid robot with two arms. Addressing the problem in scenarios where humanoid robots handling large-inertia sensitive items such as liquid containers or precision instruments experience significant peak angular velocity around the horizontal axis and torso pitch oscillations due to phase deviations of the same-side arms and legs, easily leading to liquid spillage or damage to precision instruments from transient impacts, this invention accurately estimates the rotational inertia of the object being handled by real-time acquisition of the phase deviation angle of the same-side arms and legs, the peak angular velocity of the object around the horizontal axis, the amplitude of torso pitch oscillations, and the instantaneous joint torque calculated based on joint current and deceleration ratio. This estimation is combined with the dynamic relationship between torque and angular acceleration to achieve adaptive assessment of the inertia level, and then compared with a preset large inertia threshold. For ordinary inertia items, the standard rate is directly used to correct phase deviation. However, for high-inertia sensitive items, the maximum allowable correction amplitude for a single operation is calculated based on the estimated moment of inertia and the current phase deviation, ensuring that the peak angular velocity of the object does not exceed the safe range. This determines the safe upper limit of the phase correction rate. Subsequently, the phase deviation is progressively decomposed into multiple intermediate phase targets, which are sequentially applied to the joint trajectories of the two arms to achieve a smooth transition. During the process, the peak angular velocity and trunk oscillation amplitude are continuously monitored, and the rate upper limit is adjusted in real time according to transient impact changes until the phase completely converges to the normal range. This invention effectively solves the contradiction between safety and stability in the handling of high-inertia sensitive items, significantly reduces the risk of shaking, spillage, and impact damage, and improves the adaptability and reliability of humanoid robots in collaborative handling in complex scenarios. Attached Figure Description

[0013] Figure 1 This is a flowchart of a method for generating collaborative handling motions of a humanoid robot with two arms according to the present invention.

[0014] Figure 2This is a schematic diagram of the structure of a humanoid robot dual-arm collaborative handling motion generation device according to the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0016] like Figures 1-2 This embodiment of a method and apparatus for generating collaborative handling actions of a humanoid robot with two arms may specifically include: Step S101: Obtain the phase deviation angle of the same side hand and foot, the peak value of the object's angular velocity around the horizontal axis, and the instantaneous joint torque during the humanoid robot's walking cycle. Determine the estimated value of the rotational inertia of the transported object based on the instantaneous joint torque and the peak value of the object's angular velocity around the horizontal axis.

[0017] The left arm swing phase angle and the left leg stepping phase angle are acquired during the humanoid robot's walking cycle. A phase difference calculation method is used to calculate the difference between these two angles, obtaining the phase deviation angle of the same-side hand and foot. Simultaneously, the start and end times of the current walking cycle are recorded as the cycle time window. Within this window, inertial measurement units installed at the shoulder and elbow joints of both arms collect the angular velocity time-series data of the transported object around the horizontal axis. The values ​​of each sampling point in the angular velocity time-series data are iterated, and the value of the sampling point with the largest absolute value is taken as the peak value of the object's angular velocity around the horizontal axis. Simultaneously, the pitch angle change curve is read from the torso attitude sensor, and the difference between the maximum and minimum values ​​in the pitch angle change curve is taken as the torso pitch oscillation amplitude. The real-time current value of the drive motor of each joint of the dual arms and the reduction ratio of the corresponding reducer are read. The real-time current value is multiplied by the torque constant marked on the motor nameplate to obtain the motor output torque. The motor output torque is then multiplied by the reduction ratio to obtain the instantaneous joint torque. The peak angular velocity of the object around the horizontal axis is differentiated in the time domain to obtain the angular acceleration value. The instantaneous joint torque is divided by the angular acceleration value to determine the estimated value of the moment of inertia of the transported object.

[0018] In one implementation, during walking, the humanoid robot's left arm and left leg are driven by independent joints, and their motion trajectories exhibit periodic characteristics. The left arm swing phase angle represents the angular position of the left arm in a complete swing cycle, while the left leg stepping phase angle represents the angular position of the left leg in a complete stepping cycle. By calculating the difference between the left arm swing phase angle and the left leg stepping phase angle, the phase deviation angle of the ipsilateral hand and foot can be obtained. This deviation angle reflects the degree of synchronization of the ipsilateral limb movements.

[0019] Specifically, when the robot walks while holding and carrying an object, inertial measurement units (IMUs) installed at the shoulder and elbow joints of its arms continuously collect data on the object's motion. Each IMU includes a three-axis gyroscope, capable of measuring the object's angular velocity around its horizontal axis in real time. Within the specified time window, the collected angular velocity time-series data contains multiple sampling points, each recording the angular velocity value at the corresponding moment. By iterating through all sampling points in the angular velocity time-series data, the absolute values ​​of each sampling point are compared, and the sampling point with the largest absolute value is selected as the peak value of the object's angular velocity around its horizontal axis. This peak value reflects the maximum angular disturbance intensity experienced by the object during the walking cycle.

[0020] For example, the torso attitude sensor employs an inertial measurement unit (IMU) mounted in the middle of the robot's torso, which continuously outputs the torso's attitude angle in the pitch direction. Within the periodic time window, the pitch angle variation curve exhibits periodic fluctuation characteristics, where the difference between the maximum and minimum values ​​is the torso pitch oscillation amplitude.

[0021] In one embodiment, the drive motors for each joint of the dual arms are servo motors, whose drivers have built-in current sensors that can provide real-time feedback on the current flowing through the motor assembly. The torque constant, an inherent parameter of the motor and marked on the motor nameplate, represents the output torque generated per unit current, measured in Newton-meters per ampere. Multiplying the real-time current value by the torque constant yields the motor output torque. Since a reducer is installed at the joint, the motor output torque is amplified by the reducer before being transmitted to the joint output end; the amplification factor is the reduction ratio. Multiplying the motor output torque by the reduction ratio yields the instantaneous joint torque acting on the joint output end.

[0022] It should be noted that the rate of change of the peak angular velocity of the object around its horizontal axis in the time domain is the angular acceleration. The difference between the peak angular velocities in two adjacent sampling periods, divided by the sampling period duration, yields the angular acceleration value. According to the fundamental relationship of rigid body rotational dynamics, the torque acting on a rigid body is equal to the product of its moment of inertia and angular acceleration. In this embodiment, the instantaneous joint torque is transmitted to the object being transported through both arms, constituting the external torque driving the object to rotate around its horizontal axis. Dividing the instantaneous joint torque as the numerator and the angular acceleration value as the denominator yields an estimated value of the object's moment of inertia, expressed in kilograms per square meter, used to characterize the object's resistance to changes in angular velocity.

[0023] Step S102: By comparing the estimated moment of inertia with a preset threshold, the inertia level of the transported object is determined.

[0024] The estimated moment of inertia of the object being transported is obtained and compared with a preset large inertia threshold. The estimated moment of inertia is then compared with the large inertia threshold. If the estimated moment of inertia is less than the large inertia threshold, the object is determined to be of the normal inertia level. If the estimated moment of inertia is greater than or equal to the large inertia threshold, the object is determined to be of the large inertia sensitive level.

[0025] The preset high inertia threshold is pre-set based on the load capacity and walking stability requirements of the humanoid robot's arms. When the rotational inertia of the object being transported is large, the inertial resistance experienced by the arms during phase adjustment increases accordingly. If an excessively fast correction rate is used, the object's inertia will impact the robot's torso. The high inertia threshold serves as a boundary value between ordinary inertia levels and high inertia-sensitive levels, distinguishing different phase correction processing methods.

[0026] Step S103: Determine the corresponding phase correction strategy for different inertia levels of transported objects.

[0027] When the object being transported is of a normal inertia level, a preset standard correction rate is used to adjust the phase deviation angle of the same-side hands and feet to the normal alternating phase range within a single walking cycle. For objects of a high inertia sensitivity level, an assessment of the safe upper limit of the phase correction rate is performed. The estimated moment of inertia of the object being transported is obtained along with a preset safe angular velocity threshold, which represents the maximum allowable peak value of the object's angular velocity around its horizontal axis. The maximum allowable angular acceleration is obtained by dividing the safe angular velocity threshold by the duration of a single walking cycle. Based on the estimated moment of inertia and the maximum angular acceleration, the maximum allowable driving torque applied by both arms in a single phase adjustment is obtained by multiplying the maximum angular acceleration by the estimated moment of inertia. Based on the ratio of the maximum driving torque to the torque output capacity of the shoulder joints of both arms, the maximum allowable angle adjustment of the shoulder joints within a single walking cycle is obtained. The maximum angle adjustment is converted to the same-side hand and foot phase space to obtain the maximum allowable correction amplitude for a single phase adjustment, and this maximum correction amplitude is used as the upper limit of the single correction amplitude. The upper limit of the single correction amplitude is compared with the phase deviation angle of the same side hand and foot. If the phase deviation angle of the same side hand and foot is greater than the upper limit of the single correction amplitude, the upper limit of the single correction amplitude is used as the safe upper limit of the phase correction rate. If the phase deviation angle of the same side hand and foot is less than or equal to the upper limit of the single correction amplitude, the phase deviation angle of the same side hand and foot is used as the safe upper limit of the phase correction rate.

[0028] The standard correction rate refers to the angular change in the robot's adjustment of the phase deviation angle of the same-side hand and foot within a single walking cycle. The normal alternating phase range refers to the phase difference interval between the left arm swing and the left leg step, maintaining a half-cycle interval. Within this range, the left arm swings forward corresponding to the right leg stepping forward, forming a coordinated movement rhythm between the arms and legs. When the object being transported is of a normal inertia level, its resistance to phase abrupt changes is weak; the standard correction rate can adjust the phase deviation angle to the normal alternating phase range within a single walking cycle. When the object being transported is of a high inertia-sensitive level, directly using the standard correction rate will cause trunk pitch oscillations; in this case, the evaluation process for the safe upper limit of the phase correction rate is initiated. In one embodiment, the angular velocity safety threshold is a pre-set boundary value based on the type of object being transported. When the object being transported is an open container filled with liquid, the liquid is prone to overflow during shaking; therefore, the angular velocity safety threshold is set to a smaller value. When the object being transported is a precision instrument, the internal components are sensitive to vibration and impact; therefore, the angular velocity safety threshold is also set to a smaller value. This threshold represents the upper limit of the allowable peak value of the object's angular velocity around the horizontal axis. Exceeding this limit indicates a risk of instability during the handling process.

[0029] Specifically, the maximum permissible angular acceleration is obtained by dividing the angular velocity safety threshold by the duration of a single walking cycle. Angular acceleration represents the rate of change of angular velocity per unit time. When the transported object accelerates from a stationary state to the angular velocity corresponding to the angular velocity safety threshold within one walking cycle, the average angular acceleration it experiences is the maximum angular acceleration. The duration of a single walking cycle is determined by the robot's current step frequency; the faster the step frequency, the shorter the cycle duration, and the greater the corresponding maximum angular acceleration.

[0030] It should be noted that, according to the basic principles of rigid body rotational dynamics, the torque acting on a rigid body is equal to the product of its moment of inertia and angular acceleration. The maximum driving torque that the arms are allowed to apply in a single phase adjustment is obtained by multiplying the estimated moment of inertia of the object being transported by the maximum angular acceleration. This torque represents the torsional force transmitted by the arms to the object being transported through the gripping point; the larger the torque, the more severe the angular disturbance experienced by the object. Objects with larger moments of inertia produce smaller angular accelerations under the same torque; therefore, under the same angular velocity safety threshold constraint, the maximum allowable driving torque for objects with large moments of inertia is correspondingly increased.

[0031] In one possible implementation, the bi-arm shoulder joint torque output capability refers to the maximum torque value that the shoulder joint drive motor can continuously output under rated operating conditions. Comparing the maximum driving torque with the bi-arm shoulder joint torque output capability, when the maximum driving torque is less than the shoulder joint torque output capability, it indicates that the shoulder joint has sufficient drive margin. In this case, the maximum allowable angle adjustment of the bi-arm shoulder joint within a single walking cycle is determined by the inertia constraint of the object being transported. When the maximum driving torque is greater than the shoulder joint torque output capability, it indicates that the shoulder joint drive capability has become a bottleneck. In this case, the maximum angle adjustment is determined by the output capability of the shoulder joint itself. Further, the ipsilateral hand-foot phase space is a one-dimensional angle space describing the difference between the left arm swing phase and the left leg stepping phase. Changes in the swing angle of the bi-arm shoulder joint directly affect the phase relationship between the arms and legs, but the relationship is not a simple one-to-one correspondence. When converting the maximum angle adjustment to the ipsilateral hand-foot phase space, the maximum allowable correction amplitude for a single phase adjustment is obtained by calculating based on the geometric correspondence between the arm swing amplitude and the walking gait cycle.

[0032] For example, when the phase deviation angle of the ipsilateral hand and foot is greater than the upper limit of the single correction amplitude, it indicates that the current phase deviation cannot be corrected within a single walking cycle. In this case, the upper limit of the single correction amplitude is used as the safe upper limit of the phase correction rate, and multiple walking cycles are needed to gradually complete the phase adjustment. When the phase deviation angle of the ipsilateral hand and foot is less than or equal to the upper limit of the single correction amplitude, it indicates that the current phase deviation can be corrected within a single walking cycle. In this case, the phase deviation angle of the ipsilateral hand and foot itself is used as the safe upper limit of the phase correction rate.

[0033] It is understandable that the determination process of the safe upper limit of the phase correction rate takes into account the rotational inertia characteristics of the transported object, the safety constraints of angular velocity, and the driving capability of the shoulder joint. Under the premise of ensuring that the peak value of the transported object's angular velocity around the horizontal axis does not exceed the safe range, the maximum allowable correction amplitude for a single phase adjustment is determined.

[0034] Step S104: After adjusting the phase deviation angle of the same-side hand and foot using a phase correction strategy, the phase correction strategy is corrected based on the monitoring results of the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation until the phase deviation angle of the same-side hand and foot converges to a preset range.

[0035] Based on the safety upper limit of the phase correction rate and the phase deviation angle of the ipsilateral hand and foot, the total number of correction cycles is obtained by dividing the ipsilateral hand and foot phase deviation angle by the safety upper limit of the phase correction rate. The ipsilateral hand and foot phase deviation angle is then divided equally according to the total number of correction cycles to obtain multiple intermediate phase targets in stages. These multiple intermediate phase targets are applied sequentially to the movement trajectory of the shoulder joints of both arms in chronological order. At the beginning of each walking cycle, the currently pending intermediate phase target is read, and the starting and ending angles of the arm and shoulder joint swing are adjusted according to the currently pending intermediate phase target. At the end of each walking cycle, the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation are collected. The peak value of the object's angular velocity around the horizontal axis is compared with a preset angular velocity safety threshold, and the amplitude of the torso pitch oscillation is compared with a preset pitch safety threshold. If either value exceeds the corresponding threshold, the safety upper limit of the phase correction rate is multiplied by a preset attenuation coefficient to obtain a corrected safety upper limit of the phase correction rate. The remaining ipsilateral hand and foot phase deviation angles are then re-divided equally according to the corrected safety upper limit of the phase correction rate to obtain updated intermediate phase targets. After applying the current intermediate phase target, calculate the difference between the current ipsilateral hand and foot phase deviation angle and the normal alternating phase range. If the difference is less than a preset convergence threshold, it is determined that the ipsilateral hand and foot phase deviation angle has converged to the normal alternating phase range.

[0036] In one implementation, the upper limit of the phase correction rate represents the maximum allowable change in phase angle within each walking cycle. When the phase deviation angle of the ipsilateral hand and foot is large, completing all corrections within a single walking cycle would apply excessive angular acceleration to the object being transported, causing severe angular velocity fluctuations around the object's horizontal axis. By dividing the ipsilateral hand and foot phase deviation angle by the upper limit of the phase correction rate, the number of walking cycles required to complete all corrections can be obtained; this number represents the total number of correction cycles.

[0037] Specifically, the process of dividing the phase deviation angle of the same side hand and foot equally according to the total number of correction cycles is to evenly distribute the total deviation angle into each walking cycle.

[0038] For example, if the phase deviation angle of the same-side hand and foot is 90 degrees, and the safe upper limit of the phase correction rate is 15 degrees per cycle, then the total number of correction cycles is six cycles, and the intermediate phase target of each cycle is reduced by 15 degrees compared to the previous cycle. This equal division method allows the phase of the arm swing to gradually transition to the normal alternating phase range at a constant rate.

[0039] In one embodiment, when the intermediate phase target is applied to the motion trajectory of the shoulder joints of both arms, the robot controller reads the current intermediate phase target value to be executed at the beginning of each walking cycle and adjusts the position commands of the shoulder joint servo motors accordingly. The swing start angle refers to the initial position of the arms at the beginning of the walking cycle, and the end angle refers to the target position of the arms at the end of the walking cycle. By adjusting the phase difference between the start angle and the end angle, the arm swing gradually transitions from a clockwise swing to an alternating left-right swing.

[0040] It should be noted that the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation are acquired at the end of each walking cycle. An inertial measurement unit installed at the end effector of both arms continuously records the angular velocity change of the transported object, and the sampling point with the maximum absolute value of the angular velocity within that walking cycle is taken as the peak value of the object's angular velocity around the horizontal axis. An attitude sensor in the middle of the torso continuously records the curve of the torso pitch angle change, and the peak-to-peak value of this curve within the current walking cycle is taken as the amplitude of the torso pitch oscillation. These two values ​​reflect the actual impact of the current phase adjustment on the transport stability.

[0041] In one possible implementation, both the angular velocity safety threshold and the pitch safety threshold are preset boundary values. The angular velocity safety threshold is determined based on the type of object being transported. For open containers filled with liquid, this threshold is set to a smaller value to prevent liquid sloshing and spillage; for precision instruments, this threshold is also set to a smaller value to prevent internal components from being damaged by impact. The pitch safety threshold is determined based on the stability margin of the robot's torso. When the pitch oscillation amplitude exceeds this threshold, the robot is at risk of instability and falling. Furthermore, when the peak value of the object's angular velocity around the horizontal axis or the pitch oscillation amplitude of the torso exceeds the corresponding threshold, it indicates that the current phase correction rate is too fast, and the object being transported or the robot's torso has been subjected to excessive transient impact. In this case, the safe upper limit of the phase correction rate is multiplied by a preset attenuation coefficient to obtain the corrected safe upper limit of the phase correction rate. The attenuation coefficient is a positive number less than one, for example, a value of 0.8, indicating that the correction rate is reduced to 80% of the original. After reducing the correction rate, the remaining uncompleted ipsilateral hand and foot phase deviation angles are re-divided equally according to the corrected phase correction rate safety upper limit to obtain an updated intermediate phase target sequence. Subsequent walking cycles continue to perform phase adjustment according to the updated intermediate phase target sequence.

[0042] It is understood that the normal alternating phase range refers to the interval where the phase difference between the left arm swing and the left leg step is around half a cycle. When the difference between the phase deviation angle of the same-side hand and foot and the center value of this interval is less than a preset convergence threshold, the arm swing has returned to the normal left-right alternating rhythm, and the phase correction process is terminated at this time. The convergence threshold is usually set to a small angle value, indicating that the residual phase deviation will no longer have a significant impact on the stability of the transport. The entire progressive phase correction process eliminates the unilateral movement state while ensuring the stability of the transported object's posture through a closed-loop feedback mechanism.

[0043] This invention provides a humanoid robot dual-arm cooperative handling motion generation device, mainly comprising: The data acquisition module is used to acquire the phase deviation angle of the same-side hand and foot, the peak value of the object's angular velocity around the horizontal axis, and the instantaneous joint torque during the humanoid robot's walking cycle. The inertia estimation module is used to determine the estimated value of the rotational inertia of the transported object based on the instantaneous joint torque and the peak value of the object's angular velocity about the horizontal axis. The level determination module is used to determine the inertia level of the transported object by comparing the estimated value of rotational inertia with a preset threshold. The phase adjustment module is used to determine the corresponding phase correction strategy for different inertia levels of the transported objects. After adjusting the phase deviation angle of the same-side hand and foot using the phase correction strategy, the phase correction strategy is corrected based on the monitoring results of the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation, until the phase deviation angle of the same-side hand and foot converges to the preset range.

[0044] If the technical solution of this application involves personal information, the product using this solution has clearly informed the user of the personal information processing rules and obtained the user's voluntary consent before processing the personal information. If sensitive personal information is involved, the user's separate consent has been obtained before processing, and the "express consent" requirement is met. For example, a clear sign is placed at the collection device such as a camera to inform the user that they have entered the collection area, and the user's voluntary entry is considered as consent; or the processing device clearly indicates the processing rules and obtains authorization through pop-up windows or by asking the user to upload information themselves. The personal information processing rules include the processor, the purpose of processing, the processing method, and the types of personal information.

[0045] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for generating a cooperative carrying motion of a humanoid robot's both arms, characterized by, The method comprises: acquiring ipsilateral hand and foot phase deviation angle, object roll angular velocity peak value and instantaneous joint torque in a walking cycle of a humanoid robot, determining the moment of inertia estimate value of the carried object according to the instantaneous joint torque and the object roll angular velocity peak value; comparing the moment of inertia estimate value with a preset threshold value to determine the inertia grade of the carried object; determining the corresponding phase correction strategy according to the inertia grade of different carried objects; after adjusting the ipsilateral hand and foot phase deviation angle by using the phase correction strategy, correcting the phase correction strategy according to the monitoring result of the object roll angular velocity peak value and the trunk pitch oscillation amplitude until the ipsilateral hand and foot phase deviation angle converges to a preset range.

2. The method according to claim 1, wherein The acquisition of the ipsilateral hand and foot phase deviation angle, the object roll angular velocity peak value and the instantaneous joint torque in the walking cycle of the humanoid robot comprises: acquiring the left arm swing phase angle and the left leg step phase angle in the walking cycle, and performing difference operation on the left arm swing phase angle and the left leg step phase angle by using a phase difference calculation method to obtain the ipsilateral hand and foot phase deviation angle; acquiring the angular velocity time series data of the carried object in the roll direction by using an inertial measurement unit, and taking the maximum absolute value of the angular velocity time series data as the object roll angular velocity peak value; reading the real-time current value and the reduction ratio value of the double-arm joint driving motor to calculate the instantaneous joint torque.

3. The method for generating collaborative handling motions of a humanoid robot with two arms as described in claim 1, characterized in that, The determination of the moment of inertia estimate value of the carried object according to the instantaneous joint torque and the object roll angular velocity peak value comprises: deriving the object roll angular velocity peak value in the time domain to obtain the angular acceleration value; calculating the moment of inertia estimate value of the carried object by dividing the instantaneous joint torque by the angular acceleration value.

4. The method of claim 1, wherein the method further comprises: The comparison of the moment of inertia estimate value with a preset threshold value to determine the inertia grade of the carried object comprises: comparing the moment of inertia estimate value with a preset large inertia threshold value; if the moment of inertia estimate value is less than the large inertia threshold value, it is determined that the carried object is of ordinary inertia grade; if the moment of inertia estimate value is greater than or equal to the large inertia threshold value, it is determined that the carried object is of large inertia sensitive grade.

5. The method according to claim 4, wherein the method further comprises: determining a target position of the object; and determining a target position of the robot arm based on the target position of the object and the position of the object. The determination of the corresponding phase correction strategy according to the inertia grade of different carried objects comprises: for the ordinary inertia grade, adjusting the ipsilateral hand and foot phase deviation angle to a preset range by using a standard correction rate; for the large inertia sensitive grade, entering the evaluation process of the upper limit of the phase correction rate.

6. The method according to claim 5, wherein the method is characterized by: For the large inertia sensitive grade, the evaluation process of the upper limit of the phase correction rate comprises: for the carried object of large inertia sensitive grade, acquiring the moment of inertia estimate value and a preset angular velocity safety threshold value; dividing the angular velocity safety threshold value by the walking cycle time to obtain the maximum allowed angular acceleration; calculating the maximum driving torque that the double arms are allowed to apply in a single phase adjustment according to the maximum angular acceleration and the moment of inertia estimate value; determining the maximum correction amplitude allowed in a single phase adjustment as the upper limit of the phase correction rate according to the ratio of the maximum driving torque to the shoulder joint torque output capacity of the double arms.

7. The method for generating collaborative handling motions of a humanoid robot with two arms as described in claim 1, characterized in that, The step of correcting the upper limit of the phase correction rate based on the monitoring results of the peak angular velocity of the object around the horizontal axis and the amplitude of the torso pitch oscillation includes: At the end of each walking cycle, the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation are collected; The peak value of the object's angular velocity around the horizontal axis is compared with a preset angular velocity safety threshold, and the pitch oscillation amplitude of the torso is compared with a preset pitch safety threshold. If either the peak value of the object's angular velocity around the horizontal axis or the amplitude of the torso's pitch oscillation exceeds the corresponding threshold, then the upper limit of the phase correction rate is multiplied by a preset attenuation coefficient to obtain the corrected upper limit of the phase correction rate. Based on the revised upper limit of the phase correction rate, the phase deviation angles of the same side hands and feet are redivided equally to obtain the updated intermediate phase target.

8. The method according to claim 7, wherein the method further comprises: determining a target position of the object; and determining a target position of the robot arm based on the target position of the object and the position of the object. After applying the updated intermediate phase target to the motion trajectory of the shoulder joints of both arms, the difference between the current ipsilateral hand and foot phase deviation angle and the preset range is calculated. If the difference between the current ipsilateral hand and foot phase deviation angle and the preset range is less than the preset convergence threshold, then it is determined that the ipsilateral hand and foot phase deviation angle has converged to the preset range. If the difference between the current ipsilateral hand and foot phase deviation angle and the preset range is greater than or equal to the preset convergence threshold, then the ipsilateral hand and foot phase deviation angle will continue to be adjusted according to the upper limit of the phase correction rate.

9. A device for generating collaborative handling motions of a humanoid robot with two arms, characterized in that, The device includes: The data acquisition module is used to acquire the phase deviation angle of the same-side hand and foot, the peak value of the object's angular velocity around the horizontal axis, and the instantaneous joint torque during the humanoid robot's walking cycle. The inertia estimation module is used to determine the estimated value of the rotational inertia of the transported object based on the instantaneous joint torque and the peak value of the object's angular velocity about the horizontal axis. The level determination module is used to determine the inertia level of the transported object by comparing the estimated value of rotational inertia with a preset threshold. The phase adjustment module is used to determine the corresponding phase correction strategy for different inertia levels of the transported objects. After adjusting the phase deviation angle of the same-side hand and foot using the phase correction strategy, the phase correction strategy is corrected based on the monitoring results of the peak value of the object's angular velocity around the horizontal axis and the amplitude of the torso pitch oscillation, until the phase deviation angle of the same-side hand and foot converges to the preset range.

10. The bimanual cooperative motion generating apparatus of the humanoid robot according to Claim 9, wherein The acquisition of the phase deviation angle of the same-side hand and foot, the peak value of the object's angular velocity around the horizontal axis, and the instantaneous joint torque during the humanoid robot's walking cycle includes: The phase angle of the left arm swing and the phase angle of the left leg step are obtained during the walking cycle. The difference between the phase angle of the left arm swing and the phase angle of the left leg step is calculated by the phase difference calculation method to obtain the phase deviation angle of the same side hand and foot. The inertial measurement unit collects the angular velocity time series data of the transported object around the horizontal axis, and the absolute maximum value of the angular velocity time series data is taken as the peak value of the object's angular velocity around the horizontal axis. The instantaneous joint torque is calculated by reading the real-time current value and reduction ratio value of the dual-arm joint drive motor.

Images