Method for planning dynamic motion of a robot arm

Abstract

The present application relates to the technical field of mechanical arm, and specifically relates to a kind of bionic dynamic motion target planning method of mechanical arm, comprising the following steps: defining mechanical arm task target, and quantification is kinematics parameter;End pose is converted into joint space variable, according to joint speed, load mass, end linear velocity three variables, in motion stage, load condition, end velocity three aspects, task target is classified into several kinds of motion conditions, each motion condition corresponds a trajectory planning strategy function;With minimum running time, minimum joint jerk and minimum overshoot as target, the planning strategy function selection training based on machine learning is carried out to mechanical arm, and the most suitable trajectory planning strategy function is selected for each motion condition.The bionic dynamic motion target planning method of mechanical arm of the present application solves the problem that task target different stages require different mechanical arm joint space kinematics, dynamics constraints by motion target decomposition and variable trajectory planning strategy, and improves control effect.

Description

Technical Field

[0001] This invention relates to the field of robotic arm technology, and specifically to a biomimetic dynamic motion target planning method for robotic arms. Background Technology

[0002] To improve the operational accuracy of robotic arms and control their motion parameters, existing technologies typically perform trajectory planning in the joint space of the robotic arm to ensure smooth changes in joint angular position, velocity, and acceleration values. Joint space trajectory planning involves interpolating a series of joint angular positions, velocities, and acceleration values ​​using a smooth curve between the start and end points of the joint. Commonly used interpolation methods include linear interpolation, linear interpolation with parabolic transitions, cubic polynomial interpolation, and quintic polynomial interpolation. Joint space trajectory planning does not require kinematic calculations, has low computational cost, and avoids issues such as singularities or unreachable points. However, joint space trajectory planning cannot guarantee the pose of the interpolation points.

[0003] Currently, the motion goal of robotic arms is to quickly track a target trajectory. Trajectory planning between each path point on the target trajectory often employs the same polynomial interpolation. However, real-world tasks are not simply trajectory tracking, and a single planning method is not optimal during the execution of the same task. Therefore, the control effect achieved by current robotic arm motion planning is unsatisfactory. Summary of the Invention

[0004] To address the technical problem that existing technologies often employ the same polynomial interpolation for trajectory planning between each path point of a robotic arm's motion target trajectory, resulting in suboptimal control performance, this invention proposes a biomimetic dynamic motion target planning method for robotic arms. By decomposing the motion target and employing a variable trajectory planning strategy, this method solves the problem of varying requirements for the spatial kinematics and dynamics constraints of the robotic arm joints at different stages of the task objective, thereby improving control performance.

[0005] The technical solution of the present invention:

[0006] A biomimetic dynamic motion target planning method for a robotic arm includes the following steps:

[0007] S1: Kinematic modeling of the robotic arm, defining the task objectives of the robotic arm and quantifying them into the kinematic parameters of the robotic arm;

[0008] S2: Transform the end pose into joint space variables through kinematics. Based on the three variables of joint velocity, load mass, and end linear velocity, identify and break down the task target in terms of motion phase, load condition, and end velocity, and classify it into several motion conditions. Each motion condition corresponds to a trajectory planning strategy function.

[0009] S3: With minimum running time, minimum joint abruptness, and minimum overshoot as training objectives, the robotic arm is trained using machine learning-based planning strategy function selection to select the most suitable trajectory planning strategy function for several motion scenarios.

[0010] Furthermore, in step S1, the robotic arm is a six-degree-of-freedom serial robotic arm, and the modeling method is an improved DH model.

[0011] Furthermore, in step S2, for the motion phase, when the joint velocity... Time is defined as the beginning and end stages, and the value of the motion stage is f. s =0, when the joint velocity When, it is defined as the intermediate stage, and the value of the motion stage is f. s =1; For load conditions, when the load mass m load When the load is less than 5kg, it is defined as a small to medium load, and the load value is f. l =0, when the load mass m load When the load is ≥5kg, it is defined as a large load, and the load value is f. l =1; For the terminal velocity, when the terminal linear velocity v end When the speed is less than 0.3 m / s, it is defined as medium-low speed operation, and the terminal speed is taken as f. v =0, when the terminal linear velocity v end When the speed is ≥0.3m / s, it is defined as high-speed operation, and the terminal velocity is taken as f. v =1.

[0012] Furthermore, based on the combination of motion stage, load condition, and end velocity, there are a total of eight motion conditions. The combinations of motion stage, load condition, and end velocity for the eight motion conditions are 0, 0, 0; 0, 0, 1; 0, 1, 0; 0, 1, 1; 1, 0, 0; 1, 0, 1; 1, 1, 0; 1, 1, 1.

[0013] Furthermore, in step S2, the trajectory planning strategy function corresponding to each motion case can be any one of linear interpolation, parabolic-linear interpolation, cubic polynomial interpolation, quintic polynomial interpolation, "3-5-3" interpolation which combines cubic and quintic functions, "4-1-4" mixed interpolation, cubic B-spline interpolation, quintic B-spline interpolation, or "BB" spline interpolation.

[0014] Furthermore, in step S3, the trajectory planning strategy functions for several motion scenarios are combined to form the final trajectory planning function for the robotic arm.

[0015] Furthermore, the final robotic arm trajectory planning function

[0016]

[0017] Among them, g i ([θ1(t),…,θ6(t)]),i=1,2,…,8 represent the trajectory planning strategy function corresponding to each motion case.

[0018] After adopting the above technical solution, the biomimetic dynamic motion target planning method for robotic arms provided by this invention has the following beneficial effects compared with the prior art: This invention quantifies the task target into kinematic parameter indicators of the robotic arm, and then, following the human arm motion control approach, extracts constraints for different motion targets, designs multiple robotic arm trajectory planning modes, and decomposes the quantified task target into a combination of multiple motion targets through machine learning methods, and selects the most suitable trajectory planning strategy function. Compared with the prior art that uses the same polynomial interpolation, this invention solves the problem of different requirements for the spatial kinematics and dynamic constraints of the robotic arm joints at different stages of the task target through motion target decomposition and variable trajectory planning strategy, thereby improving the control effect. Attached Figure Description

[0019] Figure 1 This is an overall flowchart of the motion target planning method in this embodiment;

[0020] Figure 2 This is a simplified schematic diagram of the six-degree-of-freedom robotic arm in this embodiment;

[0021] Figure 3 This is a schematic diagram of the linear interpolation method in this embodiment;

[0022] Figure 4 This is a schematic diagram of the parabolic-linear interpolation method in this embodiment;

[0023] Figure 5 This is a schematic diagram of the trajectory of a cubic polynomial multipath point in this embodiment;

[0024] Figure 6 This is a schematic diagram of the cubic polynomial interpolation method in this embodiment;

[0025] Figure 7 This is a schematic diagram of the "3-5-3" interpolation method in this embodiment. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0028] In the description of this invention, it should be understood that the orientation or positional relationship indicated by directional terms such as "front, back, up, down, left, right", "horizontal, vertical, horizontal" and "top, bottom" is generally based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing this invention and simplifying the description. Unless otherwise stated, these directional terms do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this invention; the directional terms "inner" and "outer" refer to the inner and outer contours relative to the outline of each component itself.

[0029] For ease of description, spatial relative terms such as "above," "on top of," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation beyond the orientation of the device as described in the figures. For example, if the device in the figures were inverted, a device described as "above" or "on top of" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatial relative descriptions used herein will be interpreted accordingly.

[0030] Furthermore, it should be noted that the use of terms such as "first" and "second" to define components is merely for the purpose of distinguishing the corresponding components. Unless otherwise stated, the above terms have no special meaning and therefore should not be construed as limiting the scope of protection of this invention.

[0031] like Figure 1 As shown, this embodiment provides a biomimetic dynamic motion target planning method for a robotic arm, which includes the following steps:

[0032] S1: Kinematic modeling of the robotic arm, defining the task objectives of the robotic arm and quantifying them into the kinematic parameters of the robotic arm;

[0033] S2: Transform the end pose into joint space variables through kinematics. Based on the three variables of joint velocity, load mass, and end linear velocity, identify and break down the task target in terms of motion phase, load condition, and end velocity, and classify it into several motion conditions. Each motion condition corresponds to a trajectory planning strategy function.

[0034] S3: With minimum running time, minimum joint abruptness, and minimum overshoot as training objectives, the robotic arm is trained using machine learning-based planning strategy function selection to select the most suitable trajectory planning strategy function for several motion scenarios.

[0035] This embodiment first quantifies the task objective into kinematic parameters of the robotic arm. Then, mimicking the motion control approach of a human arm, it extracts constraints for different motion objectives and designs multiple trajectory planning modes for the robotic arm. Through machine learning methods, it decomposes the quantified task objective into a combination of multiple motion objectives and selects the most suitable trajectory planning strategy function. Thus, compared to existing technologies that use a single polynomial interpolation, this embodiment, by employing a biomimetic approach and using motion objective decomposition and variable trajectory planning strategies, addresses the issue of varying requirements for the spatial kinematics and dynamics constraints of the robotic arm joints at different stages of the task objective, thereby improving control performance.

[0036] Furthermore, in step S1, the robotic arm kinematics model is performed, the task objective of the robotic arm is defined, and it is quantified into the kinematic parameters of the robotic arm to break down the task, form constraints, and perform trajectory planning accordingly.

[0037] First, a kinematic model of the robotic arm is established. In this embodiment, the robotic arm is a six-degree-of-freedom serial robotic arm, specifically the Aubo i5 robotic arm, and the kinematic model is established using the modified DH model (MDH). In other embodiments, robotic arms with other degrees of freedom or other types of robots may also be used.

[0038] Figure 2 A simplified model of the robotic arm is shown, in which,

[0039] d2 = 140.5 mm

[0040] a2 = 40mm

[0041] a3 = 376mm

[0042] d4 = 19mm

[0043] d5 = 102.5 mm.

[0044] The DH parameters are shown in Table 1:

[0045] Table 1 DH Parameters

[0046]

[0047] According to forward kinematics:

[0048] Based on the improved DH model described above, the homogeneous transformation matrix can be obtained:

[0049]

[0050]

[0051]

[0052]

[0053]

[0054]

[0055]

[0056] in:

[0057] c i =cosθ i

[0058] s i =sinθ i

[0059] In Matlab, symbolic variables can be created using the syms function, making it easy to calculate forward kinematics formulas with unknowns.

[0060] According to inverse kinematics:

[0061] set up:

[0062]

[0063] make:

[0064]

[0065] remember:

[0066]

[0067] Then we have:

[0068]

[0069] That is:

[0070]

[0071] The second row and fourth column are equal on both sides, which can be solved using the auxiliary angle formula. (Since the arccos function returns values ​​between [0,π] in Matlab and C++, and the arctan function returns values ​​between (-π,π], both positive and negative solutions for arccos need to be manually added to the program.)

[0072]

[0073] Since the left and right sides of the second row and second column are equal, we can solve for:

[0074] θ5=±arccos(a x s1-a y c1)

[0075] Since the second row and third column are equal on both sides, we can solve for:

[0076]

[0077] And:

[0078]

[0079] Given that the left and right sides of the first row and fourth column are equal, and the left and right sides of the third row and fourth column are also equal, after rearranging and adding their squares, we can eliminate s² and c², and thus obtain the solution:

[0080]

[0081] m = p x c1+p y s1+d5[c6(o x c1+o y s1)+s6(n x c1+n y s1)]

[0082] n = -p z -d5(o z c6+n z s6)

[0083] Substituting θ3 back, we can solve for:

[0084]

[0085] Given that the left and right sides of the first row and second column are equal, and the left and right sides of the third row and second column are also equal, after simplification and division, we can obtain the solution:

[0086]

[0087] During programming, it is necessary to determine whether the divisor is 0 (whether it is a singularity). In addition, if the divisor is too close to 0, it may also cause pathological problems, resulting in a large error in the result.

[0088] Furthermore, in step S2, the end pose is converted into joint space variables through kinematics. Based on the three variables of joint velocity, load mass, and end linear velocity, the task target is identified and decomposed in terms of motion phase, load condition, and end velocity, and classified into several motion conditions. Each motion condition corresponds to a trajectory planning strategy function.

[0089] First, break down the task objectives into a series of pre-defined combinations of motion objectives, such as high-damping motion in the initial or high-load phase, and low-damping or even negative-damping motion in the fast-moving phase.

[0090] Specifically, the end-effector pose is converted into joint space variables through kinematics, based on joint velocity. Load mass m load terminal linear velocity v end Three variables, f, during the motion phase s Load conditions f l terminal velocity f v The task is identified, split, and classified from three aspects. The specific identification and splitting methods are shown in Tables 2.1-2.3.

[0091] Table 2.1 Stages of Movement

[0092]

[0093]

[0094] Table 2.2 Load Conditions

[0095] Range of variable values Load status <![CDATA[f l Values]]> <![CDATA[m load <5kg]]> Small to medium load 0 <![CDATA[m load ≥5kg]]> High load 1

[0096] Table 2.3 Terminal Velocity

[0097] Range of variable values terminal velocity <![CDATA[f v Values]]> <![CDATA[v end <0.3m / s]]> medium and low speed operation 0 <![CDATA[v end ≥0.3m / s]]> High-speed operation 1

[0098] During the motion phase, when the joint velocity... Time is defined as the beginning and end stages, and the value of the motion stage is f.s =0, when the joint velocity When, it is defined as the intermediate stage, and the value of the motion stage is f. s =1; For load conditions, when the load mass m load When the load is less than 5kg, it is defined as a small to medium load, and the load value is f. l =0, when the load mass m load When the load is ≥5kg, it is defined as a large load, and the load value is f. l =1; For the terminal velocity, when the terminal linear velocity v end When the speed is less than 0.3 m / s, it is defined as medium-low speed operation, and the terminal speed is taken as f. v =0, when the terminal linear velocity v end When the speed is ≥0.3m / s, it is defined as high-speed operation, and the terminal velocity is taken as f. v =1.

[0099] Thus, according to the motion stage f s Load conditions f l terminal velocity f v The possible combinations of values ​​are categorized into eight motion scenarios, with each scenario having a motion stage f. s Load conditions f l terminal velocity f v The possible combinations of values ​​are 0, 0, 0; 0, 0, 1; 0, 1, 0; 0, 1, 1; 1, 0, 0; 1, 0, 1; 1, 1, 0; 1, 1, 1.

[0100] The trajectory planning strategy function corresponding to each motion case can be any one of the following: linear interpolation, parabolic-linear interpolation, cubic polynomial interpolation, quintic polynomial interpolation, "3-5-3" interpolation which combines cubic and quintic functions, "4-1-4" mixed interpolation, cubic B-spline interpolation, quintic B-spline interpolation, or "BB" spline interpolation.

[0101] Specifically, Figure 3 A schematic diagram of linear interpolation is shown, in which... Figure 3 a is a graph showing the change of joint angle over time. Figure 3 b is the curve of joint angular velocity changing with time. Figure 3 c represents the curve of joint angle acceleration versus time. Linear interpolation refers to the fact that joint variables change completely linearly between adjacent path points. Between adjacent path points, the joint angle changes linearly with time, the joint angular velocity is constant, while the joint acceleration has large abrupt changes at both the initial and final positions. In practice, these two points require very large accelerations. However, between two adjacent path segments formed by two pairs of adjacent path points, the joint velocity may be discontinuous at the nodes, resulting in infinite acceleration.

[0102] Figure 4A schematic diagram of the parabola-linear interpolation method is shown, in which... Figure 4 a is a graph showing the change of joint angle over time. Figure 4 b is the curve of joint angular velocity changing with time. Figure 4 c represents the curve of joint angle acceleration versus time. Parabolic-linear interpolation adds a buffer segment of parabolic trajectory to both ends of the straight line, thus obtaining a smooth motion trajectory with continuous angle and velocity, while avoiding infinite acceleration. In this method, the change of joint angle is smooth, and the joint angular velocity curve is a trapezoidal curve formed by uniform acceleration, uniform velocity, and uniform deceleration segments. The joint angular acceleration is composed of line segments, and its change is not smooth, with abrupt changes that can cause vibration and impact to the joint.

[0103] Figure 5 The diagram illustrates a multipath trajectory of a cubic polynomial. The general formula for a cubic polynomial function is:

[0104] θ(t)=a0+a1t+a2t 2 +a3t 3

[0105] As can be seen, a cubic polynomial curve has four unknown coefficients a0, a1, a2, and a3, which can simultaneously satisfy four independent constraints. Generally, the angle and velocity constraints at the starting and ending points are chosen, which can ensure that the trajectory at the path points is smooth on a trajectory with multiple path points.

[0106] For any segment of the trajectory of a cubic polynomial function, its joint motion information is as follows: Figure 6 As shown, where, Figure 6 a is a graph showing the change of joint angle over time. Figure 6 b is the curve of joint angular velocity changing with time. Figure 6 c represents the curve of joint angle acceleration versus time. The joint velocity curve using a cubic polynomial function is a parabola, and the acceleration curve is a straight line, which can guarantee the position and velocity constraints at the two joint points. Higher-order curves, such as quintic curves, can further satisfy the acceleration constraints.

[0107] In addition, there are other interpolation methods, such as the "3-5-3" interpolation method that combines cubic and quintic functions, such as... Figure 7 As shown, since cubic interpolation is prone to causing sudden acceleration changes, while quintic interpolation results in larger speeds and accelerations, the “3-5-3” planning method considers segmenting the path, using cubic interpolation for the beginning and end segments, and quintic interpolation for the middle segment, and reasonably setting constraints at the connection points, which can make the time curves of joint angles, angular velocities and angular accelerations smooth and continuous, and ensure that their extreme values ​​are small.

[0108] In addition, there are other interpolation methods such as "4-1-4" hybrid interpolation, cubic B-spline interpolation, quintic B-spline interpolation, and "B--B" spline interpolation.

[0109] Further, in step S3, with the training objectives of minimum running time, minimum joint abruptness, and minimum overshoot, the robotic arm is trained using machine learning-based trajectory planning strategy function selection to choose the most suitable trajectory planning strategy function for several motion scenarios. The combination of trajectory planning strategy functions for several motion scenarios forms the final robotic arm trajectory planning function.

[0110]

[0111] Among them, g i ([θ1(t),…,θ6(t)]),i=1,2,…,8 represent the trajectory planning strategy function corresponding to each motion case.

[0112] As can be seen from the above, the biomimetic dynamic motion target planning method for robotic arms provided in this embodiment solves the problem that the spatial kinematics and dynamic constraints of the robotic arm joints are different at different stages of the task target by decomposing the motion target and using a variable trajectory planning strategy, thereby improving the control effect.

[0113] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A biomimetic dynamic motion target planning method for a robotic arm, characterized in that, Includes the following steps: S1: Kinematic modeling of the robotic arm, defining the task objectives of the robotic arm and quantifying them into the kinematic parameters of the robotic arm; S2: Transform the end pose into joint space variables through kinematics. Based on the three variables of joint velocity, load mass, and end linear velocity, identify and break down the task target in terms of motion phase, load condition, and end velocity, and classify it into several motion conditions. Each motion condition corresponds to a trajectory planning strategy function. S3: With the goals of minimizing running time, minimizing joint jerks, and minimizing overshoot, the robotic arm is trained based on machine learning to select the most suitable trajectory planning strategy function for several motion scenarios. According to the stage of exercise Load status Combinations of terminal velocity values There are eight motion scenarios, and the combinations of motion stage, load condition, and end velocity for the eight motion scenarios are 0, 0, and 0, respectively. 0、0、1；0、1、0；0、1、1； 1、0、0； 1、0、1； 1、1、0； 1、1、1； In step S3, the trajectory planning strategy functions for several motion scenarios are combined to form the final trajectory planning function for the robotic arm. The final robotic arm trajectory planning function in, These represent the trajectory planning strategy functions for each type of motion.

2. The biomimetic dynamic motion target planning method for robotic arms according to claim 1, characterized in that, In step S1, the robotic arm is a six-degree-of-freedom serial robotic arm, and the modeling method is an improved DH model.

3. The biomimetic dynamic motion target planning method for robotic arms according to claim 2, characterized in that, In step S2, during the motion phase, when the joint velocity... Time is defined as the beginning and end stages, and the value of the motion stage is taken as follows. When the joint speed Time is defined as the intermediate stage, and the value of the motion stage is taken as follows. For load conditions, when the load quality When defined as a small to medium load, the load condition is taken as a value. When the load quality When defined as a large load, the load condition is taken as a value. For the terminal velocity, when the terminal linear velocity... When defined as operating at medium to low speed, the terminal speed is taken as... When the terminal linear velocity When, it is defined as high-speed operation, and the terminal speed is taken as a value. .

4. The biomimetic dynamic motion target planning method for robotic arms according to claim 1, characterized in that, In step S2, the trajectory planning strategy function corresponding to each motion case can be any one of linear interpolation, parabolic-linear interpolation, cubic polynomial interpolation, quintic polynomial interpolation, "3-5-3" interpolation which combines cubic and quintic functions, "4-1-4" mixed interpolation, cubic B-spline interpolation, quintic B-spline interpolation, or "BB" spline interpolation.

Images