CN-121973197-A - Mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization

CN121973197ACN 121973197 ACN121973197 ACN 121973197ACN-121973197-A

Abstract

The invention discloses a mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization, which relates to the field of robot motion planning, and comprises the steps of obtaining three-dimensional point cloud data of a target object, and segmenting a smooth surface area through geometric feature analysis; the method comprises the steps of screening candidate grabbing point pairs meeting normal vector parallel constraint and friction cone constraint, generating grabbing strategies comprising target clamping force and grabbing pose based on the candidate grabbing point pairs, generating a motion track from the initial pose of the mechanical arm to the grabbing pose and optimizing joint load by introducing a path planning algorithm of a gravity moment cost function with the grabbing pose as a target, controlling the mechanical arm to move along the motion track, and automatically switching control modes according to distance and tail end force feedback until grabbing is completed. The mechanical arm grabbing device has the advantages that grabbing points are screened through geometric and physical constraints, the motion track is planned through gravity optimization, and self-adaptive variable stiffness control is achieved, so that the stability, energy efficiency and safety of grabbing operation of the mechanical arm in an unstructured environment are improved.

Inventors

ZHANG ZIMING
HU WEI
AN GANG
XU SHAOCHENG
WANG WEIGUO
LI KAI
SHAN YIMENG

Assignees

国营芜湖机械厂

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. The mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization is characterized by comprising the following steps of: s1, acquiring three-dimensional point cloud data of a target object, and segmenting a smooth surface area through geometric feature analysis; S2, screening candidate grabbing point pairs meeting normal vector parallel constraint and friction cone constraint on the smooth surface area; s3, generating a grabbing strategy comprising a target clamping force and grabbing pose based on the candidate grabbing point pairs; S4, taking the grabbing pose as a target, and generating a motion track which is from the initial pose of the mechanical arm to the grabbing pose and has optimized joint load through a path planning algorithm introducing a gravity moment cost function; And S5, controlling the mechanical arm to move along the movement track, and automatically switching a control mode according to the distance and the tail end force feedback until the grabbing is completed.
2. The method for adaptively grabbing a mechanical arm based on geometric constraint and gravity optimization according to claim 1, wherein in S1, segmenting the smooth surface area through geometric feature analysis comprises: Calculating a surface normal vector of each point in the point cloud based on a principal component analysis method; Calculating the local curvature of each point; based on a region growing algorithm, using local curvature as a growing criterion, and enabling the curvature change rate to be smaller than a preset threshold value Is clustered into the same geometric smooth surface.
3. The mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization according to claim 1, wherein the step S2 comprises: searching for a point pair with parallel and opposite normal vectors of the surface in the segmented smooth surface area, wherein the error of the included angle of the normal vectors of the two points is smaller than a preset angle ; Establishing a friction cone model of a contact point of each point pair, and judging whether the force application direction of the fingertip of the mechanical arm is positioned in the friction cone, wherein the constraint conditions of the friction cone are as follows: , To form an angle between the direction of the applied force and the normal vector of the contact surface, Is a preset static friction coefficient; and judging the point pairs screened by the normal vector parallel constraint and the friction cone constraint as candidate grabbing point pairs.
4. The mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization according to claim 1, wherein the step S3 comprises: Estimating the mass of the target object according to the point cloud characteristics of the target object, and calculating the target clamping force by combining the gravity acceleration and a preset safety coefficient; and determining the midpoint coordinates and the connecting direction of the connecting lines of the two contact points in the candidate grabbing point pair as grabbing positions, wherein the grabbing positions comprise the positions and the postures of the end effector.
5. The mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization according to claim 1, wherein in S4, the path planning algorithm is an improved RRT algorithm, and an incremental cost function of path expansion is: Wherein, the A state vector representing the parent node in joint space, A state vector representing the child node in joint space, For the geometric path smoothness cost term, The weight moment cost term is used for the weight moment cost term, Is a weight coefficient.
6. The mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization according to claim 5, wherein the geometric path smoothness cost term is Weighted euclidean distance is used: Wherein n is the total number of joints of the mechanical arm, And The angle value of the ith joint in the parent node and child node state vectors respectively, Is the weighting coefficient of the i-th joint.
7. The mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization according to claim 5, wherein the weight moment cost term is as follows Based on static gravity compensation model calculations, comprising: calculating the configuration of the mechanical arm Under, the joint moment vector required by overcoming the gravity of the connecting rod and the gravity of the tail end load : Wherein, the Is a gravity term vector generated by the gravity of the connecting rod of the mechanical arm, Is a mechanical arm jacobian matrix in the current configuration, A generalized force vector that is an end load; the heavy moment cost term is defined as the normalized form of the sum of absolute values of all joint moments: Wherein, the Is a vector Is used to determine the (i) th component of the (c), Is the maximum rated torque allowed by the ith joint.
8. The mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization according to claim 1, wherein in the step S5, the automatic switching control mode comprises three stages: In the free motion stage, high-rigidity position control is adopted to control the mechanical arm to quickly track the motion trail; A contact buffering stage, when the distance d between the end effector and the nearest point of the target object surface is smaller than the preset safe distance When the control is switched to low-rigidity impedance control; A steady state clamping stage when the component of the contact force detected by the end force sensor in the normal direction Reaching a preset contact force threshold And switching to the force-position mixing control to maintain stable grabbing.
9. The mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization according to claim 8, wherein the low-rigidity impedance control law of the contact buffering stage is as follows: wherein M is an inertial parameter matrix, B is a damping parameter matrix, K is a rigidity parameter matrix, x, 、 The actual position, velocity, acceleration of the end effector, In order to achieve the desired position, the position of the device, For the contact force detected by the end sensor; At this stage, the stiffness parameter matrix K has component values in the terminal approaching direction Is set to 5% to 10% of the value of the corresponding stiffness in the free movement phase.
10. The mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization according to claim 1, wherein in S5, the method further comprises the steps of slip detection and suppression: In a steady-state clamping stage, tangential force component signals output by the end force sensor are monitored in real time, and high-frequency filtering is carried out to extract micro-vibration signal characteristics; If the characteristic amplitude of the micro-vibration signal exceeds a preset threshold, judging that a slip precursor occurs, triggering a slip inhibition strategy to increase the target clamping force until the micro-vibration signal disappears, wherein the slip inhibition strategy is expressed as follows: Wherein, the For the target clamping force at the present moment, For a predetermined force gain factor, For the estimated slip rate until the micropunching signal is extinguished.

Description

Mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization Technical Field The invention relates to the field of robot motion planning, in particular to a mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization. Background In the robot operation of unstructured environment, such as warehouse sorting, logistics loading and unloading or flexible production line, the mechanical arm needs to be self-adaptive to objects with different positions, shapes and materials, and stable, safe and efficient grabbing is achieved. The existing data-driven grabbing method based on deep learning generally directly outputs grabbing pose through a training neural network, the method is seriously dependent on a large amount of marked data, when the special-shaped object, the reflecting surface or the complex geometric structure outside the training set is faced, the identification and the positioning are easy to lose effectiveness, more importantly, the generated grabbing suggestion is often a result lacking in physical interpretability, and the surface normal direction, the local curvature, the friction characteristic and other key geometric physical constraints at the grabbing point cannot be explicitly considered, so that the robustness of the method on the physical layer is insufficient, grabbing sliding or losing is easy to generate, and the generalized grabbing stability of an unknown object is difficult to ensure. In addition, in the aspect of mechanical arm movement track planning, the mainstream algorithm usually takes the shortest geometric path or the minimum movement time as an optimization target, ignores the self dynamics characteristics of the mechanical arm, particularly the influence of gravity load, and when a heavier object is grabbed, the algorithm can plan the track which enables the mechanical arm to be in a fully-extended state, so that the root joint bears huge gravity moment, the energy consumption and the heat of a driving motor are increased, overload protection can be triggered or the abrasion of mechanical parts can be accelerated, and the active optimization of the joint load in the movement process is lacked. Meanwhile, the traditional mechanical arm position control, such as high-gain PID control, usually keeps constant high rigidity in the grabbing contact stage, the control strategy is extremely sensitive to positioning errors, and according to Hooke's law, tiny position overshoot can generate huge instantaneous impact force under rigid contact, when fragile articles such as glassware, easily-deformed foods and the like are grabbed, the rigid impact is extremely easy to damage a target object, the capability of self-adaptively adjusting flexibility according to the contact state is lacking, and the double requirements of quick positioning and safe contact cannot be met. Therefore, how to design a mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization, which can give consideration to physical robustness, motion energy efficiency and contact safety, so as to cope with the challenges of complex and various grabbing tasks in unstructured environments is a problem to be solved by the technicians in the field. Disclosure of Invention In view of the above, the invention provides a mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization, which aims to solve the problems that the existing data driving grabbing method is insufficient in grabbing stability in an unstructured environment, and the existing path planning method ignores the dynamic characteristics of the mechanical arm to cause overlarge joint load, so that the stability, energy efficiency and safety of grabbing operation of the mechanical arm in the unstructured environment are improved. In order to achieve the above purpose, the present invention adopts the following technical scheme: a mechanical arm self-adaptive grabbing method based on geometric constraint and gravity optimization comprises the following steps: s1, acquiring three-dimensional point cloud data of a target object, and segmenting a smooth surface area through geometric feature analysis; S2, screening candidate grabbing point pairs meeting normal vector parallel constraint and friction cone constraint on the smooth surface area; s3, generating a grabbing strategy comprising a target clamping force and grabbing pose based on the candidate grabbing point pairs; S4, taking the grabbing pose as a target, and generating a motion track which is from the initial pose of the mechanical arm to the grabbing pose and has optimized joint load through a path planning algorithm introducing a gravity moment cost function; And S5, controlling the mechanical arm to move along the movement track, and automatically switching a control mode according to the distance and the tail end force feedback until the grabbing is comp