CN109732605B - Robot joint friction torque compensation method and system - Google Patents
Robot joint friction torque compensation method and system Download PDFInfo
- Publication number
- CN109732605B CN109732605B CN201910052603.9A CN201910052603A CN109732605B CN 109732605 B CN109732605 B CN 109732605B CN 201910052603 A CN201910052603 A CN 201910052603A CN 109732605 B CN109732605 B CN 109732605B
- Authority
- CN
- China
- Prior art keywords
- neural network
- joint
- angular velocity
- friction
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000013528 artificial neural network Methods 0.000 claims abstract description 181
- 230000008569 process Effects 0.000 claims abstract description 34
- 238000005070 sampling Methods 0.000 claims abstract description 18
- 238000011478 gradient descent method Methods 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 4
- 210000002569 neuron Anatomy 0.000 claims description 4
- 238000005461 lubrication Methods 0.000 abstract description 6
- 230000000694 effects Effects 0.000 abstract description 4
- 238000013459 approach Methods 0.000 abstract description 2
- 230000008859 change Effects 0.000 abstract description 2
- 238000005299 abrasion Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Landscapes
- Feedback Control In General (AREA)
Abstract
本发明公开一种机器人关节摩擦力矩的补偿方法及系统。采用本发明的补偿方法及系统，工业机器人运行过程中，当关节温度、负载、润滑和磨损等参数改变后，摩擦模型也会相应改变，能够避免基于固定摩擦模型的前馈补偿方法失效的问题。在此基础上，利用辨识的摩擦模型预测生成的摩擦力矩训练神经网络，使得神经网络的训练学习过程效率更高，能够更快逼近理想效果，提高了机器人关节的跟踪速度和跟踪精度。进一步地，本发明通过高斯采样进行摩擦力矩前馈补偿，弥补了神经网络输出不具有探索性的缺点，使神经网络能更好的收敛，避免陷入局部最优，进一步确保跟踪精度，即使关节温度、负载、润滑和磨损等参数发生变化，也能快速实现高精度跟踪的目的。
The invention discloses a method and system for compensating friction torque of a robot joint. By adopting the compensation method and system of the present invention, when the parameters such as joint temperature, load, lubrication and wear are changed during the operation of the industrial robot, the friction model will also be changed accordingly, which can avoid the failure of the feedforward compensation method based on the fixed friction model. . On this basis, the neural network is trained by using the friction torque generated by the identified friction model prediction, which makes the training and learning process of the neural network more efficient, can approach the ideal effect faster, and improves the tracking speed and tracking accuracy of the robot joints. Further, the present invention uses Gaussian sampling to perform friction torque feedforward compensation, which makes up for the disadvantage that the output of the neural network is not exploratory, enables the neural network to converge better, avoids falling into local optimum, and further ensures the tracking accuracy. , load, lubrication and wear and other parameters change, but also can quickly achieve the purpose of high-precision tracking.
Description
Technical Field
The invention relates to the field of robots, in particular to a method and a system for compensating a friction torque of a robot joint.
Background
The friction of the joints of the industrial robot is one of the main factors causing the reduction of the tracking precision of the joints of the robot, and the friction torque must be compensated in order to reduce the influence of the friction on a robot control system. Usually, a feedforward compensation method based on a fixed friction model is adopted for friction torque compensation, but in the operation process of an industrial robot, parameters such as joint temperature, load, lubrication, abrasion and the like are constantly changed, so that the friction model of the feedforward compensation is remarkably changed, the feedforward compensation method based on the fixed friction model is invalid, and the tracking precision of the robot joint is reduced.
Disclosure of Invention
The invention aims to provide a method and a system for compensating a friction torque of a robot joint, which can realize the aim of tracking the robot joint quickly and accurately.
In order to achieve the purpose, the invention provides the following scheme:
a method of compensating for robot joint friction torque, the method comprising:
acquiring first joint angular velocity data and first friction torque data corresponding to the first joint angular velocity data in a robot starting process, wherein the starting process is a process that the robot is started to run to a rated rotating speed at a set speed increment;
performing friction model identification by using the first joint angular velocity data and the first friction torque data to obtain a friction model, wherein the input of the friction model is joint angular velocity, and the output of the friction model is joint friction torque;
establishing a neural network, wherein the input of the neural network is joint angular velocity, and the output of the neural network is joint friction torque;
acquiring second joint angular velocity data of the robot in a starting process;
inputting the second joint angular velocity data into the friction model to obtain first moment prediction data corresponding to the second joint angular velocity data;
taking the second joint angular velocity data and the first moment prediction data as training data of the neural network, and updating the weight and the bias of the neural network by using a gradient descent method to obtain a neural network prediction model;
acquiring random angular velocity data of the robot, wherein the random angular velocity data is angular velocity data when the robot runs at a random speed, and the random speed is less than or equal to twice of a rated rotating speed;
inputting the random angular velocity data into the neural network prediction model to obtain second moment prediction data;
carrying out Gaussian sampling by taking the second moment prediction data as a mean value and 1 as a variance to obtain a friction moment compensation value;
and performing friction torque compensation on the robot joint according to the friction torque compensation value.
Optionally, after performing friction torque compensation on the robot joint according to the friction torque compensation value, the method further includes:
acquiring joint tracking deviation;
judging whether the joint tracking deviation is greater than a deviation threshold value;
and if so, updating the weight and the bias of the neural network according to the joint tracking deviation to obtain an updated neural network prediction model.
Optionally, the updating the weight and the bias of the neural network according to the joint tracking deviation specifically includes:
according to the formula:
updating weights and biases of the neural network, wherein θnewFor the updated neural network parameter set, the neural network parameter set comprises weight and bias of the neural network, thetaoldFor the neural network parameter set before updating, α is the learning rate of the neural network, NetθIn order to be a neural network, the network is,
is the neural network parameter gradient, and e is the joint tracking deviation.
Optionally, the identifying a friction model by using the first joint angular velocity data and the first friction torque data to obtain a friction model specifically includes:
performing friction model identification by using the first joint angular velocity data and the first friction torque data to obtain a Sterbek model;
determining each parameter to be identified in the Sterbek model by adopting a nonlinear least square method to obtain each friction parameter;
and determining a friction model of the robot joint according to each friction parameter and the Sterbek model.
Optionally, the neural network includes 1 hidden layer and 10 neurons, the connection mode of the neural network is full connection, and the activation function of the neural network is a linear rectification function.
Optionally, the expression of the neural network is:
Optionally, the updating the weight and the bias of the neural network by using a gradient descent method specifically includes:
according to the formula:
updating weights and biases of the neural network, wherein θnewFor the updated neural network parameter set, the neural network parameter set comprises weight and bias of the neural network, thetaoldFor the neural network parameter set before updating, α is the learning rate of the neural network, NetθIn order to be a neural network, the network is,
is the neural network parameter gradient.
A compensation system for robot joint friction torque, the compensation system comprising:
the robot starting device comprises a first data acquisition module, a second data acquisition module and a control module, wherein the first data acquisition module is used for acquiring first joint angular velocity data and first friction torque data corresponding to the first joint angular velocity data in a robot starting process, and the starting process is a process that the robot is started at a set speed increment and runs to a rated rotating speed;
the model identification module is used for identifying a friction model by utilizing the first joint angular velocity data and the first friction torque data to obtain a friction model, wherein the input of the friction model is the joint angular velocity, and the output of the friction model is the joint friction torque;
the neural network establishing module is used for establishing a neural network, the input of the neural network is joint angular velocity, and the output of the neural network is joint friction torque;
the second data acquisition module is used for acquiring second joint angular velocity data in the starting process of the robot;
the first moment prediction module is used for inputting the second joint angular velocity data into the friction model to obtain first moment prediction data corresponding to the second joint angular velocity data;
the neural network determining module is used for taking the second joint angular velocity data and the first moment prediction data as training data of the neural network, and updating the weight and the bias of the neural network by using a gradient descent method to obtain a neural network prediction model;
the speed acquisition module is used for acquiring random angular speed data of the robot, wherein the random angular speed data is angular speed data when the robot runs at a random speed, and the random speed is less than or equal to twice of a rated rotating speed;
the second moment prediction module is used for inputting the random angular velocity data into the neural network prediction model to obtain second moment prediction data;
the Gaussian sampling module is used for carrying out Gaussian sampling by taking the second moment prediction data as a mean value and 1 as a variance to obtain a friction moment compensation value;
and the torque compensation module is used for performing friction torque compensation on the robot joint according to the friction torque compensation value.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention provides a compensation method and a compensation system for robot joint friction torque. Inputting the second joint angular velocity data into the friction model to obtain first moment prediction data corresponding to the second joint angular velocity data; and training a neural network by using the second joint angular velocity data and the first moment prediction data, and updating the weight and the bias of the neural network by combining a gradient descent method to obtain a neural network prediction model. And inputting the random angular velocity data into a neural network prediction model to obtain second moment prediction data. And performing Gaussian sampling by taking the second moment prediction data as a mean value and 1 as a variance to obtain a friction moment compensation value. And performing friction torque compensation on the robot joint according to the friction torque compensation value. Therefore, the friction model is firstly identified to obtain the friction model. In the operation process of the industrial robot, when parameters such as joint temperature, load, lubrication, abrasion and the like are changed, the friction model is also correspondingly changed, and the problem that a feedforward compensation method based on a fixed friction model is invalid can be solved. On the basis, the friction moment generated by the identified friction model prediction is used for training the neural network, so that the training and learning process efficiency of the neural network is higher, the ideal effect can be approached more quickly, and the tracking speed and the tracking precision of the robot joint are improved. Furthermore, the friction torque feedforward compensation is carried out through Gaussian sampling, the defect that the output of the neural network does not have searchability is overcome, the neural network can be converged better, the situation that the neural network is trapped into local optimization is avoided, the tracking precision is further ensured, and the purpose of high-precision tracking can be rapidly realized even if parameters such as joint temperature, load, lubrication, abrasion and the like are changed.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a flowchart of a method for compensating a friction torque of a robot joint according to an embodiment of the present invention;
fig. 2 is a block diagram of a compensation system for a robot joint friction torque according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a method and a system for compensating a friction torque of a robot joint, which can realize the aim of tracking the robot joint quickly and accurately.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a flowchart of a method for compensating a friction torque of a robot joint according to an embodiment of the present invention. As shown in fig. 1, a method for compensating a friction torque of a robot joint includes:
step 101: acquiring first joint angular velocity data and first friction torque data corresponding to the first joint angular velocity data in a robot starting process, wherein the starting process is a process that the robot is started to run to a rated rotating speed at a set speed increment;
step 102: and identifying a friction model by using the first joint angular velocity data and the first friction torque data to obtain a friction model, wherein the input of the friction model is the joint angular velocity, and the output of the friction model is the joint friction torque.
Wherein, step 102: carrying out friction model identification by utilizing the first joint angular velocity data and the first friction torque data to obtain a friction model, which specifically comprises the following steps:
performing friction model identification by using the first joint angular velocity data and the first friction torque data to obtain a Sterbek model;
determining each parameter to be identified in the Sterbek model by adopting a nonlinear least square method to obtain each friction parameter;
and determining a friction model of the robot joint according to each friction parameter and the Sterbek model.
Step 103: establishing a neural network, wherein the input of the neural network is joint angular velocity, and the output of the neural network is joint friction torque. The neural network comprises 1 hidden layer and 10 neurons, the connection mode of the neural network is full connection, and the activation function of the neural network is a linear rectification function. The expression of the neural network is as follows:
Step 104: acquiring second joint angular velocity data of the robot in a starting process;
step 105: inputting the second joint angular velocity data into the friction model to obtain first moment prediction data corresponding to the second joint angular velocity data;
step 106: and taking the second joint angular velocity data and the first moment prediction data as training data of the neural network, and updating the weight and the bias of the neural network by using a gradient descent method to obtain a neural network prediction model.
The updating of the weight and the bias of the neural network by using the gradient descent method specifically comprises the following steps:
according to the formula:
updating weights and biases of the neural network, wherein θnewFor the updated neural network parameter set, the neural network parameter set comprises weight and bias of the neural network, thetaoldFor the neural network parameter set before updating, α is the learning rate of the neural network, NetθIn order to be a neural network, the network is,
is the neural network parameter gradient.
Step 107: acquiring random angular velocity data of the robot, wherein the random angular velocity data is angular velocity data when the robot runs at a random speed, and the random speed is less than or equal to twice of a rated rotating speed;
step 108: inputting the random angular velocity data into the neural network prediction model to obtain second moment prediction data;
step 109: carrying out Gaussian sampling by taking the second moment prediction data as a mean value and 1 as a variance to obtain a friction moment compensation value;
step 110: and performing friction torque compensation on the robot joint according to the friction torque compensation value.
Preferably, step 110 is performed: after carrying out friction torque compensation on the robot joint according to the friction torque compensation value, the method further comprises the following steps:
step 111: acquiring joint tracking deviation;
step 112: judging whether the joint tracking deviation is greater than a deviation threshold value;
if yes, go to step 113;
step 113: and updating the weight and the bias of the neural network according to the joint tracking deviation to obtain an updated neural network prediction model.
In this embodiment, step 113: updating the weight and the bias of the neural network according to the joint tracking deviation, which specifically comprises the following steps:
according to the formula:
updating weights and biases of the neural network, wherein θnewFor the updated neural network parameter set, the neural network parameter set comprises weight and bias of the neural network, thetaoldFor the neural network parameter set before updating, α is the learning rate of the neural network, NetθIn order to be a neural network, the network is,
is the neural network parameter gradient, and e is the joint tracking deviation.
Fig. 2 is a block diagram of a compensation system for a robot joint friction torque according to an embodiment of the present invention. As shown in fig. 2, a compensation system for a robot joint friction torque includes:
the first data acquisition module 201 is configured to acquire first joint angular velocity data of a robot in a starting process and first friction torque data corresponding to the first joint angular velocity data, where the starting process is a process in which the robot is started at a set speed increment and runs to a rated rotation speed;
the model identification module 202 is configured to perform friction model identification by using the first joint angular velocity data and the first friction torque data to obtain a friction model, where an input of the friction model is a joint angular velocity, and an output of the friction model is a joint friction torque;
the neural network establishing module 203 is used for establishing a neural network, the input of the neural network is joint angular velocity, and the output of the neural network is joint friction torque;
the second data acquisition module 204 is used for acquiring second joint angular velocity data of the robot in the starting process;
the first moment prediction module 205 is configured to input the second joint angular velocity data into the friction model, and obtain first moment prediction data corresponding to the second joint angular velocity data;
the neural network determining module 206 is configured to use the second joint angular velocity data and the first moment prediction data as training data of the neural network, and update a weight and a bias of the neural network by using a gradient descent method to obtain a neural network prediction model;
a speed obtaining module 207, configured to obtain random angular velocity data of the robot, where the random angular velocity data is angular velocity data of the robot when the robot runs at a random speed, and the random speed is less than or equal to twice a rated rotation speed;
a second moment prediction module 208, configured to input the random angular velocity data into the neural network prediction model to obtain second moment prediction data;
a gaussian sampling module 209, configured to perform gaussian sampling with the second moment prediction data as a mean and 1 as a variance to obtain a friction moment compensation value;
and the torque compensation module 210 is configured to perform friction torque compensation on the robot joint according to the friction torque compensation value.
The implementation process of the robot joint friction torque compensation system provided by the invention is as follows:
(1) the joint motion trajectory is designed to start from zero speed and start the joint motion at a speed increment of 30 revolutions per minute until the rated speed is reached.
(2) In the joint operation process, joint angular velocity data and joint torque data are collected every 2ms to obtain first joint angular velocity data and first friction torque data corresponding to the first joint angular velocity data.
(3) Performing model identification by using the first joint angular velocity data and the first friction torque data to obtain a friction model formula of a Stribeck curve (Stribeck) as follows:
wherein f (v) is friction torque, v is joint angular velocity, fcIs the Coulomb friction torque, fsIs maximum static friction moment, vsIs the Stribeck friction velocity, delta is the empirical coefficient, fvIs the coefficient of viscous friction.
Wherein the parameter to be identified is (f)c，fs，vs，δ，fv) In this embodiment, the identification model is established by a nonlinear least square method, and the formula is as follows:
in the formula (f)i(v) Calculated friction torque, y, for the Stribeck friction modeliThe method is characterized in that the method is used for acquiring the actual friction torque of the joint, namely first friction torque data, and n is the number of data samples, namely the number of the first friction torque data.
And then solving the parameter to be identified by adopting a gradient descent method. First, arbitrarily given a set of initial solutions (f)c，fs，vs，δ，fv) When the least square error is reduced to be within the threshold range of the least square error, the Stribeck friction model parameter (f) can be obtainedc、fs、vs、δ、fv) To obtain the Sribeck friction model. In this embodiment, the least square error threshold is less than or equal to 0.0001, and the step range of the update solution is: 0.001 to 0.1.
(4) A neural network is established, the network input is joint angular velocity, the network output is joint friction torque, the neural network has a hidden layer and comprises 10 neurons, the connection mode is full connection, and the activation function is a linear rectification function (ReLU). The neural network formula is as follows:
(5) 3000 data points are uniformly sampled between zero speed and rated rotating speed, and friction torque corresponding to each rotating speed is calculated through a Stribeck model obtained through identification. And then, taking the rotating speed and the friction torque as training data of the neural network, training the neural network by using a gradient descent method, and updating parameters. If the neural network parameter set is θ, the parameter update formula is:
(6) The joint running track is designed, the control period is 2 milliseconds, and the reciprocating motion is carried out at random speed within the range from zero speed to twice of rated rotating speed.
(7) And in the operation process, collecting the angular velocity in real time as the input of the neural network, and calculating to obtain the network output, namely the joint friction torque through the neural network established in the step 5. Then, taking the joint friction torque as a mean value and 1 as a variance, and performing Gaussian sampling, wherein the Gaussian sampling formula is as follows:
U1，U2＝r and(0，1)
wherein U1 and U2 are (0, 1)]Two consistent random numbers in the interval, Z is a normal value, m is a mean value, and delta2Is variance, x is Gaussian sampling result, friction torque compensation value.
After the joint friction torque feedforward compensation value is obtained through calculation of the formula, the compensation value is output to a joint feedforward channel, and joint tracking deviation is collected in the next control period.
(8) In the operation process of the step 7, the joint angular velocity, the joint friction torque feedforward compensation value and the joint tracking error are collected once every 2 milliseconds and are stored in a database as a group of training data.
(9) Sampling in a database, training a neural network, and updating the weight and the bias of the neural network according to the joint tracking deviation by adopting the following formula:
(10) And 7, repeating the step 7 to the step 9 to enable the friction compensation of the joints of the robot to reach an ideal state, namely the joint tracking deviation is less than or equal to the deviation threshold set by the user.
Therefore, the invention firstly identifies the Stribeck friction model, and utilizes the identified Stribeck friction model to generate the training data of the neural network to train the neural network, which is equivalent to adding expert knowledge, so that the training and learning process of the neural network is more efficient and approaches to the ideal effect more quickly.
The output of the neural network is a fixed value, and the friction torque feedforward compensation is carried out through Gaussian sampling, so that the defect that the output of the neural network does not have searchability is overcome, the neural network can better converge, and the phenomenon of falling into local optimum is avoided.
Meanwhile, the updating direction of the neural network parameters is corrected by the absolute value of the joint tracking deviation, so that the training can always be carried out towards the favorable direction.
Therefore, the friction torque compensation method based on the neural network and the Gaussian sampling can correct and adjust network parameters in real time in the operation process of the industrial robot, and can quickly adjust the network parameters to a new ideal compensation state even if parameters such as joint temperature, load, lubrication, abrasion and the like change, so that the robot joint is ensured to have higher tracking accuracy.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (7)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910052603.9A CN109732605B (en) | 2019-01-21 | 2019-01-21 | Robot joint friction torque compensation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910052603.9A CN109732605B (en) | 2019-01-21 | 2019-01-21 | Robot joint friction torque compensation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109732605A CN109732605A (en) | 2019-05-10 |
| CN109732605B true CN109732605B (en) | 2021-02-26 |
Family
ID=66365403
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910052603.9A Active CN109732605B (en) | 2019-01-21 | 2019-01-21 | Robot joint friction torque compensation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109732605B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110919633A (en) * | 2019-12-06 | 2020-03-27 | 泉州市微柏工业机器人研究院有限公司 | Robot position deviation method based on torque control |
| CN111428317B (en) * | 2020-04-06 | 2023-06-09 | 宁波智诚祥科技发展有限公司 | Joint friction torque compensation method based on 5G and cyclic neural network |
| CN113821006B (en) * | 2020-05-07 | 2022-10-14 | 牧今科技 | Method and computing system for determining the value of an error parameter indicative of robot calibration quality |
| US11691285B2 (en) | 2020-05-07 | 2023-07-04 | Mujin, Inc. | Method and computing system for estimating parameter for robot operation |
| CN111639749A (en) * | 2020-05-25 | 2020-09-08 | 上海智殷自动化科技有限公司 | Industrial robot friction force identification method based on deep learning |
| CN112152539B (en) * | 2020-09-29 | 2024-12-17 | 中国船舶集团有限公司第七二四研究所 | Method for realizing neural network compensation motor load moment observer |
| CN112677156B (en) * | 2020-12-30 | 2022-10-18 | 法奥(淄博)智能装备有限公司 | A friction compensation method for robot joints |
| CN113295311B (en) * | 2021-04-27 | 2022-04-29 | 北京交通大学 | Method for determining friction torque between rolling bearing roller and raceway and testing device |
| CN113855473B (en) * | 2021-08-25 | 2023-08-15 | 上海傅利叶智能科技有限公司 | Method and device for controlling exoskeleton robot and exoskeleton robot |
| CN113742865B (en) * | 2021-09-10 | 2024-10-29 | 遨博(北京)智能科技股份有限公司 | Data processing method |
| CN114028164B (en) * | 2021-11-18 | 2024-11-08 | 深圳华鹊景医疗科技有限公司 | A control method and device for a rehabilitation robot and a rehabilitation robot |
| CN114800498B (en) * | 2022-04-20 | 2022-12-09 | 上海捷勃特机器人有限公司 | SCARA robot moment feedforward compensation method |
| CN115042191B (en) * | 2022-08-12 | 2022-11-08 | 季华实验室 | Pre-training model fine-tuning training method and device, electronic equipment and storage medium |
| CN115366107A (en) * | 2022-09-08 | 2022-11-22 | 杭州景吾智能科技有限公司 | Method and system for compensating frictional force of robotic arm joints |
| CN119458362B (en) * | 2024-12-16 | 2025-04-18 | 工禾工业软件(上海)有限公司 | A method for improving the joint accuracy of industrial robots based on CAE dynamic simulation |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS61231611A (en) * | 1985-04-05 | 1986-10-15 | Nissan Motor Co Ltd | Controller of robot |
| JPH04343690A (en) * | 1991-05-22 | 1992-11-30 | Nippon Telegr & Teleph Corp <Ntt> | Joint friction compensation method for multi-joint manipulator |
| WO1997010081A1 (en) * | 1995-09-11 | 1997-03-20 | Kabushiki Kaisha Yaskawa Denki | Robot controller |
| CN101332604A (en) * | 2008-06-20 | 2008-12-31 | 哈尔滨工业大学 | Control method of human-computer interaction manipulator |
| CN105196294A (en) * | 2015-10-29 | 2015-12-30 | 长春工业大学 | Reconfigurable mechanical arm decentralized control system and control method adopting position measuring |
| CN106313044A (en) * | 2016-09-20 | 2017-01-11 | 华南理工大学 | Feedforward torque compensating method used for industrial robot |
| CN108714896A (en) * | 2018-05-28 | 2018-10-30 | 华南理工大学 | A kind of robot flexibility torque feed forward control method based on kinetics of deformable bodies model |
-
2019
- 2019-01-21 CN CN201910052603.9A patent/CN109732605B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS61231611A (en) * | 1985-04-05 | 1986-10-15 | Nissan Motor Co Ltd | Controller of robot |
| JPH04343690A (en) * | 1991-05-22 | 1992-11-30 | Nippon Telegr & Teleph Corp <Ntt> | Joint friction compensation method for multi-joint manipulator |
| WO1997010081A1 (en) * | 1995-09-11 | 1997-03-20 | Kabushiki Kaisha Yaskawa Denki | Robot controller |
| CN101332604A (en) * | 2008-06-20 | 2008-12-31 | 哈尔滨工业大学 | Control method of human-computer interaction manipulator |
| CN105196294A (en) * | 2015-10-29 | 2015-12-30 | 长春工业大学 | Reconfigurable mechanical arm decentralized control system and control method adopting position measuring |
| CN106313044A (en) * | 2016-09-20 | 2017-01-11 | 华南理工大学 | Feedforward torque compensating method used for industrial robot |
| CN108714896A (en) * | 2018-05-28 | 2018-10-30 | 华南理工大学 | A kind of robot flexibility torque feed forward control method based on kinetics of deformable bodies model |
Non-Patent Citations (1)
| Title |
|---|
| "机器人关节摩擦建模与补偿研究";吴晓敏等;《仪器仪表学报》;20181030;第39卷(第10期);第44-49页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109732605A (en) | 2019-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109732605B (en) | Robot joint friction torque compensation method and system | |
| CN109927032B (en) | Mechanical arm track tracking control method based on high-order sliding-mode observer | |
| CN106527152B (en) | Design method and system of closed-loop fractional PDɑ type iterative learning robot controller | |
| CN110531707A (en) | The friction model of SCARA robot improves and dynamic parameters identification method | |
| CN110815190B (en) | A method and system for dragging and teaching an industrial robot | |
| CN111872937B (en) | A Control Method for Uncertain Manipulators in Task Space | |
| CN113297798A (en) | Robot external contact force estimation method based on artificial neural network | |
| CN111941432A (en) | A high-performance robotic arm artificial intelligence output feedback control method | |
| Zheng et al. | On estimation of plant dynamics and disturbance from input-output data in real time | |
| CN114169230A (en) | Robot dynamics parameter identification method | |
| CN107577146B (en) | Neural network self-adaptive control method of servo system based on friction integral approximation | |
| CN115284276B (en) | Robot joint torque control method based on long short-term memory network | |
| CN111965976B (en) | Robot joint sliding mode control method and system based on neural network observer | |
| CN111428317A (en) | Joint friction torque compensation method based on 5G and recurrent neural network | |
| CN104460704A (en) | Tracking control method for pitching position of electric rotary table based on perturbation upper bound estimation | |
| CN113043275A (en) | Micro-part assembling method based on expert demonstration and reinforcement learning | |
| CN112904713B (en) | High-precision strong-robustness control method of spatial motion servo mechanism | |
| CN109062039B (en) | Adaptive robust control method of three-degree-of-freedom Delta parallel robot | |
| CN104965413B (en) | The friciton compensation self-adaptation control method of controlledization flat pad | |
| CN115366107A (en) | Method and system for compensating frictional force of robotic arm joints | |
| CN109048995B (en) | A nonlinear joint friction compensation method for a three-degree-of-freedom delta parallel robot | |
| Xie et al. | Modeling of friction torque based on turntable servo system using neural network with memory storage unit | |
| CN112462606B (en) | A Dynamic Parameter Identification Method for Flexible Joints Based on Adaptive Control | |
| Kamalasadan | A new intelligent controller for the precision tracking of permanent magnet stepper motor | |
| CN108267970B (en) | Time-lag rotor active balance control system and method based on Smith model and single neuron PID |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |