Adaptive Projection Neural Network for Kinematic Control of Redundant Manipulators With Unknown Physical Parameters

Zhang, Yinyan; Chen, Siyuan; Li, Shuai; Zhang, Zhijun

doi:10.1109/tie.2017.2774720

Cited by 108 publications

(30 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More detailed research by utilizing the RNNs with application to robot autonomy can also be refereed in the state-of-the-art works. [148][149][150]…”

Section: Rnn In Robot Autonomymentioning

confidence: 99%

A Review on Neural Dynamics for Robot Autonomy

Chen

2018

IJRC

Self Cite

View full text Add to dashboard Cite

Exploiting neural networks to solve control problems of robots is becoming commonly and effectively in academia and engineering. Due to the remarkable features like distributed storage, parallelism, easy implementation by hardware, adaptive self-learning capability, and free of off-line training, the solutions of neural networks break the bottlenecks of serial-processing strategies and methods, and serve as significant alternatives for robotic engineers and researchers. Especially, various types and branches of recurrent neural networks (RNNs) have been sequentially developed since the seminal works by Hopfield and Tank. Successively, many classes and branches of RNNs such as primal-dual neural networks (PDNNs), zeroing neural networks (ZNNs) and gradient neural networks (GNNs) are proposed, investigated, developed and applied to the robot autonomy. The objective of this paper is to present a comprehensive review of the research on neural networks (especially RNNs) for control problems solving of different kinds of robots. Specifically, the state-of-the-art research of RNNs, PDNNs, ZNNs and GNNs in different robot control problems solving are detailedly revisited and reported. The readers can readily find many effective and valuable solutions on the basis of neural networks for the robot autonomy in this paper.

show abstract

“…More detailed research by utilizing the RNNs with application to robot autonomy can also be refereed in the state-of-the-art works. [148][149][150]…”

Section: Rnn In Robot Autonomymentioning

confidence: 99%

A Review on Neural Dynamics for Robot Autonomy

Chen

2018

IJRC

Self Cite

View full text Add to dashboard Cite

show abstract

“…Redundant manipulators are usually designed as a series of links connected by motor-driven joints which extend from a fixed base to an end-effector [2]. Until now, researches on redundant manipulators have made great progress, including [3]- [15], and have achieved the extension from single manipulator to a collection of redundant manipulators such as [8]- [15], just name a few.…”

Section: Introductionmentioning

confidence: 99%

Cooperative Kinematic Control for Multiple Redundant Manipulators Under Partially Known Information Using Recurrent Neural Network

et al. 2020

IEEE Access

Self Cite

View full text Add to dashboard Cite

In this study, we investigate the problem of cooperative kinematic control for multiple redundant manipulators under partially known information using recurrent neural network (RNN). The communication among manipulators is modeled as a graph topology network with the information exchange that only occurs at the neighbouring robot nodes. Under partially known information, four objectives are simultaneously achieved, i.e, global cooperation and synchronization among manipulators, joint physical limits compliance, neighbor-to-neighbor communication among robots, and optimality of cost function. We develop a velocity observer for each individual manipulator to help them to obtain the desired motion velocity information. Moreover, a negative feedback term is introduced with a higher tracking precision. Minimizing the joint velocity norm as cost function, the considered cooperative kinematic control is built as a quadratic programming (QP) optimization problem integrating with both joint angle and joint speed limitations, and is solved online by constructing a dynamic RNN. Moreover, global convergence of the developed velocity observer, RNN controller and cooperative tracking error are theoretically derived. Finally, under a fixed and variable communication topology, respectively, application in using a group of iiwa R800 redundant manipulators to transport a payload and comparison with the existing method are conducted. Among the simulative results, the robot group synchronously achieves the desired circle and rhodonea trajectory tracking, with higher tracking precision reaching to zero. When joint angles and joint velocities tend to exceed the setting constraints, respectively, they are constrained into the upper and lower bounds owing to the designed RNN controller. INDEX TERMS Velocity observer, multiple redundant manipulators, recurrent neural network, motion planning, zeroing neural network.

show abstract

“…Xiao et al [26] offered a novel method for avoiding obstacles using the gradient neural network method. Zhang et al [17], [23], [27] established the first adaptive RNN with online learning capabilities for redundant robots with unknown physical parameters. Chen et al [28] used an adaptive projection neural network for the kinematics control of redundant robots with unknown physical parameters.…”

Section: Introductionmentioning

confidence: 99%

Novel Method of Obstacle Avoidance Planning for Redundant Sliding Manipulators

Liu

Tong

et al. 2020

IEEE Access

View full text Add to dashboard Cite

In this paper, a weighted additional deviation velocity on the basis of the gradient projection (GP-WADV) method is proposed to solve obstacle avoidance problems of the redundant manipulator under the condition of known end-effector trajectory. The GP-WADV method aims to find the point closest to the obstacle on the link of the manipulator and attach the additional deviation velocity with weighting factors to the point on the basis of the gradient projection method. Utilizing multi-solution characteristics of redundant robots, obstacle avoidance control parameters related to the distance between obstacles and links are proposed according to the gradient projection method for solving inverse kinematic solution. Moreover, a weighted method is adopted to adjust the weighting factors of the primary motion velocity and the additional motion velocity when the link approaches the obstacle, so as to obtain the joint angle solutions more quickly and accurately. The GP-WADV method can ensure that the manipulator can avoid obstacles smoothly without turbulence and solve the highly nonlinear relationship between the position of the manipulator and the joint coordinate, which has the advantages of small amount of calculation, more flexible and convenient real-time control. The advantages of the proposed GP-WADV method are proved by simulation and experiments. INDEX TERMS Additional deviation velocity method, gradient projection method, obstacle avoidance, redundant manipulator, weighted method.

show abstract

Adaptive Projection Neural Network for Kinematic Control of Redundant Manipulators With Unknown Physical Parameters

Cited by 108 publications

References 49 publications

A Review on Neural Dynamics for Robot Autonomy

A Review on Neural Dynamics for Robot Autonomy

Cooperative Kinematic Control for Multiple Redundant Manipulators Under Partially Known Information Using Recurrent Neural Network

Novel Method of Obstacle Avoidance Planning for Redundant Sliding Manipulators

Contact Info

Product

Resources

About