Vijyant Agarwal

Intelligent control of four DOF robotic arm

In this paper; Fuzzy control and Adaptive Network based Fuzzy Interference System (ANFIS) techniques have been investigated for decentralised joint position control of an four DOF robotic arm for desired motion. Four different types of membership functions like Triangular; Pi; Gaussian; Dsigmf are implemented in the fuzzy PD and ANFIS controller. The performance of controller is evaluated and compared in terms of maximum overshoot; settling time and steady state error with conventional PID controller. Simulation results and comparative analysis validates the effectiveness of Fuzzy and ANFIS based controller with the proper selection of membership function for improving the system performance and reducing position tracking error.

Nonlinear robot teleoperation with random fluctuations in the feedback controller

A master and slave robot act in a teleoperation system. The master and slave angular position and velocities are sampled at integer multiples of a time step and the resulting difference errors at different sampling times are fed back both at the master and salve end after introducing appropriate feedback weights for each error sample. During the process of reading the master variables at the slave end, a teleoperation delay is introduced and like wise at the master end. In the presence of tremor and environmental noise at the master and slave ends and in addition random fluctuations in the feedback weights, we derive linearized delay stochastic differential equations for the state vector perturbation and calculate the correlations of the same in terms of the master and slave noise variances and the correlations of the master-slave error feedback weight vector. This calculation is based on approximate perturbation theory. Master-slave tracking is verified via MATLAB simulations.

COMPARATIVE ANALYSIS FOR KINEMATICS OF 5-DOF INDUSTRIAL ROBOTIC MANIPULATOR

Abstract This paper gives the kinematic analysis of a 5-DOF industrial robotic manipulator while considering wrist in motion. Analytical solutions have been obtained for forward kinematics and inverse kinematics to accurately position the end-effector of robotic manipulator in three dimensional spaces. For the first time, a hybrid neuro-fuzzy intelligent technique with two different membership functions has been studied and their performances are comparatively evaluated with analytical solutions. An experiment has been performed for a desired trajectory. It is seen that the results for the intelligent technique are reasonably in agreement with experiment. Also, the results obtained highlight the importance of selection of a particular membership function for robotic manipulators of industrial use.

Novel fuzzy clustering algorithm for fuzzy data

This paper presents a new fuzzy clustering algorithm for fuzzy numbers, called the weight fuzzy c-means (WFCM) clustering based on distance function [1]. We first discuss the conventional FCM algorithm for crisp data with brief overview of fuzzy set theory related to the problem at hand and indicate the disparity in the existing approaches of clustering for fuzzy data. In the proposed method, first we converted the fuzzy data matrix into respective weight matrix and then using FCM calculates the membership grade of every fuzzy data. Numerical results show that the presented algorithm performs more robust, logical and superior in performance.

Bilateral teleoperation control using recursive least squares filter with forgetting factor

In this research, we have designed and implemented recursive least squares (RLS) algorithm in master slave tracking on Geomagic® Touch™ (Phantom Omni) haptic device. RLS algorithm enables us to achieve optimal tracking in a teleoperation system in which the system parameters vary with time and the noise is weakly non-stationary. In our previous work on teleoperation, we employed Widrows least mean square algorithm instead of RLS algorithm and achieved satisfactorily high tracking accuracy. There, we employed instantaneous errors to update filter coefficients and hence slave positions. This study initiated with the idea that if we account all or some of the previous errors in updating filter coefficients and thus reducing current error, we might be probably able to achieve even higher tracking accuracy than that achieved with WLMS. Therefore, in order to understand this influence of older errors on tracking accuracy, we have applied RLS algorithm with forgetting factor. The use of forgetting factor in the least squares algorithm enables us to base our tracking on different weights of past errors that further helps us in understanding this influence at a broader level.

A Survey Paper on: Fuzzy Mathematical Morphology Techniques for Digital Image Processing

This paper puts across the various approaches and methods that have been proposed in the context of Fuzzy Mathematical Morphology. The underlying principles of Dilation & Erosion, the structuring elements used in various techniques, the unique variations put forth by researchers, new applications in spatial relationships, decision making, segmentation of medical images have been discussed.

Tracking Control of 4-DOF Robotic Arm Using Krill Herd-Optimized Fuzzy Logic Controller

Robots have become an integral part of the automated world. The motion control in joint space is a challenging problem with the objective of the controller is to force the joints of the manipulator to track the desired trajectory. In this paper, Krill Herd-Optimized Fuzzy Logic Controller (KHO-FLC) is designed, tested, and implemented for the first time for the tracking control of a 4-degree of freedom (DOF) robotic arm. The algorithm of Krill Herd (KH) is used to tune the parameters of Fuzzy Logic Controller (FLC). The proposed work is validated by simulation study of a robotic manipulator for the tracking of joint motion control. A comparison of performance analysis of the system using KHO-FLC and that of Fuzzy-PD, and Particle Swarm Optimized Fuzzy Logic Controller (PSO-FLC) controller demonstrates the effectiveness of the proposed algorithm.

Inverse kinematics of 3-DOF robotic manipulator using analytical method, ANFIS method and experiments

Inverse kinematics of industrial robotic manipulators is a very complex task. This paper involves the kinematic analysis of robotic manipulator with industrial importance. Forward kinematic analysis has been performed by analytical method using Denavit-Hartenberg convention. Inverse kinematics solutions have been obtained by geometrical method. A hybrid combination of neural networks and fuzzy logic intelligent techniques using two different membership functions has also been used to perform inverse kinematic analysis. Experimental validation has been attempted on robotic manipulator to trace different desired trajectories. Comparative analysis of joint angle errors using two different membership functions shows the importance of selection of a particular membership function. A comparison drawn on all the applied techniques, namely analytical method, adaptive neuro-fuzzy inference system method with experiments for desired trajectories shows the results for inverse kinematics to be in reasonable agreement with each other.

Wavelet transformation based tremor removal

One of the main problems in minimum invasive robotic surgery is effective control of the tracking error due to hand tremor. In this paper we review issues of hand tremor using surgical robot and propose a novel technique using wavelet transform with minimum coefficients to get a tremor free signal. In the parameter learning phase the error is minimized with minimum wavelet compression and as time progress the coefficients get stored to calculate the unknown parameters. Simultaneously the unused coefficients will be discarded to remove the data redundancy for faster processing and real time tracking.

Optimal trajectory tracking based on perturbation theory of linear systems with feedback error

In this work we propose block processing based optimal stochastic control method for tracking a desired response with a closed loop system. We assume that the feedback system contains random errors which manifest as small random perturbations to the feedback transfer function. The approximate mean square trajectory tracking error is then computed by applying perturbation theory. This mean square error is minimized with respect to the operator input force and the non-random components of the feedback transfer function.