Abstract
Robotic manipulators have been widely used in industries, mainly to move tools into different specific positions. Thus, it has become necessary to have accurate knowledge about the tool position using forward kinematics after accessing the angular locations of limbs. This paper presents a simulation study in which an encoder attached to the limbs gathers information about the angular positions. The measured angles are applied to the Kalman Filter (KF) and its variants for state estimation. This work focuses on the use of fractional order controllers with a Two Degree of Freedom Serial Flexible Links (2DSFL) and Two Degree of Freedom Serial Flexible Joint (2DSFJ) and undertakes simulations with noise and a square wave as input. The fractional order controllers fit better with the system properties than integer order controllers. The KF and its variants use an unknown and assumed process and measurement noise matrices to predict the actual data. An optimisation problem is proposed to achieve reasonable estimations with the updated covariance matrices.
Highlights
The performance may vary for different control techniques in the case of integer order PID (IOPID) and fractional order PID (FOPID) controllers (Figures 11 and 12), but qualitatively, the particle filter (PF) has a better response. This is because the PF does not generalize all of the data; rather, it operates on the areas of data that are known as particles, which are both time-consuming and computationally intensive processes
The performance and merits–demerits of the Kalman Filter (KF), Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF) and PF along with the implementation of a fractional order controller have been presented in this work
This research has focused on the improved performance of SFL and SFJ with respect to the conventional PID controller
Summary
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Linear state estimations are performed through the implementation of Kalman Filters, but this may not be a good option in the case of dynamic models, measurements that are highly nonlinear and non-Gaussian noise Filters such as the particle filter (PF) [5], Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF) [6,7] give a generalised solution for such systems, where low performance is caused due to intractable linearisation and Gaussian approximations. By using fractional and integer order sliding mode control [29], a comparison study has been conducted with a planer robotic manipulator and has shown a significant improvement in stability. Utilization of joint angle measurements in the filters for state estimations; Implementation of a fractional order PID controller; An extensive comparison of filters. Similar approaches and functions have been applied for both 2DSFJ and
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
