Simulation Experiments with Fuzzy Logic-Based Robot Control
The paper describes trajectory tracking simulation experiments with a hybrid approach to robot control that combines traditional model-based and fuzzy logic-based control techniques. The combined method is developed by extending a model-based decentralize
- PDF / 873,525 Bytes
- 6 Pages / 481.89 x 691.654 pts Page_size
- 109 Downloads / 213 Views
M. Vukobratovic and B. Karan M. Pupin Institute, Belgrade, Yugoslavia
Abstract The paper describes trajectory tracking simulation experiments with a hybrid approach to r~bot control that combines traditional model-based and fuzzy logic-based control techniques. The combined method is developed by extending a model-based decentralized control scheme with fuzzy logic-based tuners for modifying parameters of joint servo controllers. The simulation experiments conducted on a real-scale six-degree-of-freedom industrial robot demonstrate suitability of fuzzy logic-based methods for improving the performance of the robot control system.
1
Introduction
Central issues in model-based robot control schemes are complexity and uncertainty of the internal robot model [1]. It may become a very complex system of nonlinear differential equations and it is always more or less an approximation of the real robot. A viable possibility for attacking the problems of robot model complexity and uncertainty is offered by fuzzy logic-based controllers (FLC) and it has been explored by several researchers. The initial works that attempted to control manipulation robots directly with FLC [2. 3] have shown the applicability of the method, but they have also exposed considerable problems. The first problem is manifested by the lack of appropriate analytic tools for control design, i.e. selection of FLC parameters. A more tight connection between FLC and standard control methods was proposed by Tzafestas and Papanikolopoulos [-1]. who suggested to employ a two-level hierarchy in which FLC-based expert system is used for fine tuning of low-level PID control. The similar approach was applied to robot control by Popovic and Shekhawat [5]. However, the two-level hierarchy does not actually soh'e the second problem, that is manifested in weak performance. Ordinary FLC schemes displayed performance characteristics that are similar or just slightly better than with simple constant-gain PID schemes. This fact shows that knowledge of the readily available mathematical model of robot dynamics should not be ignored. Most importantly, it may be employed to decrease the nonlinear dynamic coupling between its joint subsystems. Thus. a combined approach is preferred, and it may yield superior control schemes than both simple model-based or fuzzy logic-based approaches. The similar concept was formulated by de Silva and MacFarlane [6], who tested the concept by simulation of a two-link with an assumption of idealized effectiveness of the low-level global nonlinear feedback.
A. Morecki et al. (eds.), Theory and Practice of Robots and Manipulators © Springer-Verlag Wien 1995
M. Vukobratovic and B. Karan
188
FLC-based tuner
IlKp, IlKD , ilK/ 'lIlKG
Il"t
Global gain KG
I
I
llu G
PID controller flu L Kp, KD, K[
flq, flq
U
o
U
Figure 1: Hybrid control scheme In this work, the performance of the hybrid approach is analyzed on a real-scale robot operating in free space. The purpose of this research is twofold. First, the expected improvements in rob
Data Loading...
