Optimal sliding mode control of a robot manipulator under uncertainty using PSO

This paper presents a novel optimal control approach for robot manipulators using particle swarm optimization (PSO). Although the PSO algorithm can find an optimal solution for complex problems, there are some difficulties to apply the particle swarm optimization in real-time control. This paper deals with uncertainties in using the PSO for finding an optimal sliding mode control. Being a repetitive algorithm with an offline nature, the PSO is not as fast as sufficient to apply on a dynamical system. Moreover, the actual system is not repeatable in the presence of external disturbances. As a solution, this paper suggests that the PSO should be applied to the model of a system instead of an actual system. Then some modifications are given since the model differs from the actual system. According to a range of uncertainty, a few nominal models for the system are selected. Next, the optimal designs are obtained by the PSO in offline. In real time, the system is estimated to select a model and design the controller. Finally, the control design is applied to the system. Simulation results show the efficiency of the proposed control and its superiority over the sliding mode control.