“…To eliminate the effects of these uncertainties, robust, adaptive and machine learning techniques have been used in the model-based control designs [8]- [10]. Specifically, many forms of SMC approaches have been designed for robot manipulators including traditional SMC [11], integral SMC [12], fuzzy SMC [13], backstepping SMC [7], [14], terminal SMC [15], [16], second-order SMCs [17], nonsingular terminal SMC [18], fast terminal SMC [19], time-delay estimation based SMC [20] and adaptive SMC [21], [22], but these approaches require an accurate robot model to assure a good control response.…”