Second-order sliding mode control for robot arms with time base generators for finite-time tracking (Q5959848)

This paper presents a new, dynamic sliding mode control algorithm for robotic manipulators that provides a finite-time convergence of joint tracking errors. The manipulator dynamics are assumed in classical form \(+\) a friction term, i.e. \[ Q(q)\ddot q+ C(q,\dot q)\dot q+ G(q)+ B_0\dot q= u.\tag{\(*\)} \] The following control algorithm has been proposed \[ u= -K_d\text{sat}(S_r)- Y_c\overline\Theta\text{ sat}(Y^T_c S_r)+ Y_c\Theta_0,\tag{\(\#\)} \] where sat denotes a saturation function, \(S_r= \dot q-\dot q_r\), \(\dot q_r= \dot q_d-\alpha(t)\Delta q+ S_d- K_i\sigma\), \(\dot\sigma= \text{sgn }S_q\), \(S_q= S-S_d\), \(S=\Delta\dot q+ \alpha(t)\Delta q\), \(S_d= S_0\exp(-kt)\), \(\Delta q= q- q_d\), \(q_d(t)\) -- a desirable joint trajectory, \(\alpha(t)\) -- a time base generator responsible for the finite-time error convergence, \(Y_c\) -- the regressor of \((*)\) computed for the continuous part of \(\ddot q_r\), \(\Theta\) -- the vector of unknown parameters of \((*)\), \(\Theta_0\) -- an initial estimate of \(\Theta\), and \(\overline\Theta\) -- a diagonal matrix with upper estimates of parameter errors on its diagonal. The main result of the paper states that for suitably chosen gains \(K_d\), \(K_i\) the tracking error \(\Delta q\) in the closed loop system \((*)\) and \((\#)\) is finite-time convergent. Computer simulations illustrate the theory.