Inversion-based sliding control of a flexible-link manipulator (Q4955074)

In this paper a new, robust tracking control algorithm has been derived for flexible link robotic manipulators. The algorithm is based on the input-output inversion and involves a sliding mode action. The dynamics of the manipulator and the output map are assumed in the form \[ \left \{\begin{matrix} M(q,\delta) {\ddot q\brack \ddot\delta} +{f_1(q,\dot q)+g_1(q,\dot q, \delta,\dot \delta)\brack f_2(q,\dot q)+g_2(q,\dot q,\delta, \dot\delta) +K \delta}= {u\brack 0}\\ y=q+ \Psi\delta, \end{matrix}\right. \tag{*} \] where \(q\in \mathbb{R}^n\), is a vector of joint positions, \(\delta\in \mathbb{R}^m\) denotes link deflections, \(K\) is the stiffness matrix, \(\Psi\) stands for a structure matrix depending on modal shape functions, and the remaining terms have the standard robotic meaning. Given the tracking errors \(e=y_r-y\), \(e_I= \int edt\), a sliding surface and a saturated distance function have been defined, respectively, as \[ \begin{cases} \sigma=K_p e_I+K_d e+\dot e\\ s_\sigma= \sigma-\Phi I_n sat\left( {a\over\Phi} \right), \end{cases} \] where \(\Phi\) is a measure of thickness of the boundary layer around the sliding surface. The system equations (*) are transformed to an affine control system with outputs and complemented with the time evolution equation of the distance \(s_\sigma\). In this framework a control algorithm has been designed in the form \[ u=u_N+u_U= u_n+u_\Delta +u_U, \] where \(u_N\) is responsible for tracking in the nominal system (*), \(n_U\) makes the sliding surface attractive, and \(u_\Delta\) ensures boundedness of the deflections. The main result of the paper establishes conditions under which the control law guarantees a practical stability of tracking errors \(e,\dot e\), deflections \(\delta,\dot \delta\), and distances \(s_\sigma\). The theory has been illustrated with computer simulations.