P. Boscariol, A. Gasparetto, R. Vidoni
Robust trajectory planning for flexible robots
Proc. of the 2013 ECCOMAS Multibody Dynamics Conference, Zagreb (Croatia), 1-4 July, 2013, pp. 293-294

Abstract: High speed operation is a recurring target in design and application of robotic manipulators. Moreover, maximizing the ratio between the weight of the payload and of the whole mechanism is a common objective. Therefore the use of a lightweight robot is a good practice that can help to reach these goals. On the other hand, the use of lightweight and therefore flexible manipulators requires the use of efficient control techniques and clever trajectory planning strategies [1]. For this reason, a large number of techniques have been proposed in literature to solve the problem of trajectory planning and control of such mechanisms [2]. Limiting our investigation only to the development of trajectory planning alogrithms, both model-based [3] and model-free techniques have been proposed [4]. In this paper we deal with the model-based trajectory planning of flexible-link robots. A large number of techniques have been developed for rigid-link robots, while the same problem is less frequently investigated for the case of manipulators with flexible-links, i.e. when the flexibility is distributed along the links of the robot. At the same time, the aim of this paper is to use the technique of desensitization to increase the robustness of the planned trajectory with respect to parametric perturbation of the plant. In fact several authors have emphasized [5] that the optimal control techniques which are commonly used in the case of model-based approaches, lead to a lack of robustness. This means that a trajectory that is optimal in the nominal case, is far from the optimal solution if applied to a perturbed plant. This can happen quite frequently, given the general difficulty of using (and tuning) accurate dynamic models of flexible-link mechanisms. In this paper, a possible solution to this problem is proposed. The approach is based on the definition of a Two-Point Boundary Value Problem (TP-BPV), which is augmented with the introduction of one or more sensitivity functions. A similar approach has been used in some works by Singh [6], but its application is limited to linear plants. Here, a nonlinear model of a single flexible-link mechanism, which has been validated in [7], is taken into account.

Area: Robotica

Home | Informazioni | Ricerca | Didattica | Personale | Tesi | Mappa | English