Knowledge Of Robot Dynamics Research Articles

Neurofuzzy self-tuning of the dissipation rate gain for model-free force-position exponential tracking of robots

Simultaneous force and position control of robots interacting with a rigid environment has been broadly studied assuming several contact force models, being the differential algebraic – DAE – model the most complete one, however DAE robots show complex and strong nonlinear couplings that make difficult to achieve tracking, when dynamic model is not available. In this paper, considering the fundamental structural properties of DAE robots, in particular passivity and the orthogonalization of force and velocity vectors, it is proposed a model-free neurofuzzy-based self-tuning robot controller for exponential tracking, which is composed of orthogonalized PID-position plus an I-force (If) control terms, and a feed-forward desired force term. The salient feature of this proposal is a novel neurofuzzy self-tuning scheme aimed at tuning the dissipation rate gain (DRG) so as to enforce dissipativity in closed-loop, rather than the standard scheme of tuning the feedback control gains, or the control structure, which in our case stands for a simple constant feedback gain orthogonalized PID+If controller. In fact, in virtue of such orthogonalization, it emerges a simple and low cost parallel structure of the neuro-fuzzy network that targets solely the DRG so as to drive error dynamics to zero with exponential rate, without any knowledge of robot dynamics or carrying out any approximation of inverse dynamics. Thus, this technique can be applied to other class of systems and controllers that ensure passivity in closed-loop. Simulations show the validity and the feasibility of this new approach.

Cartesian sliding PID control schemes for tracking robots with uncertain Jacobian

Owing to the fact that desired tasks are usually defined in operational coordinates, inverse and direct kinematics must be computed to obtain joint coordinates and Cartesian coordinates, respectively. However, in order to avoid the ill-posed nature of the inverse kinematics, Cartesian controllers have been proposed. Considering that Cartesian controllers are based on the assumption that the Jacobian is well known, an uncertain Jacobian will produce a non-exact localization of the end-effector. In this paper, we present an alternative approach to solve the problem of Cartesian tracking for free and constrained motion subject to Jacobian uncertainty. These Cartesian schemes are based on sliding PID controllers where the Cartesian errors are mapped into joint errors without any knowledge of robot dynamics. Sufficient conditions for feedback gains and stability properties of the estimate inverse Jacobian are presented to guarantee stability. Experimental results are provided to visualize the real-time stability properties of the Cartesian proposed schemes.

Compliant motion control for non-redundant rigid robotic manipulators

A novel robust trajectory/force controller is developed on the basis of a combination of sliding-mode and adaptive control techniques. Exact knowledge of robot dynamics and environment stiffness is not required. The controller is constructed so as to track reference trajectories in the unconstrained directions and to regulate force at a desired value in the constrained directions. Changes in constraints are also taken into account in the design of the control law. The synthesis presented assumes a non-redundant rigid manipulator and known location/geometry of the environment. Simulation and experimental results are also presented.

Neural Networks for Force Control of an Assembly Robot

The performing of robotized assembly tasks implies hard constraints like the knowledge of robot dynamics and robot-environment interaction. So classical form control laws lead to low performances. The aim of this paper is to show the ability of neural techniques for learning control under real world conditions of unknown nonlinear systems. In our control strategy neural networks are used for the identification and control of an assembly parallel robot performing a peg-in-hole insertion. Such a system is strongly non linear. In this paper, we propose a new approach of hybrid force position control based on feedforward neural networks. The proposed structure does not need any mathematical modelling of system dynamics. Experimental results show that identification and control schemes suggested in this paper are practically feasible. La réalisation de tâches d'assemblage robotisées implique des contraintes séverès telles que la connaissance de la dynamique du robot et de son interaction avec l'environnement, de sorte que les lois de contrôle classique conduisent à des performances insuffisantes. L'objectif de cet article est de montrer les capacites des techniques neuronales dans la commande des systèmes non linéaires dont les modèles ne sont pas connus. Notre stratégie de commande est basée sur l'utilisation des réseaux de neurones pour l'identification et la commande d'un robot parallèle d'assemblage qui effectue une tâche d'insertion d'une pige dans un alésage. Ce processus est fortement non linéaire. Cet article présente une nouvelle approche à base de réseaux de neurones du controle hybride force position. La structure qui est proposée ne nécessite pas de formulation mathématique de la dynamique du systéme. Les résultats expérimentaux montrent que la structure de commande et d'identification proposée conduit à des résultats prometteurs.

Adaptive control for a constrained robot without using a regressor

With the aim of controlling the motion/force of constrained robots with neither involving complex design procedurs or extensive on-line computations, this paper presents a new adaptive control strategy. The main advantage of the strategy, is its simplicity, in design for the programming code and real-time implementation. It is shown that the control objective is achieved without exact knowledge of robot dynamics or on-line calculation of non-linear dynamic functions. The control algorithm is constructed with at most four control parameters to be adjusted adaptively, and that number is determined independently of the number of links of the robot. To illustrate the developed method, a simulation example is presented.

Adaptive control scheme for a robot manipulator: direct decoupler and PID controller

A two level adaptive controller for the control of a high speed robot is proposed. Dynamic interaction effects of the robot system are eliminated by applying a centralized adaptive multivariable decoupler. Decentralized adaptive PID controllers stabilize and eliminate the static positioning errors of the robot system. No prior knowledge of robot dynamics is required for this scheme. Output transients caused by sensitivity to the system parameter variations is reducible by adjusting the gain constants of the decoupler. Computation load for the implementation of the decoupler is minimal due to its direct nature, The effectiveness of the scheme is demonstrated by simulation results in high speed repetitive motion trackings and load change conditions.