Alain Codourey

Adaptive control of the Hexaglide, a 6 dof parallel manipulator

This paper presents the dynamic equation, nonlinear control and dynamic parameters identification of the Hexaglide, a new 6-DOF parallel manipulator intended to be used as a high speed milling machine. Using a method based on the virtual work principle, the dynamic equation is found in a compact linear form and then used in a nonlinear adaptive control algorithm based on the minimization of the tracking error. The dynamic parameters are learned during motion and introduced in an inverse dynamic model used as a feedforward compensator. Specific trajectories, exciting the parameters separately, enable a fast stepwise learning of the 12 parameters. A simulation demonstrates the validity of the approach.

Dynamic Modeling of Parallel Robots for Computed-Torque Control Implementation

In recent years, increased interest in parallel robots has been ob served. Their control with modern theory, such as the computed- torque method, has, however, been restrained, essentially due to the difficulty in establishing a simple dynamic model that can be calcu lated in real time. In this paper, a simple method based on the virtual work principle is proposed for modeling parallel robots. The mass matrix of the robot, needed for decoupling control strategies, does not explicitly appear in the formulation; however, it can be computed separately, based on kinetic energy considerations. The method is applied to the DELTA parallel robot, leading to a very efficient model that has been implemented in a real-time computed-torque control algorithm.

Dynamic modelling and mass matrix evaluation of the DELTA parallel robot for axes decoupling control

The recent development of the direct drive DELTA, a 3 to 4 degree-of-freedom parallel robot dedicated to pick and place operations, opened a number of new perspectives in the domain of fast and accurate handling of light objects. The elimination of reduction gearing in these designs although improving acceleration capabilities has the disadvantage of increasing mechanical couplings between robot axes and reflects the full varying inertias directly onto each motor axis. To resolve this problem, the computed-torque control method has been applied to many serial robots over the past few years. Its application to parallel robots has however been restrained by the difficulty in establishing a simple dynamic model that can easily be calculated in real time. In this paper, an efficient method based on the direct application of the virtual work principle is proposed. Contrarily to the other methods developed up to now, it has the advantage of providing explicitly the mass matrix of the robot, which can be used in the control algorithm for axes decoupling. The use of the dynamic model in a feedforward control strategy leads to an improvement of trajectory tracking by a factor 6.

A body-oriented method for finding a linear form of the dynamic equation of fully parallel robots

In order to identify the dynamic parameters in nonlinear adaptive control the robots dynamic equation has to be written in a linear form. Many methods have been proposed for serial robots, but for parallel robots, the few solutions proposed so far lead to complicated equations that are not readily usable for real-time implementation. In this paper we propose a new method based on the virtual work principle to find a linear form of the dynamic equation of robots. Compared to other methods, it has the advantage that it does not need to open the closed loop structure into a tree-structure robot. It considers rather each body separately using its Jacobian matrix to project the forces into the joint space of the robot. Thus, simplification can be made at the very beginning of the modeling. This is very efficient when used to model fully parallel robots. As an illustration, the proposed method is applied to the 3dof DELTA parallel robot.

Visual control of a microrobot operating under a microscope

In this paper the vision-based control of a microrobot operating under a light microscope is presented. For high accuracy manipulation the parameters describing the relationship between the reference frame and the frame attached to the robot must be continuously and accurately updated These parameters are unknown a priori and are not directly measurable, but can be deduced by observing the scene change during the motion. The relative positioning accuracy is about 100 nm and trajectories can be precisely followed in 3 degrees of freedom. Experimental results are presented.

A task-oriented teleoperation system for assembly in the microworld

Operating a robot in the micro/nano-world presents challenges not found in the macro-world. Due to the small dimensions, the operator has no direct access to the objects and must assemble them by teleoperation. To allow a good manipulability, we propose a mixed teleoperation system which is composed of both direct and task-oriented teleoperation modes. The operator is provided with a set of visualization and manipulation tools on a workstation. This paper presents the micro-robot system, discusses some important issues in micro-teleoperation and finally presents some micro-manipulation tasks performed with the system.

Experimental evaluation of nonlinear adaptive controllers

Attractive methods for learning the dynamics and improving the control of robot manipulators during movements have been proposed for more than 10 years, but they still await applications. This article investigates practical issues for the implementation of these methods, Two nonlinear adaptive controllers, selected for their simplicity and efficiency, are tested on 2-DOF and 3-DOF manipulators. The experimental results show that the adaptive feedforward controller (AFFC) is well suited for learning the parameters of the dynamic equation, even in the presence of friction and noise. The control performance along the learning trajectory and other test trajectories are also better than when measured parameters are used. However, when the task consists of driving a repeated trajectory, the adaptive lookup table MEMory is simpler to implement. It also provides a robust and stable control, and results in even better performance.

Evaluation of parametric and nonparametric nonlinear adaptive controllers

In this paper, nine adaptive control algorithms are compared. The best two of them are tested experimentally. It is shown that the Adaptive FeedForward Controller AFFC) is well suited for learning the parameters of the dynamic equation, even in the presence of friction and noise. The resulting control performance is better than with measured parameters for any trajectory in the workspace. When the task consists of repeating the same trajectory, an adaptive look-up-table MEMory, introduced and analyzed in this paper, is simpler to implement and results in even better control performance.

Inertial drives for micro- and nanorobots: analytical study

The need for high precision robots dedicated to the assembly of microsystems has led to the design of new kinds of actuators able to reach very high positional accuracy over large distances. Among these, inertial sliders have received considerably interest in the last years. They have the advantage of being based on a simple principle that leads to a simple mechanical design. However, because they are based on the nonlinearity of friction, it is not easy to predict their stepsize repeatability. In order to understand the most important parameters affecting the precision of inertial drives, a theoretical study of a 1 degree of freedom inertial slider has been established. Analytical formulas describing the influence of different parameters, such as static and dynamic friction and mass distribution, have been developed. The effect of applied functions (sawtooth and parabolic), have also been studied. The theoretical cut off frequency has been found for each of the different waveforms, allowing us to predict the maximal and minimal working frequencies for the system. Thus, for each curve form, the repeatability of inertial sliders can be evaluated taking into account the uncertainties in the friction coefficients. The best suited waveforms for given constraints can therefore be selected. Simulations carried out from this have been successfully compared to experimental results.

Experiments in nonlinear adaptive control

In this paper, 2 control algorithms for learning and compensating for the dynamics of manipulators during the motion are presented and tested experimentally. It is shown that the adaptive feedforward controller (AFFC) is well suitable for learning the parameters of the dynamic equation, even, in presence of friction and noise. The resulting control performances are better than with measured parameters for any trajectory in the workspace. When the task consists of driving a repeated trajectory, an adaptive look-up-table MEMory, introduced and analyzed in this paper, is however simpler to implement and results in even better control performances.