In this paper experimental results are presented to compare the performance of kinematically redundant parallel robots with respect to their non-redundant counterparts. The main purpose is to validate existing simulated, i.e. claimed, findings demonstrating the advantages of kinematic redundancy in terms of singularity avoidance and, therefore, accuracy and precision. Exemplarily, the kinematically redundant prototype of the Institute at Mechatronic Systems is introduced. It is based on the well known planar 3RRR mechanism. In order to achieve kinematic redundancy, a prismatic actuator is added to the structure allowing one base joint to move linearly. As a result, the mechanism is able to reconfigure, i.e. optimize, its geometry according to different performance criteria and motion strategies. While performing a geometrical reconfiguration and following several desired (optimized) trajectories the pose of the end-effector is determined using an external measurement device. Hence, in addition to the encoder data of the actuators the performance can be analyzed without using any (uncertain) kinematic models. This allows for a meaningful comparative evaluation of the performance of kinematically redundant mechanisms.

In this paper five different optimization strategies for kinematically redundant mechanisms, i.e. mechanisms having additional actuator(s) in at least one kinematic chain, are presented. They are based on two main approaches, a discrete optimization and a classical continuous optimization. Exemplarily, a planar, kinematically redundant 3RRR-based mechanism is introduced. The position of its redundant actuator, i.e. the robot geometry, is optimized according to an optimization criterion that is denoted as the gain of the maximal homogenized pose error. Several analysis examples demonstrate the effectiveness of kinematic redundancy with respect to the introduced optimization procedures. It is shown that in comparison to discrete approaches, classical continuousbased optimization strategies do not necessarily lead to more appropriate results in terms of performance improvement.

In the present paper we introduce an improved method to obtain the guaranteed singularity-free workspace of planar parallel kinematic machines. A geometric condition for the existence of singularities is extended to be used in interval analysis. Therefore, we eliminate the need of calculating the inconvenient interval form of the Jacobian's determinant. Hence, an appropriate description of the singularity-free workspace is obtained and the computational effort is reduced significantly. With the interval-based approach error sources, like manufacturing tolerances, can be considered. Consequently, regions that do not satisfy the proposed condition can be guaranteed not to have any type-two singularities. Several analysis examples demonstrate the efficiency of the proposed geometrical-based solver in comparison to existing solvers, i.e. based on the Jacobian's determinant.

In the present paper kinematic redundancy is proposed in order to improve the achievable pose accuracy of a parallel robot's traveling platform. An additional prismatic actuator is applied to a 3RRR planar parallel manipulator. It allows a selective reconfiguration of the kinematic structure according to several optimization criteria. Based on an index that we denote the gain of the Jacobian matrix the position of the redundant actuator is changed in a discrete manner. The resulting optimized switching patterns lead to a great increase of the traveling platform's pose accuracy. Between two switching operations the prismatic actuator's position is supposed to be locked. Hence, sources of error, e.g. the joint clearance, can be minimized. Several analysis examples demonstrate the effectiveness of the proposed concept in combination with the developed optimization procedure.

This article provides a tracking scheme for 6 DOFpsilas parallel robots. It is based on a controller/observer combination with an additional compensation of the desired dynamics. The latter is necessary to avoid an online calculation of the direct kinematics. However, it introduces systematic errors. To take such deviations into account a passivity based formalism is used. The resulting approach is experimentally substantiated and proven to be highly powerful and robust against initial errors and model uncertainties.

This article provides a systematic insight into planning time-minimal trajectories for 6-dof parallel manipulators. Most of wide spread model simplifications, especially known from serial manipulators, are removed in order to give a general solution. All rigid-body forces, dry friction, viscous damping in passive and in actuated joints are considered. This yields a considerable escalation of the task’s complexity. Therefore, an adapted algorithm is presented, that ensures the computational efficiency of trajectory planning, even in cases of high complex dynamics.

This paper presents a complete approach for parametrization of model- and knowledge-based controller for parallel robots. By combining and merging methodologies from mechanics, system theory, information processing and intelligent control, an accurate and compact method resulted and is substantiated with experimental results achieved on an innovative hexapod PaLiDA. An appropriate form of excitation trajectories helps to overcome classical identification problems, like disturbances in the acceleration signals. The Gauss-Markov estimator is applied for solving the over determined linear equation system. A novel method is presented that uses statistical and uncertainty attributes of the estimate for choosing an optimal structure and parameter number of the dynamics model.

In this paper, a complete framework for the task of grasping a naturally textured object is being presented. The lack of calibration in the components of the eye-in-hand system is overcome by an image trajectory based visual servoing scheme (ITBVS). Using an object model reconstructed off-line, consecutive rough pose estimation and refinement of the estimate by scale optimized 3D model to object registration solves the feature correspondence problem in the absence of artificial landmarks. Experimental results using a zoom camera and a 7-DOF serial manipulator prove the effectiveness of the approach.