摘要：The paper is devoted to the elastostatic calibration of industrial robots, which is used for precise machining of large-dimensional parts made of composite materials. In this technological process, the interaction between the robot and the workpiece causes essential elastic deflections of the manipulator components that should be compensated by the robot controller using relevant elastostatic model of this mechanism. To estimate parameters of this model, an advanced calibration technique is applied that is based on the non-linear experiment design theory, which is adopted for this particular application. In contrast to previous works, it is proposed a concept of the user-defined test-pose, which is used to evaluate the calibration experiments quality. In the frame of this concept, the related optimization problem is defined and numerical routines are developed, which allow generating optimal set of manipulator configurations and corresponding forces/torques for a given number of the calibration experiments. Some specific kinematic constraints are also taken into account, which insure feasibility of calibration experiments for the obtained configurations and allow avoiding collision between the robotic manipulator and the measurement equipment. The efficiency of the developed technique is illustrated by an application example that deals with elastostatic calibration of the serial manipulator used for robot-based machining.

摘要：The design of H∞ reduced order controllers is known to be a non-convex optimization problem for which no generic solution exists. In this paper, the use of Particle Swarm Optimization （PSO） for the computation of H~ static output feedbacks is investigated. Two approaches are tested. In a first part, a probabilistic-type PSO algorithm is defined for the computation of discrete sets of stabilizing static output feedback controllers. This method relies on a technique for random sample generation in a given domain. It is therefore used for computing a suboptimal Ha static output feedback solution, In a second part, the initial optimization problem is solved by PSO, the decision variables being the feedback gains. Results are compared with standard reduced order problem solvers using the COMPIeib （Constraint Matrix-optimization Problem Library） benchmark examples and appear to be much than satisfactory, proving the great potential of PSO techniques.