In the Departement of Aeronautics and Space Engineering of Tohoku University in Sendai.

The Space Machinery Laboratory is directed by Professor Masaru UCHIYAMA.

Touch :

Between August 1st 1997 and July 31st 1999.

Using a grant provided by the JSPS (Japan Society for the Promotion of Science).

(JSPS : 5-3-1 Kojimachi, Chiyoda, Tokyo 102, Phone : +81-3-3263-1721)

• Introduction : my present research

CAD/CAM systems provide the necessary tools to perform realistic animations of robots. However, these systems are not widely used for the off-line programming of robots because they fail to generate trajectories which accurately depict those expected from the actual robot in its environment. Robot calibration is the determination of the parameters of a kinematic model that most accurately depicts an actual robot. One result of robot calibration is that the realistic behaviour of the robotics cell is being closely simulated by the CAD/CAM system. Several comprehensive studies cover the modelling, measurement, identification and compensation techniques necessary to perform robot calibration. However, the problem of robot calibration fundamentally relies on the techniques used to perform data acquisition during the measurement stage. Commercial systems are available to obtain the data but are expensive, time consuming and difficult to implement in an industrial environment. The focus of my research was the development of kinematic calibration procedures. The result was a method allowing one to perform robot registration in the environment and to identify the entire set of robot kinematic parameters necessary to improve the robot model accuracy. The goal of my research was to reduce the cost, duration and complexity of the measurement phase. Accordingly, the proposed methods are based on the use of a low cost displacement sensor which can be easily mounted on the robot end-effector. This sensor allows detecting specific targets placed in the robot work area or to constrain the robot end-effector to remain parallel to a physical plane secured in the cell. The kinematic error model is then derived from the detected target geometry or from the performed constraint. With the validation of these procedures to the calibration of serial structures, the new objective becomes to extend our procedures to the calibration of parallel robots.
• Proposed research in Japan

Many comprehensive works have also been written in the area of parallel robots. Parallel architectures in comparison to serial robots provide improved dynamic capabilities, increased rigidity and high positioning repeatability. These features make these robots appropriate for several applications such as pick and place and conditioning tasks. Some recent applications concern the development of new machine tools.

The goal of my research project is to develop calibration techniques for parallel robot and in particular for the existing HEXA robot and the new parallel manipulator being developed in the Department of Aeronautics and Space Engineering of Tohoku University. This new robot will be one of the fastest six degree-of-freedom parallel manipulators. The developments that have already been done in the laboratory include test of fast motion, dynamic control for precise trajectory motion, kinematic control with singularity consistency and compliance control dextrous motion. The kinematic calibration technique, however, has not been developed yet even though this technique is very important for the control of the manipulator. Both the development of kinematic calibration techniques and experimental tests with the manipulators have to be performed. In addition, software packages for the calibration and packages to control the manipulator have to be developed.

• Expected outcomes

An application in which calibration is strongly needed is a new project of the Aeronautics and Space Engineering Department concerning the development of a moving base simulator. Using the HEXA robot as a model of this simulator, the kinematic calibration will guarantee accurate simulation of the moving base that is floating in space or fixed on a flexible structure. A close link between the parallel robot calibration and the moving base simulator development will be kept. Moreover, the synthesis of calibration techniques should lead to a real improvement of the parallel manipulators accuracy and thus should contribute to the design of new accurate machine tools. The development of these machine tools is one of the new research topics of our laboratory.

Finally, this project will ensure the continuation of the productive co-operation between Tohoku University and the LIRMM.

• Beginning in 1991, the co-operation between Tohoku University, the LIRMM and Toyoda (Toyota Machine Tool Department) led to the development of the HEXA manipulator.
• Other joint studies have concerned the development of new schemes for the control of two-arm robots : two researchers from LIRMM worked in Tohoku University during 1993.
• During 1994-1995, the CNRS-MONBUSHO co-operation involving the LIRMM and Tohoku University dealt with fast and accurate new parallel robots.

HexaM Machine Tool
(source: TOYODA home page)