成果報告書詳細
管理番号20150000000603
タイトル*平成26年度中間年報 環境・医療分野の国際研究開発・実証プロジェクト ロボット分野の国際研究開発・実証事業 Choreonoidフレームワークを用いた災害対応ロボットシミュレータの研究開発(1)
公開日2016/1/27
報告書年度2014 - 2014
委託先名独立行政法人産業技術総合研究所
プロジェクト番号P12001
部署名ロボット・機械システム部
和文要約
英文要約Title: R&D of the disaster response robot simulator based on the Choreonoid framework(FY2014-FY2015)FY2014 Annual Report

(1) Improvement of compatibility with Gazebo
A simulation environment consists of (i) robot/environment models, (ii) control software and (iii) a simulator. In order to improve compatibility between Gazebo and Choreonoid, it is important to make it possible to share (i) robot/environment models and (ii) control software between these two simulators.
(a) Bidirectional conversion between URDF, SDF and VRML
We have developed a conversion tool called “simtrans”. It can convert URDF, SDF and VRML bilaterally. An input file is converted into an intermediate format once, and then converted to the specified format. This intermediate format makes it easy to support a new file format. Adding a convertor between the new format and the intermediate format enables conversions with other formats.
(b) Control software development platform
We extended an existing control software development platform “hrpsys-base” so that it can be used with Gazebo. As the result, control software developed using “hrpsys-base” can connect both simulators. Since “hrpsys-base” can work with both of RT-Middleware and ROS, users of both middleware can develop their software that can be used with both simulators.
(2) Accuracy improvement of simulator
(d) Identification of simulation models
We have developed a new approach to dynamic parameter identification for humanoid robots based on systematic exciting trajectory generation and optimization. The method derives the exciting trajectories with an iterative computation through observation of condition numbers of sub-regressor matrices created from the columns of the regressor matrix, while taking into account such constraints as balancing and mechanical limits. The identification method has been applied to identify mass parameters of HRP-2 humanoid robot. By using the proposed method, we could recognize changes of mass properties of the humanoid robot after mechanical modifications.
The performance of identification depends on the accuracy of the internal sensors mounted on a robot. We evaluated the accuracy of them by using an optical motion capture system and a force plate. The total external forces acting on the base-link were evaluated from the data obtained from the motion capture system and the internal sensors of the robot, respectively. The pilot measurement of humanoid robot HRP-4 showed that, the poor-calibrated force sensors leads the error of the generalized forces; identifying the gain matrix of the sensors could improve the accuracy of the estimation of the generalized forces.
The software for systematic excitation motion generation and parameter identification on MATLAB has been made available to public since the end of fiscal year 2014.
(3) Demonstration of simulator
We have released the enhanced Choreonoid with its manual, tutorials and its binary packages.
We have decided name, logo, date and place of the competition of disaster response robots by computer simulations. The name is “Japan Virtual Robotics Challenge” (JVRC for short). We promoted JVRC using leaflets at several domestic/international conferences on robotics.
ダウンロード成果報告書データベース(ユーザ登録必須)から、ダウンロードしてください。

▲トップに戻る