日本フィジカルAI新聞

世界のフィジカルAIを、日本語で。

モデル縮約arXiv:1705.10143

パラメータ削減によるマルチボディモデルの簡略化

Simplification of multibody models by parameter reduction

シェア:XThreadsFacebookLINEはてブBluesky

マルチボディ系の力学モデルをモデル選択手法で少数パラメータに縮約し、精度を保ちつつ計算コストを大幅に削減する手法を提案・検証した論文。

著者: Javier Ros, Xabier Iriarte, Aitor Plaza, Vicente Mata

分類: cs.RO

原文アブストラクト

Model selection methods are used in different scientific contexts to represent a characteristic data set in terms of a reduced number of parameters. Apparently, these methods have not found their way into the literature on multibody systems dynamics. Multibody models can be considered parametric models in terms of their dynamic parameters, and model selection techniques can then be used to express these models in terms of a reduced number of parameters. These parameter-reduced models are expected to have a smaller computational complexity than the original one and still preserve the desired level of accuracy. They are also known to be good candidates for parameter estimation purposes. In this work, simulations of the actual model are used to define a data set that is representative of the system's standard working conditions. A parameter-reduced model is chosen and its parameter values estimated so that they minimize the prediction error on these data. To that end, model selection heuristics and normalized error measures are proposed. Using this methodology, two multibody systems with very different characteristic mobility are analyzed. Highly considerable reductions in the number of parameters and computational cost are obtained without compromising the accuracy of the reduced model too much. As an additional result, a generalization of the base parameter concept to the context of parameter-reduced models is proposed.

関連論文

  1. 1Boston Dynamics、車輪付きAtlasの研究試作機を紹介研究・技術
  2. 2newmoとKDDIが資本業務提携 2028年に堺市で自動運転タクシー商用化目指す提携・買収
  3. 3NVIDIA、CoRLでGR00T 2とIsaac Lab 3.0を発表新製品・発表
  4. 4GXOとApptronik、Apolloの物流現場試験を拡大 最大1,000台の展開を構想量産・展開
  5. 5UC BerkeleyとNVIDIAの研究者、AIエージェントでロボット制御コードを仮想環境で改善研究・技術

PR・アフィリエイト広告

OZ GAMING(GPU搭載BTOゲーミングPC)

OZ GAMING(GPU搭載BTOゲーミングPC)の公式広告バナーOZ GAMINGで完成品PCを見る ↗
PR本紙発行元 EmplifAI