慣性計測からの姿勢推定:動的用途におけるルンゲ効果の緩和
Attitude Reconstruction from Inertial Measurement: Mitigating Runge Effect for Dynamic Applications
等間隔サンプリングの慣性計測で生じるルンゲ現象を、隣接サンプルを利用し中央区間のみ使うBAC戦略で抑え、姿勢計算の精度を大幅に改善した論文。
著者: Yuanxin Wu, Maoran Zhu
分類: cs.RO, cs.SY, eess.SY
原文アブストラクト
Time-equispaced inertial measurements are practically used as inputs for motion determination. Polynomial interpolation is a common technique of recovering the gyroscope signal but is subject to a fundamentally numerical stability problem due to the Runge effect on equispaced samples. This paper reviews the theoretical results of Runge phenomenon in related areas and proposes a straightforward borrowing-and-cutting (BAC) strategy to depress it. It employs the neighboring samples for higher-order polynomial interpolation but only uses the middle polynomial segment in the actual time interval. The BAC strategy has been incorporated into attitude computation by functional iteration, leading to accuracy benefit of several orders of magnitude under the classical coning motion. It would potentially bring significant benefits to the inertial navigation computation under sustained dynamic motions.
