ヒューマノイドロボットの状態推定フレームワーク
State Estimation for a Humanoid Robot
脚の運動学と固有感覚センサのみを用い、足裏の回転制約を考慮したEKFでヒューマノイドの状態を推定する手法を提案し、歩行・スクワット動作のシミュレーションで有効性を示した。
著者: Nicholas Rotella, Michael Bloesch, Ludovic Righetti, Stefan Schaal
分類: cs.RO
原文アブストラクト
This paper introduces a framework for state estimation on a humanoid robot platform using only common proprioceptive sensors and knowledge of leg kinematics. The presented approach extends that detailed in [1] on a quadruped platform by incorporating the rotational constraints imposed by the humanoid's flat feet. As in previous work, the proposed Extended Kalman Filter (EKF) accommodates contact switching and makes no assumptions about gait or terrain, making it applicable on any humanoid platform for use in any task. The filter employs a sensor-based prediction model which uses inertial data from an IMU and corrects for integrated error using a kinematics-based measurement model which relies on joint encoders and a kinematic model to determine the relative position and orientation of the feet. A nonlinear observability analysis is performed on both the original and updated filters and it is concluded that the new filter significantly simplifies singular cases and improves the observability characteristics of the system. Results on simulated walking and squatting datasets demonstrate the performance gain of the flat-foot filter as well as confirm the results of the presented observability analysis.
