日本フィジカルAI新聞

世界のフィジカルAIを、日本語で。

歩行arXiv:1703.10337

不整地歩行のためのヒューマノイドオンライン適応アルゴリズム

Online Adaptation for Humanoids Walking On Uncertain Surfaces

シェア:XThreadsFacebookLINEはてブBluesky

高さの不確かな不整地を二足歩行するため、遊脚が最大高さに達した後に接地を検出して軌道をオンライン修正し、滑らかな着地を実現する適応アルゴリズムを提案し、SURENA IIIで実験検証した。

著者: Majid Khadiv, S. Ali. A. Moosavian, Aghil Yousefi-Koma, Hessam Maleki, Majid Sadedel

分類: cs.RO

原文アブストラクト

In this paper, an online adaptation algorithm for bipedal walking on uneven surfaces with height uncertainty is proposed. In order to generate walking patterns on flat terrains, the trajectories in the task space are planned to satisfy the dynamic balance and slippage avoidance constraints, and also to guarantee smooth landing of the swing foot. To ensure smooth landing of the swing foot on surfaces with height uncertainty, the preplanned trajectories in the task space should be adapted. The proposed adaptation algorithm consists of two stages. In the first stage, once the swing foot reaches its maximum height, the supervisory control is initiated until the touch is detected. After the detection, the trajectories in the task space are modified to guarantee smooth landing. In the second stage, this modification is preserved during the Double Support Phase (DSP), and released in the next Single Support Phase (SSP). Effectiveness of the proposed online adaptation algorithm is experimentally verified through realization of the walking patterns on the SURENA III humanoid robot, designed and fabricated at CAST. The walking is tested on a surface with various flat obstacles, where the swing foot is prone to either land on the ground soon or late.

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

FRONTIER(GPU搭載BTOパソコン)

FRONTIER(GPU搭載BTOパソコン)の公式広告バナーFRONTIERでGPU搭載PCを見る ↗
PR本紙発行元 EmplifAI