본 논문에서는 수정된 작업 야코비안과 자체 충돌 방지를 위한 완화된 장벽 함수를 사용하여 상체 휴머노이드 로봇을 효과적으로 원격 조작하기 위한 VR 인터페이스 접근 방식을 제시합니다.
本稿では、上半身ヒューマノイドロボットの直感的かつ安全な遠隔操作を実現するため、修正タスクヤコビアンを用いたVRトラッカーの関節マッピングと、緩和バリア関数による自己衝突回避を統合した新規アプローチを提案する。
By modifying task Jacobians to simplify joint control mapping and incorporating relaxed barrier functions for real-time self-collision avoidance, this approach makes VR teleoperation of an upper-body humanoid robot more intuitive, safe, and efficient.
The author presents a reinforcement learning-based framework for real-time whole-body teleoperation of humanoid robots using only an RGB camera. The main thesis is to enable seamless integration of human cognitive skills with versatile humanoid capabilities through learning-based real-time teleoperation.