The article discusses the development of ReachBot, a robotic platform utilizing extendable booms for mobility in challenging environments like martian caves. The platform acts as a parallel robot, enabling manipulation-focused scientific objectives through tools operation and sample handling. The two-part solution presented optimizes for robustness against task uncertainty and stochastic failure modes. A mixed-integer stance planner determines boom positioning to maximize task wrench space, while a convex tension planner calculates boom tensions for desired task wrenches. Improvements in robustness metrics are demonstrated, along with an increase in the manipulation workspace volume. Monte-Carlo simulation validates the methods' robustness across randomized tasks and environments, including cable-driven morphologies.
다른 언어로
소스 콘텐츠 기반
arxiv.org
더 깊은 질문