Inverse Dynamics of NonRedundant ClosedChain Mechanisms.pdfVIP

Inverse Dynamics of Non-Redundant Closed-Chain Mechanisms ERIK SPENCER Abstract—The following paper serves as an overview and interpretive framework for a particular set of closed chain inverse dynamic derivations. These derivations are generally applicable to robots containing revolute or prismatic joints and are described by a known set of independent coordinates. A reduced form of the dynamics is formulated by inclusion of the closed-chain constraints. Such representations are useful for control and simulation. I. INTRODUCTION Solving for the dynamics of closed-chain mechanisms generally involves converting the constrained system into an open tree structure by virtually cutting an unactuated j oint of the closed chain. The dynamics of the open tree structure are found using well known methods [1]. Constraints are then reapplied at the virtually cut joints, reducing the system to independent coordinates, thus allowing representation of the closed form dynamics, τc = Mq + V(q,q) + G(q). (1) A c represents the Actuated torques of the closed chain. The subscript P will be used A rather than A in reference to the passive j oints of the closed chain. The superscript o will be used rather than c in reference to the open tree structure. For the open tree structure of the robot, let n denote the total number of j oints and let m denote the number of passive j oints. For non-redundant closed chain mechanisms, there will be n-m j oints requiring actuation. These two variables will often be included as subscripts indicating matrix dimension for clarity and quick reference. It is convenient to represent the forces and j oint vari