| We('1) advance computational robot dynamics to automate dynamic robot modeling for computer control engineering (including mechan- ical design, dynamic simulation, linearized and trajectory sensitivity modeling, parameter identification, and controller design and imple- mentation). Computational robot dynamics synthesizes classical mechanics and list processing to address the symbolic and numeric efficiencies of robot dynamics formulations and algorithms. We develop customized algorithms to compute the dynamics of modern six degree-of-freedom manipulators in less than one millisecond for uniprocessor real-time controllers. Our software package, Algebraic Robot Modeler (ARM) applies seven robot dynamics formulations to generate symbolic closed-form, recursive, linearized and discrete dynamic robot models for physical interpretation and engineering insight. We implement a systematic organization procedure to gener- ate customized algorithms by recognizing and removing redundant computations from both closed-form and recursive dynamic models.;kinematic and dynamic structure of the manipulator and the elimi- nation of repetitive calculations appearing throughout dynamic models.;('1)For expository convenience, the author has chosen to use such personal pronouns as "we" and "our".;Our classical Lagrange-Christoffel formulation is the most sym- bolically efficient generator of closed-form dynamic robot models. From our comparative evaluation of the computational requirements of customized algorithms (for fourteen benchmark manipulators), customized closed-form algorithms for three DOF manipulators out-perform customized recursive algorithms (which, in turn, out-perform general-purpose algorithms). For six DOF predominantly rotational manipulators, customized recursive algorithms out-perform custom- ized closed-form algorithms (which, in turn, out-perform general-purpose algorithms). The increased numeric efficiencies over doc- umented implementations stem from the exploitation of the entire. |
