Efficient robot dynamics for high sampling rate motions: case studies and benchmarks

Abstract

The implementation of on-line algorithms for robot dynamics is made difficult by the complex nature of the equations. However, most of the advanced control techniques, especially model-based algorithms, require the inclusion of the dynamic model in the ‘control-law’. In this case, the dynamics are computed at each sampling instant. Hence, the development of simplified and computationally viable dynamic models is crucial for the enhancement of controller design. These models should satisfy two criteria. First, the simplification of the model should not lead to loss of accuracy which will destabilize the robot performance. Second, the model must meet real-time constraints and help to achieve high sampling rate motions. In this paper, a simplified formulation of robot dynamics based on the Lagrange-Euler is introduced. However, the work emphasizes the computational efficiency of this technique as compared to some existing ones. The advantage of the proposed technique emanates from its simple and well-defined structure which lends itself to parallel processing applications. The results of real-time implementations and parallel realization are also included.