Key points are not available for this paper at this time.
Six different transmission types suitable for robotic manipulators were compared in an experimental and theoretical study. Single-degree-of-freedom mechanisms based on the different transmissions were evaluated in terms of force control performance, achievable bandwidth, and stability properties in hard contact tasks. Transmission types considered were (1) cable reducer, (2) harmonic drive, (3) cycloidal disk reducer, (4) cycloidal cam reducer, (5) ball reducer, and (6) planetary/cycloidal gear head. Open loop torque following error, attenuation and phase lag, and closed loop bandwidth and stability margin were found to be severely dominated by levels of inertia, stiffness distribution and variability, stiction, coulomb and viscous friction, and ripple torque. These aspects were quantified and shown to vary widely among all transmissions tested. The degree of nonlinearity inherent in each transmission affected its open and closed loop behavior directly, and limited the effectiveness of controller compensation schemes. Simple transmission models based on carefully measured transmission characteristics are shown to predict stability margins and achievable force-control bandwidths in hard contact to within a 5 to 15 percent error margin.
Schempf et al. (Wed,) studied this question.