MINLP-Based Design Optimization of Backdrivable 3K Planetary Gear Drive for Robot Actuator

Abstract

The 3K planetary gear reducer has a higher gear reduction ratio and higher driving efficiencies for both forward and backward transmissions. It is particularly suitable for wearable exoskeleton robot actuator to improve the performance of human-robot interaction due to its backward driving ability. To further improve its forward and backward driving efficiencies with a given range of gear reduction ratios, the optimization model of the 3K planetary gear reducer is formulated. The design parameters include the number of gear tooth, modification coefficient and working pressure angle, which contain both integer and continuous variables. To guarantee the feasibility of the design, design constraints such as assembly constraints, contact ratio constraints, and modification coefficient constraints are established. The BONMIN solver in MATLAB optimization toolbox is employed to solve such an MINLP (Mixed Integer Nonlinear Programming) problem. The optimization results show that the forward and backward driving efficiencies of the optimized 3K planetary gear reducer are 91.5% and 91.0%, respectively, which validates the effectiveness of the proposed design method.