Pareto size optimization study of a y-shaped six-axis force/moment sensor: an alternative to cross-beam design

Abstract

Purpose Six-axis force/moment sensor studies traditionally focused only on maximizing sensitivity or isotropy. This study aims to expand optimization by also considering structural stability, dynamic characteristics and sensor size, with the goal of developing a y-shaped sensor body for robot end effector with high force/moment capacity (250 N/25 Nm), high structural stability (σvm lt;72 MPa) and high fundamental frequency (ωn gt; 4,000 Hz) while maintaining 1% accuracy and high sensitivity relative to size. Design/methodology/approach A novel optimization problem formulation is developed that cannot be solved with a generic weighted sum approach. Instead, Pareto optimization is performed using a population-based genetic algorithm, with the optimal solution selected via the technique for order of preference by similarity to ideal solution (TOPSIS) multi-criteria decision-making technique. Sensitivity analysis is carried out using random balance designs to determine the influence of sensor dimensions on Pareto objectives and constraints. The selected design is prototyped, calibrated and experimentally characterized using custom-built test benches. Findings The optimization process is less than 6 min and the relative error is under 5% compared to the finite element model. The y-shaped sensor prototype has 0.44% nonlinearity, 0.25% repeatability, 0.24% hysteresis, 0.31% time drift and 0.80% crosstalk error, leading to 1.03% overall accuracy. The strain-based y-shaped and cross-beam sensor prototypes optimized with the same methodology revealed that the y-shaped sensor has higher sensitivity due to coupled Wheatstone bridges while maintaining equivalent accuracy and compactness. Originality/value This study introduces a novel sensor optimization formulation and combines Pareto optimization with genetic algorithm, TOPSIS selection and random balance designs in multi-axis sensor design. It provides an efficient and accurate design methodology, and investigates the advantages of y-shaped sensor structures over conventional cross-beam types, contributing to the development of more compact, sensitive and structurally stable multi-axis sensors.