Constraint Mechanism Design for Fast Dielectric Elastomer Actuator-Driven Capsule Robots

Dielectric elastomer actuator (DEA)-driven capsule robots convert periodic excitations into net translation through passive constraints, yet a systematic study on optimal constraint design remains largely unexplored. To address this gap, we present a comparative numerical study of six passive constraint families that range from linear and bistable springs to limit stops and polynomial shapes and encompass 2447 configurations under a reference condition. The best configuration of each family spans more than an order of magnitude in speed, from 2.62 mm/s (linear spring) to 33.86 mm/s (asymmetric bistable). We find that the sign of the constraint-force derivative at the rest position separates every family with negative rest-position stiffness from every family with positive stiffness, whereas the second derivative (curvature) shows no correlation with speed. Building on this observation, we formulate a library-based inverse-design framework that treats velocity as a specification and minimizes the peak driving force. The comparisons reported here are made under one reference condition, and the ranking is therefore a property of the library as sampled rather than a general design rule. For example, relaxing the target velocity from 25 mm/s to 23 mm/s reduces the peak force from 0.31 N to 0.19 N. These results move passive-constraint design from empirical trial-and-error toward a requirement-driven methodology for future soft capsule robots.