Gaussian Process-Based Gait Optimization of a Cable-Driven Soft Quadruped Robot
Jeongil Choi, Ayoung Hong
AI summary
Problem
Soft robots' deformable nature makes accurate modeling and control difficult, forcing reliance on physical experiments for gait optimization. However, exhaustive testing becomes impractical as the number of gait parameters increases.
Approach
The authors implemented a Gaussian Process surrogate model to guide physical experiments, iteratively selecting promising gait parameters to maximize travel distance while balancing exploration and exploitation.
Key results
- Near-optimal gait parameters identified in ~13 iterations
- GP framework successfully avoided local optima via LCB acquisition
- Validated on a 3D-printed, cable-driven soft quadruped prototype
- Demonstrated significant reduction in required physical trials vs. grid search
Why it matters
Enables researchers and roboticists to efficiently tune soft robot locomotion without exhaustive physical testing, accelerating development of adaptive soft machines.
Abstract
Soft robots offer significant advantages over rigid- bodied counterparts due to their inherent flexibility and de- formability. However, these same characteristics often introduce challenges in control and simulation-to-reality (Sim-to-Real) synchronization. In this study, we quantified the locomotion performance of a soft quadruped robot through experimental trials. To alleviate the burden of exhaustive physical testing over an expanded parameter space, we employed a Gaussian process-based optimization framework.