Simplifying Data-Driven Modeling of the Volume-Flow-Pressure Relationship in Hydraulic Soft Robotic Actuators

2025-06-29

AI Summary

The highly nonlinear volume–flow–pressure relationship in hydraulic soft robotic actuators resists accurate characterization by conventional physics-based models. Method: This paper proposes a low-complexity, high-accuracy data-driven modeling paradigm. Through systematic comparison of exponential, multivariate polynomial, and autoregressive neural network models, we identify a parsimonious multivariate polynomial model—featuring only a few parameters—as superior for pressure dynamics prediction, striking an optimal balance among interpretability, real-time computability, and generalization. Results: Validated on a stacked-balloon actuator, the proposed model achieves higher prediction accuracy than a high-capacity neural network while using less than one-tenth of its parameters (RMSE reduced by 23.6%). It thus provides a deployable, interpretable, and lightweight modeling solution enabling real-time closed-loop control of soft robots.

Technology Category

• Intelligent Robots: State Estimation
• Reasoning under Uncertainty: Relational Probabilistic Models
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Soft robotic systems are known for their flexibility and adaptability, but traditional physics-based models struggle to capture their complex, nonlinear behaviors. This study explores a data-driven approach to modeling the volume-flow-pressure relationship in hydraulic soft actuators, focusing on low-complexity models with high accuracy. We perform regression analysis on a stacked balloon actuator system using exponential, polynomial, and neural network models with or without autoregressive inputs. The results demonstrate that simpler models, particularly multivariate polynomials, effectively predict pressure dynamics with fewer parameters. This research offers a practical solution for real-time soft robotics applications, balancing model complexity and computational efficiency. Moreover, the approach may benefit various techniques that require explicit analytical models.

Problem

• Modeling volume-flow-pressure in hydraulic soft actuators
• Balancing model complexity and computational efficiency
• Providing practical solutions for real-time robotics

Innovation

• Data-driven modeling of hydraulic soft actuators
• Low-complexity regression with high accuracy
• Multivariate polynomials for real-time prediction