Machine learning for the design of biodegradable materials in soft robotics

Abstract Biodegradable materials are emerging as key enablers of next-generation soft robotic systems designed for biomedical, environmental, and transient applications. In parallel, machine learning (ML) has become a powerful tool for accelerating materials discovery, property prediction, and process optimization. This review examines the convergence of biodegradable materials, soft robotics, and ML from a material-centered perspective, focusing on how data-driven approaches can bridge polymer chemistry, processing conditions, and degradation behavior with application-specific functional requirements. We discuss recent advances in ML-enabled prediction of mechanical properties, degradation kinetics, and processability, as well as emerging strategies for inverse design, multitask learning, and closed-loop optimization. Particular emphasis is placed on the role of ML in navigating trade-offs between compliance, durability, and controlled biodegradation, as well as in optimizing high-dimensional process–structure–property relationships. A critical comparison of modeling approaches highlights the strengths and limitations of neural networks, interpretable models, and data-driven frameworks in the context of biodegradable soft robotic systems. Key challenges, including data scarcity, lack of standardized datasets, limited model interpretability, and scalability constraints, are analyzed with reference to recent studies and application scenarios. Finally, we outline future research directions, including multi-stimuli responsive biodegradable materials, lifecycle-aware datasets, and ML-guided design frameworks for adaptive and sustainable soft robotic systems. This integrated perspective provides a unified framework for advancing ML-driven development of environmentally compatible soft robotic materials.