Soft robots built from compliant materials and deformable structures are increasingly used in medical intervention, wearable assistance, delicate manipulation, and environmental exploration, where conventional rigid robots are limited by high mechanical impedance and poor morphological adaptability. However, their transition from laboratory prototypes to deployable systems remains constrained by coupled bottlenecks in materials, actuation, sensing, modeling, control, energy supply, and manufacturing. This review summarizes recent advances in soft robotics through an evaluative framework covering actuation and materials, modeling and simulation, control strategies, multimodal sensing, and representative applications. Instead of treating these topics as independent descriptions, we compare the underlying mechanisms, measurable performance indicators, strengths, limitations, and application boundaries. Three conclusions emerge. First, no single actuation strategy can simultaneously maximize output force, response speed, energy efficiency, durability, miniaturization, and untethered operation. Second, high-fidelity continuum models improve physical accuracy but remain difficult to use for real-time control, whereas reduced-order and data-driven models improve efficiency at the cost of generalization, interpretability, or contact fidelity. Third, practical soft robots will depend on system-level integration of embedded sensing, physics-informed learning, robust control, reliable materials, and scalable fabrication. Future progress should therefore prioritize standardized benchmarks, lifecycle reliability, energy-autonomous operation, and task-specific comparisons with rigid robotic systems.
Wang et al. (Thu,) studied this question.