AI-Driven Adaptive Control System for Soft Robotics: Bridging Perception and Actuation in Dynamic Environments

This journal discusses the challenges of controlling soft robotics operating in dynamic and unstructured environments, primarily due to their flexible, nonlinear material properties and multiple degrees of freedom. The study aims to map how artificial intelligence can enhance adaptive control systems by bridging sensory perception and robot action. The method used is a PRISMA-based Systematic Literature Review of 25 empirical articles from the period 2020–2025 from ScienceDirect. The findings indicate that AI-based adaptive control, particularly reinforcement learning, sensor fusion, and hybrid architectures, can improve real-time response, task accuracy, energy efficiency, and the robot's adaptability. The implication is that this technology is crucial for the development of medical robots, assistive devices, and exploration systems that are safer, more autonomous, and responsive.