This paper presents the design, fabrication, and control of a clothing-type artificial-muscle wearable suit that integrates electromagnetic (EM) f iber coils and shape-memory-alloy (SMA) filaments to realize distributed full-body actuation. The system functions as a flexible “robotic garment,” enabling active motion assistance, haptic feedback, and human-in-the-loop training under a Physical AI framework. The proposed hybrid actuator structure combines the rapid response of EM fibers with the high contraction ratio of SMA fibers, achieving both dynamic motion support and sustained load generation. The soft textile architecture incorporates a multi-layer sensor–actuator network capable of monitoring deformation, temperature, and contact pressure. Finite-element analysis (FEA) verified strain amplification and uniform stress distribution within the auxetic knit geometry, while thermal–electromagnetic co-simulation demonstrated controllable bidirectional motion. Experiments on a full-scale prototype achieved peak actuation stress of 18 kPa, average contraction strain of 4.1 %, and latency below 25 ms. Integrated feedback control using adaptive proportional-integral compensation ensured smooth force transitions during gait and arm lift tasks. The resulting platform provides an efficient route toward scalable, textile-integrated robotic systems for rehabilitation, industrial assistance, and embodied Physical AI research.This study emphasizes practical deployment feasibility by demonstrating stable operation under wearable constraints, including thermal safety, limited power supply, and distributed control scalability.