<p>This study introduces a novel approach for slip detection during the manipulation of deformable objects, focusing specifically on doll assembly, a process that demands high precision and delicacy. By integrating matrix tactile sponge sensors with long short-term memory (LSTM) networks, the proposed method captures detailed tactile data during handling, detecting subtle interactions with soft materials that are often missed by traditional sensing techniques. The LSTM networks process this data to model temporal dependencies, enabling the system to identify and predict slip events with remarkable accuracy. The approach addresses key challenges in automating tasks that involve deformable objects, bridging the gap between manual precision and robotic consistency. Moreover, the solution is designed to be scalable, providing a versatile framework that can be adapted to other manufacturing sectors requiring delicate manipulation of soft or flexible materials. By combining traditional production techniques with cutting-edge automation, this research supports the evolution of smart factories and demonstrates a pathway to automating complex processes previously thought to require human expertise. The results emphasize the potential of advanced sensor integration and machine learning in transforming manufacturing practices.</p>