This thesis demonstrates that cost-effective, reliable embedded acquisition systems and biocooperative controls can enhance robotic-assisted upper-limb rehabilitation. The main contributions include the development of novel EMG and multimodal wearable platforms, validation of EMG-driven and multimodal control strategies, and evidence of improved user interaction through physiological feedback. Collectively, the results support the feasibility of translating biocooperative systems into real-world clinical practice by lowering costs and improving usability. Future research should focus on clinical validation with stroke patients, usability studies, and expansion of multimodal approaches to include more comprehensive emotional and cognitive monitoring, fostering widespread clinical adoption of these technologies.