 The study aimed to develop a wearable robotic limb system to assist individuals with upper limb motor disorders due to stroke. The system utilized a hybrid control approach combining motor imagery, MI, and object detection to provide assistance for movement and grasping tasks. The MIEEG recognition method based on graph convolutional neural networks, GCN, and gated recurrent units, GRUs, was employed to recognize the subject's intentions from EEG signals. The system was tested offline and online, where the subjects were able to successfully complete the target object grasping tasks within 23 seconds. The results demonstrated that the proposed MIEEG recognition method improved the MI classification accuracy compared to traditional methods, and the SRL system was effective in providing assistance for upper limb tasks in daily life. This article was authored by Zhichuan Tang, Lingtao Zhang, Xiyan Chen, and others. We are article.tv, links in the description below.