Upper limb moto disorders are the main symptoms of stroke patients. Based on deep learning algorithms and object detection technology, we developed a brain-controlled supernumerary robotic limb system for upper-limb motion assistance. The system makes use of the motor imagery electroencephalogram (MI EEG) recognition model with graph convolutional network (GCN) and gated recurrent unit network (GRU) to obtain the patient’s motion intentions and control the supernumerary robotic limb to move. The object detection technology can compensate for the disadvantages when using MI EEG alone like fewer control instructions and lower control efficiency. We also validated the feasibility and effectiveness of the system by designing model training experiment and target object grasping experiment. The results showed that the highest EEG classification accuracy using GCN+GRU algorithm achieved 92.32%, and the average success rate of grasping tasks achieved 88.67±3.77%.