A Robotic Arm Grasping System Based on Computer Vision and Brain-Computer Interface

Object recognition and grasping by the robotic arm are crucial prerequisites for realizing brain-controlled robotic arms. Excellent robotic arm design and efficient algorithms can compensate for the shortcomings of Brain-Computer Interface (BCI) based on motor imagery. This paper utilizes a depth camera as the sensor for environmental perception and employs a six-degree-of-freedom robotic arm with computer vision capabilities to achieve object recognition, pose estimation, and autonomous grasping in multi-target scenarios. The robotic arm is controlled to grasp a recognized object online via BCI based on the analysis of evoked potentials of the user's EEG signals upon presentation of visual stimuli.