Pipeline of our framework. First, brain commands generate neural signals transmitted to the arm muscles, producing surface electromyography (sEMG) signals captured by sensors. The raw sEMG signals are collected and filtered to improve signal quality. (IMAGE)
Caption
The processed signals are input into a classifier (A) to identify gesture intent, triggering predefined prosthetic hand tool grasping commands. In dynamic tool interaction tasks, rigid impacts can easily cause prosthetic hand grasp instability (B). To address this issue, we developed a biomimetic controller (tactile, kinesthetic, and EMG bionic gripping controller [TKE-BGC]) specifically for high-dynamic scenarios (C). This controller is trained on the basis of EMG, tactile, and joint angle data generated by able-bodied individuals during tool manipulation (D). When sensors detect a risk of grasp instability, TKE-BGC immediately adjusts the joint angles using the amputee’s real-time tactile, joint angle, and EMG signals (E). This stabilizes the grasp and maintains task continuity.
Credit
Bin Fang, School of Artificial Intelligence, Beijing University of Posts and Telecommunications.
Usage Restrictions
News organizations may use or redistribute this image, with proper attribution, as part of news coverage of this paper only.
License
Licensed content