Sensor Fusion for Predictive Control of Human-Prosthesis-Environment Dynamics in Assistive Walking

Powered prostheses are effective for helping amputees, who lose their legs, towalk on level ground, but these devices are inconvenient to use in complex en-vironments, such as stair ascent/descent and ramp ascent/descent, and can causeproblems. In order to resolve this issue, prostheses need to understand the environ-ments and the motion intent of amputees (the wearer of the prosthesis—the user).To realize these objectives, the present thesis develops a sensor fusion system in acomplete human-prosthesis-environment loop to recognize the environments, pre-dict the motion intent of different users, and control the motion of the prosthesis.It is verified that the proposed and developed sensor fusion system and algorithms,when implemented in a prosthesis, are able to predict the motion intent of differentamputees accurately (at an accuracy≈97%) and efficiently (computing time<30ms). The proposed method will increase the intelligence of the wearable robots,improve the human-robot interaction, and assist people in need to convenientlywalk in complex environments.