Multisensor data fusion for prosthetic control

A fully neurally integrated high degree of freedom prosthetic limb system is the goal of the Revolutionizing Prosthetics program, a large-scale Defense Advanced Research Projects Agency (DARPA) project. Multi-Sensor Data Fusion (MSDF) for prosthetic control is one of the research areas. This paper proposes a novel approach of applying MSDF technology to prosthetics. The paper proposes a prosthetic data fusion framework, analyzes characteristics of MSDF techniques, and suggests a new algorithm for motile fusion. Specifically, this study compares the fidelity of single-classifier decisions with multi-classifier decision fusions on simulated isometric bicep contractions. It reveals some interesting findings and insights and provides a better understanding of MSDF applications in the medical field. The results confirm the envisioned potentials of applying MSDF for prosthetic control.