Sensor fusion in upper limb area networks: A survey

Body sensor networks (BSNs) have been increasingly used in medical applications such as exoskeleton control, powered prosthesis control, tremor suppression, gesture and sign language recognition systems, and human computer interfaces. This review explores the use of multi-modal sensor fusion in BSNs for the detection, measurement and classification of upper limb for the control of dynamic systems. Specifically, the review will look into the most common multi-modal sensor combinations found in literature, namely inertial measurement units (IMUs) with electromyography (EMG), IMUs with camera systems, EMG with electroencephalography (EEG), and IMUs with flexible force sensors. The advantages and challenges associated with these sensor combinations is discussed, as well as the challenges of sensor fusion in a broad nature, with particular focus on the use of data, feature, or decision level fusion.