A Multi-Sensor Approach for Multi-Joint Tracking

This paper deals with the development of a low-cost multi-sensor and multi-node system aimed to reconstruct the orientation of a multi-joint model compatible with several application fields, such as robotics, human rehabilitation or monitoring systems. As an example, human kinetic parameters can be harnessed to prevent and diagnose diseases, enhance physical performance, and expedite rehabilitation. The approach adopted exploits features of an embedded multi-sensor system and signal processing aimed at the extraction of quantities of interest for the reconstruction of the model pose. In particular, the main contribution of this work is a dedicated signal processing implemented to reconstruct time series of joint angles and to provide a real-time 3D representation of the model. The assessment phase, consisting of a series of angles configurations imposed to the model, proved that mean and standard deviation of the residuals (as the absolute value of the difference between nominal and reconstructed values) are, in the worst case, 1.23° and 0.98° respectively, proving the capability of the algorithm to appropriately reconstruct the pose of the structure in real time.