Sequential Human Gait Classification With Distributed Radar Sensor Fusion

This paper presents different information fusion approaches to classify human gait patterns and falls in a radar sensors network. The gaits classified this work are both individual sequential, continuous collected by FMCW three UWB pulse placed at spatial locations. Sequential those containing multiple styles performed one after the other, with natural transitions between, including fall events developing from walking some cases. proposed operate signal decision level. For level combination, simple trilateration algorithm is implemented on range data 3 sensors, achieving good classification results Bi-LSTM (Bidirectional LSTM neural network) as classifier, without exploiting conventional micro-Doppler information. fusion, of radars using network combined robust Naive Bayes Combiner (NBC), showed subsequent improvement compared single case thanks multi-perspective views subjects. Compared SVM Random Forest classifiers, approach yields +20% +17% accuracy for range-only method NBC method, respectively. When classifying sequential gaits, overall two methods reaches 93% 90%, validation via 'leaving participant out' test robustness subjects unknown