A fuzzy logic system tuned with particle swarm optimization for gait segmentation using insole measured ground reaction force

To segment gait phases of human locomotion tasks, such as level walking and stairway walking, ground reaction force (GRF) data during walking are collected. This paper presents a fuzzy logic inference (FLI) technique tuned by particle swarm optimization (PSO), which is used to find optimal triangular membership functions. To discriminate flexion and extension of swing leg, ground reaction force of the other foot is supplemented. As a result, a whole gait cycle is divided into five phases: heel strike(H-S), stance(S), heel off(H-O), swing one(Sw1), swing two(Sw2). The proposed method is simple and particularly suitable for human gait phases segmentation. Six healthy human were tested while maintaining a constant average walking velocity as far as possible. Results show that swing phase period accounts for around 31%, 41% and 42% in three gait motion patterns separately, which is nearly 40% in clinical gait analysis. After comparison, the proposed method is superior to conventional FLI(C-FLI) system. These results suggest that the proposed method can function as an efficient detector for gait phases.