Real-time detection of step direction based on plantar pressure pattern during walk

摘要

Walking is an essential activity for physical fitness. Alongside usual forward walk “FW”, the backward walk “BW” is also proved useful for lower limb rehabilitation. However, current real-time walk detection systems have been evaluated on FW only. This paper introduces a technique for bidirectional (forward and backward) distinction in real-time during communal walk. The proposed technique utilises an algorithm to monitor pressure loading pattern on a walking foot while transiting from swing into stance phase. The technique is validated under normal walk styles at slow, normal, and fast speeds in forward and backward directions by 16 healthy walkers. Overall the detection accuracy is found 100% for both directions at all speeds with all the participants. The technique is promising to upgrade existing gait detection systems to accurately detect and count the steps in either direction during a communal or fitness walk including forward and backward steps.