The accurate identification of initial and final foot contacts is a crucial prerequisite for obtaining a reliable estimation of spatio-temporal parameters of gait. Well-accepted gold standard techniques in this field are force platforms and instrumented walkways, which provide a direct measure of the foot–ground reaction forces. Nonetheless, these tools are expensive, non-portable and restrict the analysis to laboratory settings. Instrumented insoles with a reduced number of pressure sensing elements might overcome these limitations, but a suitable method for gait events identification has not been adopted yet. The aim of this paper was to present and validate a method aiming at filling such void, as applied to a system including two insoles with 16 pressure sensing elements (element area = 310 mm²), sampling at 100 Hz. Gait events were identified exploiting the sensor redundancy and a cluster-based strategy. The method was tested in the laboratory against force platforms on nine healthy subjects for a total of 801 initial and final contacts. Initial and final contacts were detected with low average errors of (about 20 ms and 10 ms, respectively). Similarly, the errors in estimating stance duration and step duration averaged 20 ms and <10 ms, respectively. By selecting appropriate thresholds, the method may be easily applied to other pressure insoles featuring similar requirements.

准确识别初始和最终足部接触是获得可靠估计步态时空参数的关键先决条件。该领域广为接受的黄金标准技术是力平台和仪器化走道，它们提供了对脚-地面反作用力的直接测量。尽管如此，这些工具价格昂贵、不便携，并且将分析限制在实验室环境中。减少压力传感元件数量的仪表鞋垫可能会克服这些限制，但尚未采用合适的步态事件识别方法。本文的目的是提出并验证一种旨在填充此类空隙的方法，该方法应用于包括两个鞋垫和 16 个压力传感元件（元件面积 = 310 mm ²)，采样频率为 100 Hz。利用传感器冗余和基于集群的策略识别步态事件。该方法在实验室中针对 9 名健康受试者的力平台进行了测试，总共进行了 801 次初始和最终接触。检测到初始和最终接触的平均误差较低（分别约为 20 毫秒和 10 毫秒）。同样，估计站立持续时间和步长持续时间的平均误差分别为 20 ms 和 <10 ms。通过选择合适的阈值，该方法可以很容易地应用于具有类似要求的其他压力鞋垫。