1Anthropology, University of Washington, 2Orthopaedics and Sports Medicine, University of Washington
March 27, 2015 , Archview Ballroom
The kinematic parameters of walking can often reveal important characteristics of an animal’s interaction with their environment. Parameters such as heel strike (HS) and toe-off (TO) are particularly important, because they describe the period of contact with the substrate. Visual inspection of motion data is a benchmark method for determining these parameters, but it is laborious. Automatic algorithms, which typically rely on assessment of landmark velocity, require little input to implement, but have proven unreliable. Our goal was to create a Matlab algorithm that replicated the process that researchers use when they visually assign HS and TO, but that can be automated.
We collected the kinematic data on three groups of people using an eight-camera Qualisys motion capture system with a Kistler force plate. Participants walked at self-selected slow, normal, and fast velocities (30 trials). We used marker data to predict HS and TO and compared these predicted values to the HS and TO values assessed using a force plate. The groups include: 20 shod and unshod women, used to develop the algorithms (DEV); 16 women and 10 men who walked in their normal walking shoes, used as the first validation group (VAL1); and 8 women who walked unshod, used as the second validation group (VAL2).
Our algorithm predicted HS and TO reliably: all r2 > 0.95 and all p < 0.001. Our results are, at this time, limited to humans with normal gait walking on level surfaces, but offer the opportunity to expand this approach to other situations.