Fig. 1

Experimental protocol for transfemoral amputee subjects. The experiment was divided into an undisturbed calibration phase, followed by blocks containing perturbations. In each block, trials of 10 s containing a single perturbation were separated by at least 60 s of walking. A total of 16 trials (4 points in swing phase * 2 repetitions * 2 legs) were recorded per block, in random order. Resting breaks were mandatory between data collection blocks, and could occur throughout the experiment at subjects’ request