Online performance estimated (a) before and (b) after donning/doffing for all metrics, and (c) statistical test results showing the significant difference (gray colored entry, p < 0.05) between the performance obtained before and after donning/doffing at the corresponding combinations of training and test positions.

The error bars in (a) and (b) represent the standard deviations of the performance values of the thirteen normally limbed subjects. No significant performance loss is observed even when test arm positions are different from training positions, compared to the performance obtained from the same training and test positions (a) before donning/doffing. Only two cases show significant performance differences between test arm positions (P1 and P2 training positions for user effort before donning/doffing). When comparing the online performance in terms of the training position, there is only one case showing statistically better performance than the others (P2 > P1 for path efficiency before donning/doffing). A similar trend is also observed (b) after donning/doffing even if overall online performance decreases, compared to before donning/doffing. (c) Statistically significant performance loss is frequently observed between before and after donning/doffing, except one performance metric (stopping path).