In the current study, three methods to detect turn switches in alpine ski racing based on wearable systems were assessed. Accuracy and precision for method A were within time resolution of the reference system (20 ms) and precision was about twice as large for methods B and C (table 1). The observed differences in accuracy and precision could be explained by functionally different definitions of turn switches for methods B (inflection point in LJC line) and C (left and right vertical distance crossing), while method A used the same turn switch definition as the reference method (Supej et al., 2003). Another noteworthy finding is the fact that, despite the absence of a full-body inertial sensor setup and additional snow surface measurements, method B was able to provide similar accuracy and precision as method C. These results indicate that the inflection point in the LJC line happens at the same time as the crossing point of the left-right vertical distance. However, since the CoM line is not identical with the LJC line, the turn switch is detected at a slightly different time. A factor which might add a time difference for the turn switches detected with methods B and C compared to the definition from Supej et al. (2003) could be an oblique skiing direction (e.g., during a traverse where gates are not aligned along the slope`s fall line). In this case the ankle positions cannot be at the same height during a turn switch (asymmetric leg positions) and the time of equal distance between the left and right ankles and CoM may no longer correspond to the true moment of turn switch. A previous study by SpOrri, Krdll, Schwameder, & Muller (2012) reported section time differences between the fastest and slowest runs ranging from 20 ms to 60 ms. Thus, if a section started and stopped at a turn switch, the proposed methods might not provide sufficient accuracy and precision to reliably detect such performance differences. Moreover, precision might also be a limiting factor for analyses of turn cycle structure or timing aspects (e.g., the relative time between turn start and gate crossing). In cases where the highest precision is needed, method A should be used, since it provided similar performance as the reference system. However, if simpler setups are required, then methods B or C could be used with only a slight loss of performance. For method B, since absolute position information is not needed, the differential GNSS could also be replaced with a low-cost GNSS, further reducing the setup complexity significantly. The precision of method B could be further increased by fixing the GNSS antenna closer to the LJC instead of fixing it to the athlete`s head. To conclude, turn switches can be detected with wearable systems only. The simplest setup was based on GNSS and one inertial sensor and did not require an estimation of the CoM; nevertheless, its accuracy and precision were similar to the complex setups based on differential GNSS and multiple inertial sensors.
© Copyright 2018 Science and Skiing VII. Julkaistu Tekijä Meyer & Meyer. Kaikki oikeudet pidätetään. / Kääntänyt https://www.DeepL.com/Translator