On the efficiency of an autonomous cyclic motion grading system

In sport activities and rehabilitation it is of great benefit to provide an autonomous and individual assessment system of motor activity. Providing quantitive data about the movement is not only valid for characterising the courses of training processes, it will also help the athlete to improve the technique or to enhance learning in motor rehabilitation. In this regard, significant features are required to monitor and score the movement-pattern. We analyzed different rowing parameters including grip pull out, grip force, slinding seat position and foot-rest force to derive relevant rowing features. An indoor rower was used to record several movement parameters of elite and novice athletes. Different features were extracted for each rowing parameter by assessing the difference of each feature value on elite and novice athletes. The most relevant features as well as rowing parameters were selected based on the signal-to-noise-ratio ranking. The efficiency of Naive Bayes classifier to differentiate the rowing activity between elite and novice athletes was investigated. According to the results five features are sufficient to achieve a high classification rate of 100%. Using the proposed feature set a grading system was introduced enabling to rank and score the quality related to motor activity in rowing.