Coordination analysis of human movements with body sensor networks: A signal processing model to evaluate baseball swings

Becoming proficient in a sport requires significant investment in training. Traditional training approaches such as training with a partner or an expert, and training with the help of videotaping can significantly increase progress. These techniques, however, do not provide fine grain detail about movements of the player, are time consuming, or are limited to specific locations. In contrast, wearable sensor devices can improve training due to the high level of mobility, ubiquity and intelligent feedback offered. In this paper, we present a wearable platform that provides baseball players with corrective feedback based on multidimensional physiological data collected from a body sensor network. We employ a swing model that specifies actions that must be performed properly, in the correct order, and with precise timing between limbs. The system evaluates a baseball swing using motion transcripts. Transcripts simplify interpretation of complex movements and can be used to reduce the size of data that need to be transmitted across the network. Using transcripts, we measure coordination among limb segments and joints of the body. The starting times of key events are found in the transcripts, and the coordination between these times is analyzed. The swing quality is then assessed by comparing the intersegment coordination of a test swing to that of a template swing.