Punch Detection During Boxing

This is an undergraduate student’s project (by Simon-Peter Biggs) supervised by Dr. Stefan. As the wearable technology market continues to grow the demand for new products health and fitness increases. People are becoming more interested in fitness gadgets that allow them to track progress over time. Boxing is a popular sport and current fitness trend, however as a combat sport it is associated with injuries. These injuries are most likely to occur in the upper body extremities.

An Internet of Things boxing system was developed using the following sub systems; sensor data acquisition, sensor data transfer, sensor data exchange, data storage, data analysis and data visualisation. The systems purpose was to detect and collect data from boxing movements and classify the data. The data could then be used to design future boxing devices.

The system consisted of a senor connected to a microntroller, which communicated to a custom mobile phone application by Bluetooth Low Energy. The data collected was used to classify boxing movements. The data collected from the application was analysed using a Support Vector Machine algorithm. The data was classified to determine the difference between two movement; Cross and Hook. The classification algorithm was successful producing a score of 87.5%.

A successful prototype had been designed but further iterations were needed to improve the system and make it ready for production.