Dog agility is an increasingly popular sport in the United States and other countries. Injuries occur in up to 40% of agility dogs. Agility handlers and trainers need a reliable, accurate way to measure and monitor athlete activity. Advances in human- and canine-wearable sensors provide new ways of accurately recognizing and measuring movement, and machine learning techniques can find patterns in movement data to perform activity recognition. For this project, researchers hypothesize that wearable sensor technology paired with machine learning can measure discrete canine agility movements and can automatically recognize and quantify the activities that occur during agility. The goal is to develop a canine-wearable system to recognize agility activity to lay the groundwork for future prospective, hypothesis-driven studies. Future research will measure canine training workload, speed, fatigue, and other parameters to optimize performance, evaluate the relation between training parameters and injury risk, and study injury prevention strategies.