Analysis of Front Leg Kinematics of Cricket Bowler Using Wearable Sensors and Machine Learning

Abstract

A fast bowler in the game of cricket is exposed to high-intensity actions and is more prone to lumbar stress fractures, ankle sprains, and knee, shoulder, and lower back injuries. There are different phases of bowling, from running to ball release, viz., run-up, predelivery stride (PDS), mid-bound (MB), front foot contact (FFC), ball release, and follow through. The front leg motion of the bowler plays a crucial role in delivering the ball efficiently. An improper landing after the jump causes most of the injuries as more strain is induced on the knee, ankle, and groin region. Loss of balance during the follow-through is the other most common reason for injury. A braced front leg at FFC helps the bowler transfer the momentum gained during the run-up into a fast-bowling action. For a refined understanding of the motion of the bowler and the causes of injuries, the identification of bowling phases is crucial. This article proposes the design of a miniature and economical wearable sensor unit using a Bosch BNO055 inertial measurement unit (IMU) sensor that the players can comfortably wear during the game. A novel method for calculating knee flexion of bowlers at FFC using quaternion data gathered from sensors is presented. Two wearable units were placed on the bowler’s front leg to collect motion data. Supervised machine-learning algorithms have been evaluated for classifying bowling phases using statistical features extracted from the sensor data to assist coaches and players with data-driven and evidence-based bowlers’ training.