Real-time tracking of game assets in american football for automated camera selection and motion capture

Abstract

In fast-paced games like American football, the action (and the viewers’ interest) often revolves around the location of key game assets, such as the football. It is difficult to track the football because it is often obscured or moving rapidly, essentially meaning that the broadcast-camera operators must manually track the football to provide feeds of the most interest. In this paper we describe the design and implementation of sensor-system that uses received signal strength data from multiple strategically placed sensor nodes to localize the game asset to an accurate camera region within a latency of 100ms. Using this real-time localization strategy, we can automate the control of broadcast cameras even in a fast-paced game with dramatically gamechanging moments (such as a kickoff return for a touchdown). This paper describes the design of the sensors, the network protocol, and the architecture of a deployable system in the field of play.