Event-Based Camera Simulation Wrapper for Arcade Learning Environment

Charles P. Rizzo, Catherine Schuman, James Plank

Abstract: Event-based cameras are cameras with high dynamic range that measure changes in light intensity at each pixel instead of capturing frames like traditional cameras. There are several event-based camera simulation software libraries that can convert videos or collections of frames to a stream of simulated camera events. To the authors’ knowledge, with the exception of vehicle control projects, there are no software libraries that simulate event-based camera activity online during an agent’s training phase on other various control applications. This work introduces ALE_EBC, a software wrapper around the Arcade Learning Environment that converts game frames to simulated event-based camera event streams, allowing agents to be trained on a variety of control-based Atari games through the lens of a neuromorphic camera.