SPARSE: A Simulation Platform for Autonomous Research and System Evaluation

State of the art research in AI for the autonomous systems space requires the collection of large, annotated datasets of real-world scenarios, especially for edge case events. Take for example, autonomous driving, an edge case might be that while driving, the system encounters a situation in which boxes have suddenly fallen out of the bed of a truck, requiring quick action by the autonomous system. It is difficult to collect enough data for these edge cases in the real world, since by definition, they are edge cases. The Unreal Game Engine provides the perfect environment to simulate realistic driving situations in real-time with minimal costs and with no safety risks. In a simulated environment edge cases can be easily tested repeatedly, accruing the large datasets needed to train an autonomous system.