Probabilistic Programming Bots in Intuitive Physics Game Play

Recent findings suggest that humans deploy cognitive mechanism of physics simulation engines to simulate the objects. We propose a framework for bots probabilistic programming tools interacting with intuitive environments. The employs in way infer about moves performed by an agent setting governed Newtonian laws motion. However, methods programs can be slow such due their need generate many samples. complement model model-free approach aid sampling procedures becoming more efficient through learning from experience during game playing. present where combining approaches (a convolutional neural network our model) and model-based (probabilistic simulation) is able achieve what neither could alone. This outperforms all or approach. discuss case study showing empirical results performance on Flappy Bird.