LatRec: Recognizing Goals in Latent Space

Recent approaches to goal recognition have progressively relaxed the requirements about the amount of domain knowledge and available observations, yielding accurate and efficient algorithms. These approaches, however, assume that there is a domain expert capable of building complete and correct domain knowledge to successfully recognize an agent’s goal. This is too strong for most real-world applications. LATREC applies modern goal recognition algorithms directly to real-world data (images) by building planning domain knowledge using an unsupervised learning algorithm that generates domain theories from raw images. We demonstrate this approach in an online simulation of simple games, such as the n-puzzle game.