AUTHOR=Scholz Fedor , Gumbsch Christian , Otte Sebastian , Butz Martin V. 

TITLE=Inference of affordances and active motor control in simulated agents

JOURNAL=Frontiers in Neurorobotics

VOLUME=Volume 16 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2022.881673

DOI=10.3389/fnbot.2022.881673

ISSN=1662-5218

ABSTRACT=Flexible, goal-directed behavior is a fundamental aspect of human life. Based on the free energy minimization principle, the theory of active inference formalize the generation of such behavior from a computational neuroscience perspective. Based on these theories, we introduce an output-probabilistic, temporally predictive, modular artificial neural network architecture, which processes sensorimotor information, infers behavior-relevant aspects of its world, and invokes highly flexible, goal-directed behavior. We show that our architecture, which is trained end-to-end minimizing an approximation of free energy, develops latent states that can be interpreted as affordance maps. That is, the emerging latent states signal which actions lead to which effects dependent on the local context. In combination with active inference, we show that flexible, goal-directed behavior can be invoked, incorporating the emerging affordance maps. As a result, our simulated agent flexibly steers through continuous spaces, avoids collisions with obstacles, and prefers pathways that lead to the goal with high certainty. Additionally, we show that the learned agent is highly suitable for zero-shot generalization across environments: After training the agent in a handful of fixed environments with obstacles and other terrains affecting it's behavior, it performs similarly well in procedurally generated environments containing different amounts of obstacles and terrains of various sizes at different locations. To improve and focus model learning even more, we plan to apply active inference also while learning the temporally predictive model itself in an information-gain-oriented manner in the near future. Moreover, we intend to foster the development of both deeper event-predictive abstractions and compact, habitual behavioral primitives.