Active Inference in Robotics and Artificial Agents: Survey and Challenges
It provides a review and discussion of challenges for researchers in robotics and AI interested in biologically plausible, unified approaches to goal-driven behavior.
The paper surveys the active inference framework, originally from computational neuroscience, as applied to robotics and artificial agents for state-estimation, control, planning, and learning under uncertainty, highlighting its potential in adaptation, generalization, and robustness.
Active inference is a mathematical framework which originated in computational neuroscience as a theory of how the brain implements action, perception and learning. Recently, it has been shown to be a promising approach to the problems of state-estimation and control under uncertainty, as well as a foundation for the construction of goal-driven behaviours in robotics and artificial agents in general. Here, we review the state-of-the-art theory and implementations of active inference for state-estimation, control, planning and learning; describing current achievements with a particular focus on robotics. We showcase relevant experiments that illustrate its potential in terms of adaptation, generalization and robustness. Furthermore, we connect this approach with other frameworks and discuss its expected benefits and challenges: a unified framework with functional biological plausibility using variational Bayesian inference.