Home Run: Finding Your Way Home by Imagining Trajectories
This addresses the limitation of existing hierarchical active inference models that only allow planning on previously explored paths, offering a potential improvement for navigation in AI systems, though it is incremental as it builds on prior work.
The paper tackled the problem of enabling agents to plan novel, unexplored paths in navigation tasks, inspired by mice behavior, and demonstrated a proof of concept in a grid-world environment where the agent accurately predicted a new, shorter path to its starting point using a generative model.
When studying unconstrained behaviour and allowing mice to leave their cage to navigate a complex labyrinth, the mice exhibit foraging behaviour in the labyrinth searching for rewards, returning to their home cage now and then, e.g. to drink. Surprisingly, when executing such a ``home run'', the mice do not follow the exact reverse path, in fact, the entry path and home path have very little overlap. Recent work proposed a hierarchical active inference model for navigation, where the low level model makes inferences about hidden states and poses that explain sensory inputs, whereas the high level model makes inferences about moving between locations, effectively building a map of the environment. However, using this ``map'' for planning, only allows the agent to find trajectories that it previously explored, far from the observed mice's behaviour. In this paper, we explore ways of incorporating before-unvisited paths in the planning algorithm, by using the low level generative model to imagine potential, yet undiscovered paths. We demonstrate a proof of concept in a grid-world environment, showing how an agent can accurately predict a new, shorter path in the map leading to its starting point, using a generative model learnt from pixel-based observations.