Learning and Planning with a Semantic Model
This work addresses the problem of adaptation in visually diverse environments for AI agents, representing an incremental improvement with a novel hybrid method.
The paper tackles the challenge of generalizing deep reinforcement learning agents to unseen man-made environments by proposing LEAPS, a hybrid model-based and model-free approach that uses a semantic model for planning, and it outperforms baselines in visual navigation tasks on House3D.
Building deep reinforcement learning agents that can generalize and adapt to unseen environments remains a fundamental challenge for AI. This paper describes progresses on this challenge in the context of man-made environments, which are visually diverse but contain intrinsic semantic regularities. We propose a hybrid model-based and model-free approach, LEArning and Planning with Semantics (LEAPS), consisting of a multi-target sub-policy that acts on visual inputs, and a Bayesian model over semantic structures. When placed in an unseen environment, the agent plans with the semantic model to make high-level decisions, proposes the next sub-target for the sub-policy to execute, and updates the semantic model based on new observations. We perform experiments in visual navigation tasks using House3D, a 3D environment that contains diverse human-designed indoor scenes with real-world objects. LEAPS outperforms strong baselines that do not explicitly plan using the semantic content.