ReALFRED: An Embodied Instruction Following Benchmark in Photo-Realistic Environments
This work addresses the gap between simulated training and real-world deployment for embodied AI agents, though it is incremental as it extends an existing benchmark with enhanced realism.
The authors tackled the problem of training robotic agents for household tasks in simulated environments that lack realism, by proposing the ReALFRED benchmark which uses real-world scenes and objects to reduce the visual domain gap. They found that existing methods from the ALFRED benchmark consistently performed worse on ReALFRED, highlighting the need for new approaches in more realistic settings.
Simulated virtual environments have been widely used to learn robotic agents that perform daily household tasks. These environments encourage research progress by far, but often provide limited object interactability, visual appearance different from real-world environments, or relatively smaller environment sizes. This prevents the learned models in the virtual scenes from being readily deployable. To bridge the gap between these learning environments and deploying (i.e., real) environments, we propose the ReALFRED benchmark that employs real-world scenes, objects, and room layouts to learn agents to complete household tasks by understanding free-form language instructions and interacting with objects in large, multi-room and 3D-captured scenes. Specifically, we extend the ALFRED benchmark with updates for larger environmental spaces with smaller visual domain gaps. With ReALFRED, we analyze previously crafted methods for the ALFRED benchmark and observe that they consistently yield lower performance in all metrics, encouraging the community to develop methods in more realistic environments. Our code and data are publicly available.