Sonicverse: A Multisensory Simulation Platform for Embodied Household Agents that See and Hear
This addresses the limitation of deaf agents in silent environments for researchers in embodied AI, though it is incremental by adding audio to existing visual simulators.
The authors tackled the problem of training embodied household agents with multisensory perception by introducing Sonicverse, a simulation platform that integrates realistic audio-visual rendering in real-time, enabling tasks like semantic audio-visual navigation and achieving state-of-the-art performance with sim-to-real transfer to real-world environments.
Developing embodied agents in simulation has been a key research topic in recent years. Exciting new tasks, algorithms, and benchmarks have been developed in various simulators. However, most of them assume deaf agents in silent environments, while we humans perceive the world with multiple senses. We introduce Sonicverse, a multisensory simulation platform with integrated audio-visual simulation for training household agents that can both see and hear. Sonicverse models realistic continuous audio rendering in 3D environments in real-time. Together with a new audio-visual VR interface that allows humans to interact with agents with audio, Sonicverse enables a series of embodied AI tasks that need audio-visual perception. For semantic audio-visual navigation in particular, we also propose a new multi-task learning model that achieves state-of-the-art performance. In addition, we demonstrate Sonicverse's realism via sim-to-real transfer, which has not been achieved by other simulators: an agent trained in Sonicverse can successfully perform audio-visual navigation in real-world environments. Sonicverse is available at: https://github.com/StanfordVL/Sonicverse.