A Deep Reinforcement Learning Approach to Audio-Based Navigation in a Multi-Speaker Environment
This addresses a novel challenge in reinforcement learning for autonomous agents in audio-rich settings, though it is incremental as it applies existing methods to a new sensory modality.
The paper tackles the problem of audio-based navigation in a multi-speaker environment using deep reinforcement learning, achieving successful identification and movement towards a target speaker while avoiding collisions and handling pitch shifting and limited training data.
In this work we use deep reinforcement learning to create an autonomous agent that can navigate in a two-dimensional space using only raw auditory sensory information from the environment, a problem that has received very little attention in the reinforcement learning literature. Our experiments show that the agent can successfully identify a particular target speaker among a set of $N$ predefined speakers in a room and move itself towards that speaker, while avoiding collision with other speakers or going outside the room boundaries. The agent is shown to be robust to speaker pitch shifting and it can learn to navigate the environment, even when a limited number of training utterances are available for each speaker.