OpenSocInt: A Multi-modal Training Environment for Human-Aware Social Navigation
🤖 AI Summary
This work addresses the limited multimodal perception and human-centric awareness of autonomous agents operating in human-dense environments by proposing and open-sourcing OpenSocInt—a modular, extensible simulation platform for multimodal social interaction. OpenSocInt is the first framework to unify multimodal perception (including vision, audio, and social signals), feature fusion mechanisms, and training of social navigation policies within a single architecture, supporting learning paradigms such as reinforcement learning and behavioral cloning. Experimental results demonstrate that OpenSocInt flexibly accommodates diverse perception modalities and agent architectures, significantly enhancing agents’ navigation and interaction capabilities in complex social scenarios.
Application Category
📝 Abstract
In this paper, we introduce OpenSocInt, an open-source software package providing a simulator for multi-modal social interactions and a modular architecture to train social agents. We described the software package and showcased its interest via an experimental protocol based on the task of social navigation. Our framework allows for exploring the use of different perceptual features, their encoding and fusion, as well as the use of different agents. The software is already publicly available under GPL at https://gitlab.inria.fr/robotlearn/OpenSocInt/.
Problem
Research questions and friction points this paper is trying to address.
social navigation
multi-modal interaction
human-aware agents
training environment
social intelligence
Innovation
Methods, ideas, or system contributions that make the work stand out.
multi-modal
social navigation
modular architecture
human-aware agents
simulation environment
🔎 Similar Papers
No similar papers found.
