Search papers, labs, and topics across Lattice.
This paper introduces SurveilNav, a collaborative navigation framework that integrates active camera scheduling, joint 2D/3D mapping, and value estimation using vision-language models to enhance robot performance in multi-view environments. By leveraging a novel indoor collaborative object navigation dataset with 206 cameras across 74 floors, the framework allows robots to utilize both dynamic local perception and static global surveillance views. Experimental results indicate that SurveilNav significantly improves exploration efficiency and navigation success rates compared to existing methods, highlighting its potential for practical applications in large-scale search and rescue operations.
SurveilNav achieves state-of-the-art navigation success rates by seamlessly integrating robot perception with multi-view surveillance, transforming how robots can navigate complex environments.
With the growing deployment of surveillance systems in factories, offices, and homes, integrating them with robots offers a promising direction for collaborative and efficient task execution. However, existing approaches largely focus on single-robot scenarios and struggle with multi-view collaboration in large-scale environments. In this paper, we present a novel indoor collaborative object navigation dataset built on Habitat-Sim, featuring 206 cameras across 74 floors. The dataset enables systematic evaluation of an agent's ability to exploit multi-view surveillance information. To address the limitations of single-robot perception, we propose SurveilNav, a collaborative navigation framework that integrates active camera scheduling, joint 2D/3D mapping, VLM-based value estimation, and collaborative target verification. By synergizing the robot's dynamic local perception with the static global view of surveillance, this architecture effectively overcomes both the limited perception range of single agents and the inherent blind spots of fixed cameras, resolving inefficient exploration. Experimental results on the HM3D dataset demonstrate that SurveilNav substantially outperforms existing methods, achieving state-of-the-art performance in both exploration efficiency and navigation success rate. Moreover, the system shows strong potential for applications in large-scale search, home environments, and rescue missions.
