Swarm-ReID: Decentralized Self-Adaptive Gallery Construction for Multi-Robot Open-World Person Re-Identification
Marios Kaplanis, Miquel Kegeleirs, Lorenzo Garattoni, Mauro Birattari, Gianpiero Francesca
AI summary
Problem
Traditional person re-identification relies on static, closed-world galleries and single-robot setups, making it unsuitable for open-world robotics where new identities continuously appear and scalability across multiple robots is required.
Approach
Swarm-ReID is an unsupervised, decentralized method where each robot extracts multimodal features, maintains a self-adaptive local gallery, and exchanges data with peers using specific communication protocols and informed exploration behaviors to collaboratively recognize and track people.
Key results
- Consistently outperforms the baseline swarm perception method across all tested scenarios (CMC@1 up to 0.859 vs 0.644)
- Demonstrates that decentralized communication and informed robot movement strategies significantly boost recognition accuracy
- Validates the approach through extensive simulations varying environments, swarm sizes, and communication ranges, plus a real-world physical robot demonstration
- Introduces a gallery optimization module that maintains compact, diverse identity models by filtering redundant or low-quality samples
Why it matters
Provides a scalable, robust framework for multi-robot open-world perception, enabling practical human-robot interaction and collaborative surveillance in dynamic environments.
Abstract
Swarm perception enables a robot swarm to col- lectively sense and understand the environment by integrating sensory inputs from individual robots. We explore its applica- tion to person re-identification (re-id), the task of recognizing previously observed individuals. Traditional re-id systems rely on static offline galleries, which restricts their use in open-world scenarios where new identities appear over time. In robotics, most methods address single-robot re-id in person-following tasks, limiting scalability to multi-person settings, while swarm perception studies largely overlook the role of re-id algorithms. To address these gaps, we propose Swarm-ReID, an unsupervised method for decentralized swarm re-identification. Our method introduces mechanisms for robot-to-robot communication and informed movement strategies, enabling the swarm to collabo- ratively construct adaptive galleries online without centralized control. Simulations across diverse environments, number of people, swarm sizes, communication protocols, and exploration behaviors show that Swarm-ReID consistently outperforms existing swarm perception methods. Our results highlight how communi- cation and informed movement improve recognition performance, establishing Swarm-ReID as a state-of-the-art method for open- world multi-robot person re-identification.