Implicit Virtual Leader: Decentralized Vision-Only Relative Pose Estimation for Multi-Robot Formations

2026-07-17

AI Summary

This work addresses the limitations of traditional leader-follower formation control, which is vulnerable to single-point failures and error propagation and relies on absolute positioning, rendering it impractical in GPS-denied environments. The authors propose a fully decentralized visual relative pose estimation framework that leverages graph neural networks to fuse monocular images with inter-robot communication, implicitly constructing a virtual reference frame—termed an implicit virtual leader—that is decoupled from any physical entity. This approach eliminates dependence on a physical leader or global localization, supports heterogeneous platforms and variable formation sizes, and incorporates heteroscedastic Gaussian negative log-likelihood loss with Monte Carlo Dropout to model both aleatoric and epistemic uncertainties. Experiments demonstrate that the method achieves high-precision pose estimation and strong generalization capabilities in both simulated and real-world scenarios.

Abstract

Classical leader-follower formation control suffers from single points of failure and error propagation, and relies on absolute localization sensors that are ill-suited for GPS-denied environments. We address these limitations by introducing a fully decentralized, vision-only relative pose estimation framework based on Graph Neural Networks (GNNs). The key idea is the implicit virtual leader (IVL): a non-physical formation reference frame that is not tied to any individual robot but is implicitly learned within the GNN using only monocular images and inter-robot communication. We attach a heteroscedastic GNLL head for aleatoric uncertainty and MC~Dropout for epistemic uncertainty, and conduct a systematic comparison across simulation and real-world test sets. Our framework achieves competitive pose estimation accuracy and generalizes naturally to heterogeneous robot platforms and varying formation sizes.

Problem

• multi-robot formations
• relative pose estimation
• decentralized control
• GPS-denied environments
• single point of failure

Innovation

• Implicit Virtual Leader
• Graph Neural Networks
• Vision-only Pose Estimation
• Decentralized Multi-Robot Systems
• Uncertainty Quantification