SLAMFormer-$\infty$: Infinite SLAM Transformer for Unbounded Frontend and Backend Processing

2026-08-04

AI Summary

This work addresses the limitations of conventional SLAM systems in unbounded, ultra-long trajectories—specifically, inefficient front-end processing and insufficient global consistency caused by fixed coordinate frames and explicit distance constraints. To overcome these challenges, we propose Memory-Conditioned Geometric Transformer (MCGT), the first end-to-end unbounded SLAM architecture. MCGT dynamically defines coordinate frames and scale, eliminating reliance on anchoring to the first frame or hard-coded distance bounds, and enables joint learning of local front-end feature extraction and global back-end optimization. Evaluated on large-scale datasets, our system efficiently handles trajectories exceeding 17 kilometers and achieves state-of-the-art performance in both trajectory estimation and scene reconstruction, demonstrating exceptional scalability, robustness, and generalization.

Abstract

We introduce the Infinite SLAM Transformer (SLAMFormer-$\infty$), the first geometric transformer capable of supporting both long-range frontend and backend processing without an explicit distance bound. Instead of relying on a first-frame-anchored formulation, SLAMFormer-$\infty$ employs memory conditions to define flexible coordinate systems and scales for input frames, enabling more expressive structural conditioning. Built upon this formulation, the frontend preserves efficient local computation, while the backend jointly optimizes long-range trajectories and scene geometry in a globally consistent manner. Experimental results demonstrate that SLAMFormer-$\infty$ achieves superior or highly competitive performance in both trajectory estimation and scene reconstruction across large-scale datasets. Notably, SLAMFormer-$\infty$ generalizes to extremely long trajectories, successfully operating on sequences exceeding $17\mathrm{km}$.

Problem

• SLAM
• unbounded processing
• long-range trajectory
• scene reconstruction
• coordinate system

Innovation

• SLAMFormer
• geometric transformer
• unbounded SLAM
• memory-conditioned coordinate system
• long-range trajectory optimization