Event
FAME Multi-Anchor Exploration Method Selected as NeurIPS 2026 Spotlight
3 reports · 3 sources
A team from Tianjin University and Hefei University of Technology proposed FAME (Flow-driven Anchor-based Multimodal Exploration), which reformulates long-horizon goal prediction in open-environment active 3D mapping as using Conditional Flow Matching to simultaneously generate multiple exploration anchors, replacing traditional single-point goal prediction. The paper has been accepted to NeurIPS 2026 and selected as a Spotlight. The team states that information about unseen environments is limited, and methods based on closed-set assumptions (assuming the test environment is similar to the training environment) struggle to generalize, hence this reformulation. FAME uses an A*-based path planner to generate candidate paths for each anchor according to the current obstacle map, then determines routes through path clustering and hierarchical selection. The research team states that after training on AiMDoom, the model was transferred directly to MP3D for testing without any fine-tuning.
Report timeline
- Tianjin University proposes FAME: using multi-anchor exploration to help robots plan routes in unknown environments
A team from Tianjin University and Hefei University of Technology proposes FAME, which uses Conditional Flow Matching to generate multiple exploration anchors simultaneously, then determines exploration routes through A* path planning, path clustering, and hierarchical selection. The paper was accepted to NeurIPS 2026 and selected as a Spotlight.
CVer · 微信公众号 ·
- Tianjin University team proposes FAME: an active 3D mapping method using Flow Matching to generate multiple exploration goals, selected for NeurIPS 2026 Spotlight
A team from Tianjin University and Hefei University of Technology proposes FAME, which uses Conditional Flow Matching to simultaneously generate multiple exploration anchors, then determines routes through A* path planning, path clustering, and hierarchical selection. The paper was accepted to NeurIPS 2026 and selected for Spotlight.
机器之心 · 微信公众号 ·
- Flow-driven multi-anchor exploration method improves generalization of active 3D mapping in open environments
The paper proposes using Conditional Flow Matching to reformulate long-horizon goal prediction in open-environment active 3D mapping as conditional multimodal anchor generation, replacing traditional single-point prediction.
arXiv cs.RO ·