BEACON: Language-Conditioned Navigation Affordance Prediction under Occlusion

Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• Existing vision-language spatial grounding methods struggle to infer target locations in occluded regions.limitation 95%
• BEACON predicts an ego-centric Bird's-Eye View (BEV) affordance heatmap over a bounded local region including occluded areas.method 95%
• BEACON injects spatial cues into a Vision-Language Model (VLM) and fuses its output with depth-derived BEV features.method 95%
• An occlusion-aware dataset was built in the Habitat simulator to validate BEACON.method 90%
• BEACON improves accuracy by 22.74 percentage points over the state-of-the-art image-space baseline on the validation subset with occluded target locations.result 95%
• The proposed method validates both its BEV space formulation and the design choices of each module through experimental analysis.method 90%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card