PanoVLN

PanoVLN is a vision-and-language navigation policy that follows instructions from 360ยฐ RGB observations. It combines a Qwen3.5-4B backbone with PanoVGGT geometry features and predicts 18-action sequences. Confidence-guided execution selects how far to move before observing and planning again.

This model is presented in the paper PanoVLN: Towards Effective Panoramic Vision-and-Language Navigation.

For usage, training, evaluation, and deployment instructions, please refer to the GitHub repository.

Downloads last month
82
Safetensors
Model size
6B params
Tensor type
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Paper for wangzhen-w/PanoVLN