Expand Your SCOPE: Semantic Cognition over Potential-Based Exploration for Embodied Visual Navigation

---

Installation

You can set up the conda environment either by following the step-by-step instructions below or by using the provided environment.yml file.

Option 1: Step-by-step installation (Linux, CUDA 11.8, Python 3.9.21)

conda create -n scope python=3.9.21 -y && conda activate scope

conda install pytorch==2.4.1 torchvision==0.19.1 torchaudio==2.4.1  pytorch-cuda=11.8 -c pytorch -c nvidia
conda install -c conda-forge -c aihabitat habitat-sim=0.2.5 headless faiss-cpu=1.7.4 -y
conda install https://anaconda.org/pytorch3d/pytorch3d/0.7.4/download/linux-64/pytorch3d-0.7.4-py39_cu118_pyt201.tar.bz2 -y

pip install omegaconf==2.3.0 open-clip-torch==2.26.1 ultralytics==8.2.31 supervision==0.21.0 opencv-python-headless==4.10.* \
 scikit-learn==1.4 scikit-image==0.22 open3d==0.18.0 hipart==1.0.4 openai==1.35.3 httpx==0.27.2                                                      

Option 2: Using environment.yml

We provide an environment.yml file for easy setup. This file includes all the dependencies (both conda and pip) and their versions.

conda env create -f environment.yml
conda activate scope

Note: The environment.yml file was generated on a Linux system with CUDA 11.8. If you are using a different system, you may need to adjust the CUDA version or the packages accordingly.

Run Evaluation

1 - Preparations

Dataset

Please download the train and val split of HM3D, and specify the path in cfg/eval_goatbench.yaml. For example, if your download path is /your_path/hm3d/ that contains /your_path/hm3d/train/ and /your_path/hm3d/val/, you can set the scene_data_path in the config files as /your_path/hm3d/.

OpenAI API Setup

Please set up the endpoint and API key for the OpenAI API in src/const.py.

2 - Run Evaluation on GOAT-Bench

You can directly run the following script:

python run_goatbench_evaluation.py -cf cfg/eval_goatbench.yaml

The results will be saved and printed after the script finishes.

Acknowledgement

The codebase is built upon 3D-Mem, OpenEQA, Explore-EQA, and ConceptGraph. We thank the authors for their great works.

Citing SCOPE

@misc{wang2025expandscopesemanticcognition,
      title         = {Expand Your SCOPE: Semantic Cognition over Potential-Based Exploration for Embodied Visual Navigation}, 
      author        = {Ningnan Wang and Weihuang Chen and Liming Chen and Haoxuan Ji and Zhongyu Guo and Xuchong Zhang and Hongbin Sun},
      year          = {2025},
      eprint        = {2511.08935},
      archivePrefix = {arXiv},
      primaryClass  = {cs.RO},
      url           = {https://arxiv.org/abs/2511.08935}, 
}