Lost On Campus: Vision-Language Model Benchmark for Outdoor Navigation

Lost On Campus benchmark evaluates Embodied Scene Representation (ESR) of Vision-Language Models in large-scale real-world outdoor 3D environments reconstructed by 3D Gaussian Splatting. It introduces a unified reasoning-action evaluation framework integrating diagnostic QA and closed-loop interactive navigation under multimodal instructions. The dataset is authored by lost-on-campus-project and was last updated on 2026-05-07.

Use Cases

• Benchmarking Embodied Scene Representation (ESR) based on the described evaluation framework.
• Evaluating active reasoning for navigation based on multimodal instructions.
• Testing diagnostic question-answering capabilities in isolated reasoning scenarios.
• Developing models for interaction in large-scale outdoor 3D environments.

Strengths

• Focuses on a specific benchmark for Vision-Language Model evaluation in embodied AI.
• Integrates both diagnostic QA and interactive navigation within a unified framework.
• Targets large-scale real-world outdoor 3D environments.

Limitations

• Dataset structure and column-level documentation are unknown, requiring manual inspection after download.
• Row count, file formats, and data size are unspecified, limiting suitability assessment.
• Last updated 2026-05-07 08:27:50; freshness should be verified.

Provenance

Source

lost-on-campus-project

Collection Method

Likely constructed for benchmarking Vision-Language Models in reconstructed 3D environments.

Freshness

2026-05-07 08:27:50