README.md

metadata

license: mit
tags:
  - robotics
  - papers
  - embeddings
  - lancedb
  - semantic-search
size_categories:
  - 10K<n<100K

Robotics Papers Vector Database

Semantic search over 63,381 academic papers from 30 conference-year combinations in robotics, CV, and ML.

Contents

• Embeddings: BAAI/bge-m3 (1024 dimensions) via SiliconFlow
• Conferences: CoRL, CVPR, ECCV, ICCV, ICLR, ICML, ICRA, IROS, NeurIPS, RSS, WACV (2023-2026)
• Fields: title, abstract, author, conference, year, arxiv, github, citations, keywords
• Format: LanceDB (compacted, single fragment)

Usage

Remote query (no download needed)

import lancedb

db = lancedb.connect("hf://datasets/Litian2002/robotics-papers-vecdb/lancedb")
table = db.open_table("papers")
print(f"Papers: {table.count_rows()}")

# Semantic search (requires embedding your query first)
# See the companion repo for the full search pipeline

Local usage

git clone https://huggingface.co/datasets/Litian2002/robotics-papers-vecdb
# or use huggingface_hub to download

With vec-db CLI

# Clone the tool repo
git clone <vec-db-repo-url>
cd vec-db
VECDB_LANCE_DIR=/path/to/downloaded/lancedb npx tsx src/cli.ts search "robot grasping"

Schema

Column	Type	Description
vecId	string	{conf}_{year}_{id} compound key
title	string	Paper title
abstract	string	Paper abstract
author	string	Authors (semicolon-separated)
conference	string	Conference name
year	float	Publication year
arxiv	string	arXiv ID
github	string	GitHub repo URL
gsCitation	float	Citation count (from OpenAlex)
vector	float[1024]	bge-m3 embedding