A Firehose of Robot Papers, Organized by Acronym

What it does This repository is a curated index of 700-plus papers and projects at the intersection of computer vision, language models, and robotic control. It tracks the embodied-AI landscape—Vision-Language-Action (VLA) models, diffusion policies, sim-to-real transfer, navigation benchmarks, and training simulators—by linking directly to arXiv preprints, project pages, and code. It is essentially a reading list for a field that is currently publishing faster than most people can read.

The interesting bit The maintainer’s bet is that robotics is heading for its own “LLM moment,” and the list functions as a real-time chronicle of that buildup. Instead of a flat link dump, it partitions the research into concrete categories—physics-aware policies, world-action models, MLLM-based reasoning—so you can see exactly where a paper sits in the larger stack.

Key highlights

• Covers 2024 through 2026 research, including work from NVIDIA, Physical Intelligence, Ant Group, and Qwen.
• Sections span foundational VLA and World-Action Models, Vision-Language Navigation, diffusion policies, and sim-to-real transfer.
• Most entries link to both the arXiv paper and, where available, an official project page or GitHub repository.
• Papers are grouped by year in reverse chronological order, with particularly influential works pinned to the top regardless of date.

Verdict Worth bookmarking if you are a robotics researcher or engineer trying to stay current without manually trawling arXiv. Skip it if you are looking for a unified code framework or turnkey model weights—this is strictly a reading list.