BaiShuanghao/Awesome-Robotics-Manipulation

What it solves

This project provides a structured, curated map of the rapidly evolving field of robot manipulation. It helps researchers and developers navigate the vast landscape of academic literature, ranging from traditional control methods to modern learning-based approaches, by organizing them into a clear, logical taxonomy.

How it works

The repository categorizes research papers into a two-stage framework: high-level planning and low-level action modeling. It connects task-level reasoning (such as task planning or geometric constraints) with the specific learning strategies (like imitation or reinforcement learning) used to generate executable robot actions. It also tracks essential resources, including simulators, benchmarks, and datasets.

Who it’s for

It is designed for robotics researchers, students, and engineers who need to stay current with state-of-the-art developments in embodied AI, foundation models for robotics, and various manipulation tasks like dexterous, soft robotic, or humanoid control.

Highlights

• Curates a comprehensive collection of research papers on robot manipulation.
• Features a systematic taxonomy connecting high-level planning with low-level action modeling.
• Includes dedicated sections for simulators, benchmarks, and datasets to support practical research.
• Covers diverse manipulation domains including dexterous, soft robotic, mobile, and humanoid manipulation.
• Continuously updated with new research and community contributions.