Awesome Embodied Robotics and Agent
A curated, actively updated list of research on embodied robotics and agents built with vision-language and large language models, covering scene understanding, navigation, planning and manipulation.
Awesome Embodied Robotics and Agent is a curated list of research on embodied robotics or agents built with vision-language models and large language models. It tracks papers and their code across areas such as scene understanding, navigation, planning algorithms and robotic manipulation, and organizes a survey of efficient vision-language-action models around three pillars: efficient model design, efficient training, and efficient data collection.
Entries include benchmarks such as a spatial-intelligence navigation benchmark with an open dataset and baseline model, lightweight perception-control-learning frameworks, notable conference papers, and surveys on self-evolving large language models.
It is maintained on GitHub, accepts pull requests for new papers, and is updated with dated news entries.