Research surveys and paper notes on embodied intelligence and robotics. The English collection begins with vision-language navigation (VLN), from task definitions to real-robot evaluation.
Vision-language navigation
VLN Papers: Instruction Following
55 detailed readings on instruction-following VLN, with benchmark leaderboards, technical comparisons, methods, experiments, and limitations.
Vision-Language Navigation: A Survey
A survey of VLN tasks, state and action interfaces, training signals, and closed-loop evaluation, comparing streaming policies, fast-slow control, spatial me...
Suggested reading path
- Start with tasks and system interfaces to understand what a navigation agent observes and controls.
- Compare spatial memory, hierarchical control, and training methods.
- Choose datasets and read the evaluation guide before comparing results.
Companion paper readings
The instruction-following collection includes 55 detailed paper readings, benchmark leaderboards, and technical comparisons in English. The goal-navigation and extensions collection is currently available in Chinese.
- VLN Papers: Instruction Following
- VLN Papers: Goal Navigation and Extensions (Chinese only)
- Full research collection (Chinese only)
About the author
I am Tingde Liu, an AI engineer at Tyndall Labs. I maintain these notes to make research on embodied intelligence, navigation, and robot learning easier to study and apply.
For corrections or suggestions, use the feedback links at the end of each article or visit the site repository.