Update README.md

README.md

@@ -11,7 +11,7 @@ tags:
   - dual-arm
 pretty_name: AgiBot World 2026
 size_categories:
-  - 1K<n<10K
+  - "1K<n<10K"
 language:
   - en
 ---
@@ -26,8 +26,15 @@ language:
 [![LeRobot](https://img.shields.io/badge/LeRobot-v2.1-blue)](https://github.com/huggingface/lerobot)
 [![Homepage](https://img.shields.io/badge/Homepage-agibot--world.com-green)](https://agibot-world.com)
 
+<br/>
+
+<img src="./assets/teaser.gif" alt="AgiBot World 2026 dataset preview (animated)" style="width: 100%; max-width: 100%; height: auto; display: block; margin: 0 auto;" />
+
+<br/>
+
 </div>
-<img src="./assets/teaser.gif" alt="AgiBot World 2026 — real-world embodied intelligence preview" style="width: 100%; max-width: 100%; height: auto; display: block; margin: 0 auto;" />---
+
+## Overview
 
 As robotics research advances into real-world scenarios, the demand for authentic, high-quality data has become increasingly urgent. Following AGIBOT WORLD's "ImageNet moment," we now release the **AGIBOT WORLD 2026** dataset. Built upon massive real-world scenes, it systematically spans pivotal research directions in embodied intelligence, designed to power the next generation of embodied agents.
 
@@ -35,7 +42,7 @@ The AGIBOT WORLD 2026 dataset is collected from **100% real-world environments**
 
 We invite researchers worldwide to leverage AGIBOT WORLD 2026 to drive robotic intelligence from the lab into the real world, empowering every industry and tangibly boosting production and service efficiency.
 
-
+---
 
 ## Get Started
 
@@ -342,8 +349,8 @@ Common built-in frame type names: `Error Frame`, `Success Frame`, `Intervention
 
 ## Episode Data and Video Data
 
-| | Episode Data | Video Data |
-|---|---|---|
+| Aspect | Episode data | Video data |
+|--------|--------------|------------|
 | **Location** | `data/` | `videos/` |
 | **Format** | Apache Parquet | MP4 |
 | **Granularity** | One file per episode | One file per camera per episode |
@@ -395,7 +402,7 @@ A companion script `split_episode.py` is provided to convert the dataset into st
 
 Run `split_episode.py` to split long-horizon episodes into single-instruction episodes compatible with standard LeRobot training scripts (e.g., `lerobot/scripts/train.py`). The extended annotation fields are stripped from `info.json` in the output.
 
-Tool Script: [split_episodes_tool.zip](split_episodes_tool.zip)
+**Tool:** [split_episodes_tool.zip](split_episodes_tool.zip)
 
 **Prerequisites:** Python 3.8+ &nbsp;|&nbsp; FFmpeg in `PATH`