Add model-specific Description summary

Adds a specialized Description paragraph + bullets (analogous to Cosmos3-Super-Image2Video-4Step) summarizing what this model is and how it differs, while keeping the existing generic Cosmos3 overview.

README.md

@@ -46,6 +46,15 @@ countDownloads:
 
 ## Description
 
+Cosmos3-Edge-Policy-DROID is a robot-manipulation policy in the compact 4B Cosmos3-Edge family, specialized for the DROID robot platform. Given language instructions and visual observations from a DROID robot, it generates robot action trajectories for manipulation and control tasks.
+
+- **What's different from the base model:** Where Cosmos3-Edge is a general-purpose Omnimodal world model, this variant is specialized into a control policy for the DROID platform — optimized to map language instructions plus visual observations directly to executable robot action trajectories for manipulation and control.
+- **Input / Output:** Language instruction + DROID visual observations in → robot action trajectories out.
+- **Example usage and output:** See the [Usage](https://huggingface.co/nvidia/Cosmos3-Edge-Policy-DROID#usage) / Quickstart section.
+- **Hardware:** The compact 4B size is intended for efficient single-GPU deployment; see [Usage](https://huggingface.co/nvidia/Cosmos3-Edge-Policy-DROID#usage).
+
+Cosmos3-Edge-Policy-DROID was developed by NVIDIA as a part of Cosmos3.
+
 Cosmos3 is a collection of Omnimodal world models capable of generating dynamic, high-quality video, image, audio, and action commands from combinations of text, image, video, and action trajectory inputs. It serves as a foundational building block for a broad range of Physical AI applications and research spanning world understanding, world generation, simulation, and embodied policy learning.
 
 This model is ready for commercial and non-commercial use.