Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,62 @@
+---
+license: mit
+tags:
+- reinforcement-learning
+- robotics
+- ppo
+- navigation
+---
+
+# JhonDOE Robot Models
+
+Trained models for the JhonDOE humanoid robot project.
+
+## Models
+
+### Navigate 2D (Phase 1)
+- **Task**: 2D navigation with obstacle avoidance
+- **Algorithm**: PPO (Proximal Policy Optimization)
+- **Input**: Position, velocity, orientation, target, 8 proximity sensors (13D)
+- **Output**: Forward velocity, angular velocity (2D continuous)
+- **Training**: 1M timesteps, ~30-60 minutes on T4 GPU
+- **Success Rate**: ~90%+
+
+Models uploaded automatically from: https://huggingface.co/spaces/DanielDel89/JhonDOE
+
+## Usage
+
+Download models using the hub manager:
+
+```bash
+cd execution
+python model_hub.py --mode download --repo DanielDel89/jhondoe-models --model navigate_2d_v01.zip
+```
+
+Or use in Python:
+
+```python
+from stable_baselines3 import PPO
+
+# Load model
+model = PPO.load("navigate_2d_v01.zip")
+
+# Use for inference
+obs = env.reset()
+action, _ = model.predict(obs)
+```
+
+## Project Links
+
+- **Training Space**: https://huggingface.co/spaces/DanielDel89/JhonDOE
+- **Documentation**: See training Space for full architecture details
+
+## Training Pipeline
+
+1. Configure training in Space (timesteps, parallel envs, etc.)
+2. Start training with GPU (T4 recommended)
+3. Model auto-uploads to this repository
+4. Download and evaluate locally
+
+## Model Versions
+
+Models are versioned with timestamps and tracked here. Latest versions are always recommended.