AutoSeatAdjust-EmbeddedAI

An AI-powered adaptive seat positioning system that predicts ideal longitudinal seat distance and seat angle from the driver’s height. Designed for embedded use on systems like Raspberry Pi or NVIDIA Jetson, this project trains and runs lightweight regression models with user feedback support.

---

📌 Features

• One input, two predictions: Maps driver Height (cm) → True Longitudinal Distance and Seat Angle
• XGBoost-based regressors with support for real-time inference
• User feedback loop: Adjustments can be saved and used to retrain the model instantly
• Excel dataset support with safe appends via openpyxl
• Cross-platform (Mac, Windows, Ubuntu)

---

🧠 How It Works

1. Train two separate XGBoost regressors using historical seat fitment data
2. Save the trained models in a single .pth file using joblib
3. During inference, prompt the user for height
4. Predict seat distance and seat angle
5. If the prediction is wrong, collect corrected values, save them back to the Excel file, and retrain the models

---

💾 Dataset Format

Your Excel file (Extrapolated_Participant_Data.xlsx) should contain the following columns:

Height (cm)	True Long. Distance (cm)	True Seat Angle (from vertical)
172	85.94	14.75
...	...	...

---

🛠️ Setup Instructions

1. Clone the Repository

git clone https://github.com/YourUsername/AutoSeatAdjust-EmbeddedAI.git
cd AutoSeatAdjust-EmbeddedAI

2. Set Up a Virtual Environment

macOS / Ubuntu

python3 -m venv venv
source venv/bin/activate
pip install -r dependencies.txt

Windows (PowerShell)

py -m venv venv
venv\Scripts\Activate.ps1
pip install -r dependencies.txt

---

🧪 Training the Model

python src/DTreeClassifier.py

• Loads the Excel dataset
• Trains two regressors
• Saves them to models/pred_model.pth

---

🤖 Running Inference with Feedback

python src/infer_height.py

• Input: Driver height
• Output: Predicted seat position and angle
• If the prediction is off, enter your corrected values and the system will:
  • Append them to the dataset
  • Retrain the model with the new datapoint

---

🔄 Real-Time Updates

Any new data collected during use is:

• Stored safely in the same Excel file
• Used for immediate retraining
• Persisted via joblib in a .pth file

---

🚀 Future Work

• Add GUI for in-car use
• Replace height input with real-time eye-tracking using OpenCV
• Deploy on Jetson Orin or Raspberry Pi with model quantization
• Integrate calibration profiles per driver

---

📜 License

MIT License – use, modify, or deploy freely.

---

🙌 Acknowledgements

• XGBoost
• scikit-learn
• openpyxl

---

Need help? Open an issue or start a discussion! 💬