Screw Dynamics SINDy

A computational framework for modeling robotic screw-driving dynamics using Sparse Identification of Nonlinear Dynamics (SINDy).

Table of Contents

• Screw Dynamics SINDy
  • Table of Contents
  • Overview
    • Key Features
  • Paper
  • Quick Start
  • Repository Structure
  • Installation
  • Installation
    • Prerequisites
    • Setup
  • Usage
    • Training Models
    • Data Processing
    • Analysis
  • Data Access
  • Contributing
  • Citation
  • License

Overview

This project develops interpretable dynamical models for robotic screw-driving operations using the SINDy framework. The approach combines computer vision-based feature extraction with sparse regression to identify governing equations that describe screw tip dynamics.

Key Features

• SINDy modeling: Sparse identification of nonlinear dynamics
• Computer vision: Feature tracking and pose detection
• Baseline models: LSTM and MLP for performance comparison
• Data processing: Collection and preprocessing utilities
• Analysis tools: Jupyter notebooks for visualization

Paper

Research Paper: Link

This work is documented in our research paper in detail.

Quick Start

# 1. Setup environment
source activate_env.sh

# 2. Train SINDy model
python src/main.py --data_dir ./data/data_1 --epochs 1000

# 3. Run baseline comparison
python baseline/train.py --model_type lstm --data_dir ./data/data_1

# 4. Open analysis notebooks
jupyter lab notebook/

Repository Structure

├── src/                    # SINDy implementation → [README](src/README.md)
├── baseline/               # Baseline models (LSTM, MLP) → [README](baseline/README.md)
├── scripts/                # Data processing utilities → [README](scripts/README.md)
├── notebook/               # Analysis notebooks → [README](notebook/README.md)
├── data/                   # Sample data → [README](data/README.md)
├── archived/               # Previous implementations
├── .github/                # CI/CD workflows
├── requirements.txt        # Python dependencies
├── setup.py               # Package installation
└── LICENSE                # MIT License

Installation

Installation

Prerequisites

• Python 3.8+
• PyTorch, OpenCV, scikit-learn, pandas, numpy, matplotlib, jupyter

Setup

1. Clone the repository:

git clone https://github.com/l0g1c-80m8/screw-dynamics-sindy.git
cd screw-dynamics-sindy

2. Setup environment:

source activate_env.sh

Usage

Training Models

# SINDy model
python src/main.py --data_dir ./data/data_1 --epochs 1000

# Baseline models  
python baseline/train.py --model_type lstm --data_dir ./data/data_1

Data Processing

# Extract screw tip poses
python scripts/screwtip_pose_detection.py --data_dir ./raw_data

Analysis

# Open Jupyter notebooks
jupyter lab notebook/

Data Access

Experimental data: Google Sheets Dataset

• Observation data: Screw tip pose and position tracking
• Sensor data: Force/torque measurements and system states
• Image sequences: Visual data for feature extraction

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Submit a pull request

Citation

Manyar, O. M., Patel, R., & Gupta, S. K. (2026). Autonomous robotic screwdriving for high-mix manufacturing. Robotics and Computer-Integrated Manufacturing, 98, 103172. https://doi.org/10.1016/j.rcim.2025.103172

@article{Manyar_Patel_Gupta_RCIM_2026,
  author = {Manyar, Omey M. and Patel, Rutvik and Gupta, Satyandra K.},
  title = {Autonomous robotic screwdriving for high-mix manufacturing},
  journal = {Robotics and Computer-Integrated Manufacturing},
  publisher = {Elsevier},
  year = {2026},
  volume = {98},
  pages = {103172},
  doi = {10.1016/j.rcim.2025.103172}
}

License

This project is licensed under the MIT License - see the LICENSE file for details.