LDED-FusionNet Dataset: Multisensor Audio-Visual Data for Laser-Directed Energy Deposition

LDED-FusionNet Dataset: Multisensor Audio-Visual Data for Laser-Directed Energy Deposition

The LDED-FusionNet Dataset is a comprehensive multisensor dataset designed for audio-visual fusion in Laser-Directed Energy Deposition (LDED) processes. This dataset supports research in defect detection, quality monitoring, and multimodal machine learning models for metal additive manufacturing.

This dataset corresponds to the LDED-FusionNet GitHub repository:
🔗 GitHub Repository

It has been used in multiple peer-reviewed publications focusing on feature-level fusion of acoustic and visual data, including:

• Multisensor Fusion-Based Digital Twin for Localized Quality Prediction in Robotic LDED (RCIM 2023).
• In-situ Defect Detection in LDED with Machine Learning and Multi-Sensor Fusion (JMST 2024).
• Inference of Melt Pool Visual Characteristics in LAM Using Acoustic Signal Features and Robotic Motion Data (ICCAR 2024).

Key Features

• Synchronized Audio-Visual Data: High-fidelity coaxial melt pool images and multi-domain acoustic signals recorded during LDED processing.
• Cross-Modality Feature Fusion: Enables research on acoustic-vision feature relationships for process quality monitoring.
• Defect Detection Labels: Ground truth annotations for melt pool stability, defect types, and process variations.
• Robotic Motion Data: Captures spatial dependency of process dynamics, providing tool-center-point (TCP) motion logs.
• Benchmark for Machine Learning: Supports feature extraction, deep learning models, and anomaly detection.

Usage

This dataset is useful for:

• Developing AI models for real-time defect detection in LDED
• Exploring acoustic-visual correlations for quality prediction
• Improving multisensor fusion techniques in metal additive manufacturing
• Benchmarking ML algorithms for process monitoring and control

Citation

If you use this dataset in your research, please cite:

@dataset{chen2024lded_fusionnet,
  author    = {Chen, Lequn},
  title     = {LDED-FusionNet Dataset: Multisensor Audio-Visual Data for Laser-Directed Energy Deposition},
  year      = {2024},
  publisher = {Zenodo},
  url       = {https://zenodo.org/record/[Dataset-ID]}
}