3D PAUT Visualization and Real-time Robotics Integration

Results (Ongoing)

Editor.2.mp4

Editor.mp4

Overview

This project focuses on real-time 3D visualization of Phased Array Ultrasonic Testing (PAUT) data and integrates it with robotic control for industrial applications. It leverages Vulkan for high-performance rendering and includes plans for AI-based optimization in the future.

Note: The source code is private due to privacy policies. For inquiries, contact:

📧 Email: [email protected]

🔗 LinkedIn: Nguyen Tuan

Key Features

1. Phased Array Ultrasound in Realtime

• Real-time control and processing of OmniScan PAUT data.
• Full rendering and handling of A, B, S, and C-scan data.
• 3D data visualization using Vulkan.
• Performance:
  • Rendered 1,508,832 pixels across 4 buffers:
    • SViewBuf: 292,666 pixels
    • CViewBuf: 49,000 pixels
    • BViewBuf: 1,166,000 pixels
    • AViewBuf: 1,166 pixels
  • Achieved 2ms total frame time (~500 FPS), meeting industrial-grade real-time requirements.
  • System demonstrates excellent GPU utilization and optimized data processing pipeline.
• Designed for high scalability with support for increased resolution and advanced features.

2. Robot Control in Realtime

• Inverse Kinematics for 6 FreeBase Robot RealTime
• Real-time robotic arm control with IPC and C++.
• Vision-based detection for autonomous PAUT scanning.

3. AI Application (Planned)

• AI integration to enhance FEM analysis and optimize performance.
• Plans to use GPU-based rendering and OpenCV for AI detection at high frame rates.

4. FEM Integration

• Finite Element Method (FEM) applied to processed PAUT data.
• CUDA acceleration for FEM calculations.
• Integrated into Vulkan viewport for simulation.

Technologies Used

• PAUT: Real-time Phased Array Ultrasound processing (C++)

• Robotics: Real-time Yaskawa robot arm control (C++)

• Vulkan: High-performance 3D rendering (C++)

• IPC: Inter-process communication for real-time systems (C++)

• FEM: Finite Element Method simulation (C++)

• CUDA: Accelerated FEM computation (C++)

• AI (Planned): AI optimization and real-time detection.

• Render Performance:

  • Successfully rendered ~1.5 million pixels in 2ms (~500 FPS).
  • Exceeds real-time industrial standards (60-120 FPS).

• Buffer Details:

  • SViewBuf: 292,666 pixels
  • CViewBuf: 49,000 pixels
  • BViewBuf: 1,166,000 pixels
  • AViewBuf: 1,166 pixels

• Efficiency:

  • Maintains high scalability and GPU utilization.
  • Provides a strong foundation for integrating additional features like AI and FEM.

Applications

• Real-time robotics control and monitoring.
• Advanced NDT (Non-Destructive Testing) applications.
• Autonomous industrial inspections with enhanced AI integration.

---

This project combines real-time robotics, advanced 3D rendering, and scalable data processing to support high-performance industrial applications.