FANUC iRVision System Implementation

Author: Dawson Burgess, Computer Science Department, University of Idaho, Moscow, ID, United States
Email: burg1648@vandals.uidaho.edu

Overview

This repository contains the research paper and documentation for implementing the FANUC iRVision system on the CRX-10iA collaborative robot to detect and pick up large block dice. The project was conducted using the robot’s teaching pendant (TP) exclusively, without external APIs, focusing on vision-guided robotics for object recognition and manipulation.

Key Features

• Vision Setup: Configured a robot-mounted iRVision camera with auto-calibration.
• Process Design: Developed a 2D single-view vision process to locate dice.
• Robot Control: Programmed TP code for precise pick-and-place operations.
• Objective: Achieved autonomous dice detection and handling in a controlled workspace.

Project Structure

• FANUC_iRVision_System_Implementation.pdf: Research paper detailing setup, methods, and results.
• README.md: This file, providing an overview and instructions.

Methodology

iRVision Object Camera Setup

• Camera Installation: Mounted on the CRX-10iA near the end-of-arm tooling with optimized lighting.
• Calibration: Used "Robot-Generated Grid Calibration" with ~20 orbital positions (30+ minutes).
• Training: Captured dice images, masked pips for edge detection.

Vision Process Setup

• Type: 2D Single-View Vision Process, storing data in Vision Register (VR[1]).
• Offset Mode: "Found Position" for direct Cartesian coordinates, avoiding fixed frame issues.
• Tools: Configured Snap Tool (200ms exposure) and GPM Locator Tool (60% score threshold).

TP Program Implementation

• Core Commands: VISION RUN_FIND, VISION GET_OFFSET, and position register assignments.
• Motion Handling: Used Wjnt to manage wrist configuration and avoid "flipped" states.
• Workflow: Detects dice, moves to position, picks, and places at table center, looping until stopped.

Results

• Success: The system reliably detects and picks dice, returning to a waiting position above the workspace.
• Performance: Consistent operation under controlled lighting, mimicking industrial pick-and-place tasks.

Challenges and Limitations

• Documentation: Limited official resources; relied on FANUC iRVision manual and peer assistance.
• Calibration: Long process requiring uninterrupted button hold; prone to restart if errors occur.
• Frames: Initial confusion with robot motion and configuration settings (e.g., "flipped" states).

Future Work

• Extend to multi-workspace detection (e.g., full table coverage).
• Explore Python API integration if camera access becomes available.
• Enhance with dynamic lighting adjustments for robustness.

Installation and Usage

This project was implemented directly on the FANUC CRX-10iA teaching pendant, requiring no external software setup beyond the robot’s firmware. To replicate:

1. Hardware: FANUC CRX-10iA with iRVision camera module.
2. Setup: Follow the paper’s camera and vision process instructions.
3. Program: Input the TP code from Figure 11 in the paper.
4. Run: Execute on the TP, ensuring proper lighting and dice placement.

Note: No Python code is included, as the project avoids the API per constraints.

License

This project is licensed under the MIT License.

Contact

For inquiries, contact Dawson Burgess at burg1648@vandals.uidaho.edu.

Citation

If you use this project, please cite: Dawson Burgess. (2025). FANUC iRVision System Implementation. University of Idaho.

Acknowledgments

Special thanks to Jacob Friedberg for assistance with troubleshooting and resetting robot settings.