Neuro-Grasp is a modular computer vision system designed to calculate 6-DOF grasp poses for unknown objects without relying on large machine learning datasets. It utilizes Principal Component Analysis (PCA) to geometrically determine an object's primary axis (Orientation) and secondary axis (Grasp Approach Vector) in real-time.
Visual verification of the PCA Solver identifying Major and Minor axes for optimal grasp planning.
| Cycle 1: Orientation | Cycle 2: Alignment | Cycle 3: Precision |
|---|---|---|
  |
  |
  |
| Angle: -19.4° | Angle: -23.9° | Angle: 94.4° |
| Status: ✅ Target Detected | Status: ✅ Target Detected | Status: ✅ Target Detected |
Instead of stochastic guessing, we calculate the object's Eigenvectors from the covariance matrix of its contour points.
- First Eigenvector (Major Axis - Red): Defines the object's length and primary orientation.
- Second Eigenvector (Minor Axis - Blue): Defines the object's width, used to set the Grasp Approach Vector.
- Center of Mass (Yellow Dot): The geometric centroid target for the end-effector.
This repository adheres to strict Systems Engineering software standards to ensure reliability in industrial environments.
| Metric | Status | Standard |
|---|---|---|
| Static Analysis | 10.00/10 |
Pylint (PEP 8 Strict) |
| Unit Testing | 12/12 PASSED |
unittest (PCA Math & PACE Drivers) |
| Fault Tolerance | PACE | Self-healing Driver Recovery (Primary/Alt/Cont/Emerg) |
| Kinematics | VERIFIED | Accurate Pixel-to-MM Coordinate Mapping |
The system includes test_neuro_grasp.py, covering:
- PCA Orientation Math: Validating eigenvector stability for horizontal and vertical parts.
- Driver Resilience: Testing the "PACE" self-healing logic during transient hardware signal loss.
- Coordinate Integrity: Verifying the pixel-to-millimeter transformation for robot frame integration.
The system performs a pre-flight unit test check before initializing the camera driver.
python neuro_grasp.py
python test_neuro_grasp.py
- Python 3.10+
- OpenCV (Geometric Perception)
- NumPy (Matrix Math)
- Matplotlib (Telemetry Visualization)
Author: Charles Austin (Senior Systems Architect) Focus: Computer Vision, Robotics Perception, and Deterministic AI.