A simultaneous calibration technique of the extrinsic and turntable for structured-light-sensor-integrated CNC system

Zhaoyao Shi,Tao Wang,Jiachun Lin

DOI: https://doi.org/10.1016/j.optlaseng.2020.106451

IF: 5.666

2021-03-01

Optics and Lasers in Engineering

Abstract:<p>The center of the mandrel-target is unique and does not get affected by the extrinsic angle of the projector. On this basis, a mandrel-target-based method is described to simultaneously calibrate the extrinsic parameters and the turntable of a multi-axis, structured-light-sensor-integrated, three-dimensional (3-D) scanning system. Besides, a cross-section-arc algorithm is developed to identify the conjugate pair. By computing the mandrel-center in the light plane, some collinear scanning points are generated along with the spindle scanning. Further, these calibration points are employed to the tip-tilt self-calibration of the turntable. In the two-scanning strategy, three sets of non-colinear points are derived simultaneously by relatively moving the sensor at distinct world coordinates. These generated points are fulfilled to calibrate the extrinsic parameters with the relative motion model. By determining the direction vectors, the skewed sensor coordinates are converted into the Cartesian world coordinates. The effectiveness of the proposed method is verified using a set of experiments. The accuracy of the proposed model is justified by comparing it with the measurements of a CMM trigger probe. The absolute measurement accuracy for a mandrel-target (about 7.4827 mm in radius and 16 mm in depth) is 0.002 mm. The target is cost-effective and simple in shape. The calibration procedure is easy to carry out and highly accurate.</p>

optics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: in the three - coordinate measurement system integrated with multi - axes and structured - light sensors, the problem of simultaneously calibrating extrinsic parameters and the turntable. Specifically, the author proposes a method based on the mandrel - target to achieve the synchronous calibration of the structured - light sensor and the turntable, and develops a cross - sectional arc algorithm to identify conjugate point pairs.

### Problem Background

Structured - light sensors (especially line - structured - light sensors) are widely used in the field of 3D visual measurement because of their high precision, non - contact operation, flexibility and other advantages. However, when these sensors are integrated with other multi - axis devices (such as CNC machine tools), accurate extrinsic parameter calibration is required to ensure the accuracy of measurement data. Traditional calibration methods usually require additional auxiliary tools and the process is complex and time - consuming.

### Paper Solutions

To solve the above problems, this paper proposes the following methods:

1. **Selection of Mandrel - Target**:
- Use the mandrel as a calibration target because its geometric center is not affected by the external angle of the projector.

2. **Identification of Conjugate Point Pairs**:
- Develop a cross - sectional arc algorithm. Generate collinear scanning points by calculating the mandrel center in the light plane, and use these points for the self - calibration of the tilt of the turntable.

3. **Dual - Scanning Strategy**:
- Scan the mandrel at two different positions respectively to generate non - collinear point sets for solving extrinsic parameters. This method can avoid the problem of multiple solutions generated by a single matrix equation and improve the calibration accuracy.

4. **Mathematical Model and Optimization**:
- Establish the overall transformation relationship from the non - orthogonal sensor coordinate system to the Cartesian world coordinate system, and optimize parameter estimation by the Levenberg - Marquardt iterative method.

### Calibration Process

1. **Overall Transformation Relationship**:
\[
\mathbf{P}_o=\mathbf{P}_{oL}+\mathbf{P}_{Lm}
\]
where \(\mathbf{P}_{oL}\) and \(\mathbf{P}_{Lm}\) are the vector representations of the same point on the target surface in the sensor coordinate system and the reference coordinate system respectively, and \(\mathbf{P}_o\) is the representation of the converted point in the working coordinate system.

2. **Identification of Conjugate Point Pairs**:
- Generate conjugate point pairs by matching common points in different coordinate systems. These point pairs are used to determine the mandrel center, and the sum of squared deviations is minimized by geometric fitting optimization.

3. **Turntable Tilt Calibration**:
- Use the collinear point set to fit a straight line and calculate the tilt angle \(\xi\) of the turntable:
\[
\xi = \arctan\left(\frac{\sqrt{(\Delta d_x)^2+(\Delta d_y)^2}}{\Delta z_m}\right)
\]

4. **Dual - Scanning Strategy**:
- Generate non - collinear point sets by scanning in two different directions for solving extrinsic parameters.

### Experimental Verification

The experimental results show that this method has high calibration accuracy, with an absolute measurement error of only 0.002 mm, and the calibration process is simple and easy to implement.

In conclusion, this paper aims to provide a simple and efficient method to synchronously calibrate the extrinsic parameters of the structured - light sensor and the turntable tilt, thereby improving the accuracy and reliability of the 3D visual measurement system.