TL;DR
This paper introduces a novel "composite standard" – a precisely machined cone and cylinder – designed to overcome the limitations of traditional reference spheres for multi-sensor registration in high-precision Coordinate Measuring Machines (CMMs). By measuring features on this composite standard with a video camera, tactile probe, and chromatic confocal sensor, the researchers achieve stable and accurate alignment of these heterogeneous sensors, crucial for complex dimensional metrology.
What Problem Does This Paper Address?
In high-precision dimensional measurement, especially with complex workpieces, integrating data from multiple sensors (like cameras, tactile probes, and optical sensors) is essential. This integration, known as registration, requires aligning the coordinate systems of each sensor. Traditional methods often rely on reference spheres. However, the authors highlight significant drawbacks of spheres for high-precision CMMs: optical sensors struggle with the sphere's gradually changing edge and light diffraction, leading to inaccurate center detection. Optical sensors also have angle limitations, and tactile probes may not achieve sufficient coverage on a sphere for high accuracy. These issues compromise the overall measurement quality. This paper aims to provide a more robust and accurate registration solution for multi-sensor CMMs.
Methods
The core of this work is the development and validation of a composite standard. This standard is an ultra-precision machined component featuring a cone and a cylinder sharing a common central axis. The key idea is that specific features on this standard are designed to be precisely measured by each of the three sensors:
- Video Camera (VC): Captures the top circle of the cone, which provides a sharp feature suitable for optical measurement.
- Tactile Probe (TP): Contacts the cylinder's top plane and generatrix to determine the intersection point with the cylinder's symmetry axis.
- Chromatic Confocal Displacement Sensor (CC): Scans the cone's generatrix and the cylinder's top plane to find the center of the bottom circle of the cone.
The registration principle relies on these sensors measuring features that converge to a common point. The paper details the mathematical models for fitting these features (circles, planes, and cone intersections) using least-squares methods and discusses error propagation analysis. The method involves transforming the standard to an ideal state, calculating Jacobian matrices, and using a Gauss-Newton iteration for optimization, particularly for the CC sensor's intersection circle measurement.
Datasets
The paper does not use pre-existing datasets for training or evaluation. Instead, it relies on:
- Simulated Data: 200 simulations were performed for each sensor to evaluate the registration precision under controlled noise conditions.
- Real Experimental Data: Measurements were taken using the composite standard on a high-precision CMM equipped with the three specified sensors. The standard was rotated to assess stability.
The paper introduces the composite standard itself as a new benchmark for multi-sensor registration, specifically designed to address the shortcomings of reference spheres.
Key Insights and Ideas
- Composite Standard Superiority: The composite standard, with its distinct geometric features (cone and cylinder), offers more reliable and precise measurement points for each sensor type compared to a single reference sphere.
- Addressing Sensor-Specific Limitations: The design directly tackles the known issues with each sensor: the VC benefits from a sharp cone edge, the TP uses a planar and linear feature, and the CC can precisely locate an intersection.
- Coaxiality is Key: The shared central axis of the cone and cylinder is crucial for establishing a unified coordinate system.
- Error Analysis is Rigorous: The paper provides detailed theoretical analysis of error propagation for each sensor's measurement, which is essential for understanding the overall registration uncertainty.
- Stability Demonstrated: Both simulations and real experiments show that the proposed method is stable, with consistent results across different rotations of the standard.
Key Results
- Stable Registration: The proposed registration method using the composite standard demonstrated stability in both simulations and real experiments.
- Reduced Uncertainty: While specific quantitative accuracy figures for the final registration are not explicitly stated as a single number, the paper shows that the standard deviation (SD) of the offsets between sensors is generally low, indicating good precision. For instance, the SD for offsets between the VC and TP was in the order of micrometers.
- Comparison to Sphere: The paper suggests that the composite standard offers advantages over reference spheres, particularly in avoiding systematic errors associated with optical sensor interactions with spherical surfaces.
- Feasibility Confirmed: The results from simulations and real experiments validate the feasibility of using the composite standard for multi-sensor registration in high-precision CMMs.
Limitations and Future Work
- VC Measurement Limitations: The paper notes that offsets in the x and y directions between the VC and TP were "liable to the rotation of the standard," suggesting that the top circle of the cone might not be perfectly coaxial with the cylinder's axis, or that the VC's light source might not be ideal. The z-direction measurement for the VC is also considered less critical as it's primarily used for 2D measurements.
- CC Optimization Challenges: For the CC sensor, the paper mentions that while simulations showed good results, the SDs were larger than for other sensors, possibly due to the difficulty in obtaining precise results without adequate constraints using the optimization method.
- Future Work: The authors suggest that further work should focus on accuracy improvement through modified methods or compensation techniques for the current registration approach.
Summary updated