Abstract:In robotic inspection of aviation parts, achieving accurate pairwise point cloud registration between scanned and model data is essential. However, noise and outliers generated in robotic scanned data can compromise registration accuracy. To mitigate this challenge, this article proposes a probability-based registration method utilizing Gaussian Mixture Model (GMM) with local consistency constraint. This method converts the registration problem into a model fitting one, constraining the similarity of posterior distributions between neighboring points to enhance correspondence robustness. We employ the Expectation Maximization algorithm iteratively to find optimal rotation matrix and translation vector while obtaining GMM parameters. Both E-step and M-step have closed-form solutions. Simulation and actual experiments confirm the method's effectiveness, reducing root mean square error by 20% despite the presence of noise and outliers. The proposed method excels in robustness and accuracy compared to existing methods.

What problem does this paper attempt to address?

The paper mainly addresses the issue of point cloud registration in the robotic inspection of aerospace parts and proposes a new solution. Specifically, the study aims to solve the following problems: 1. **Improving Registration Accuracy**: In the robotic inspection of aerospace parts (such as blades and aircraft skins), the scanned data usually contains noise and outliers, which severely affect the accuracy of point cloud registration. 2. **Enhancing Robustness**: Existing methods often perform poorly when dealing with data containing noise and outliers, especially in complex measurement environments. Therefore, a method that can effectively address these issues is needed. ### Solution Overview To overcome the above challenges, the authors propose a registration method based on the Locally Consistent Gaussian Mixture Model (LCGMM). The main contributions of this method are: 1. **Probabilistic Model Framework**: The registration problem is transformed into a probabilistic model fitting problem, using the Gaussian Mixture Model (GMM) to describe the correspondence between point clouds. By constraining the similarity of the posterior distributions between neighboring points, the correctness and robustness of the correspondences are improved. 2. **Iterative Expectation Maximization Algorithm**: The Iterative Expectation Maximization (EM) algorithm is used to optimize the model parameters, including the rotation matrix, translation vector, and the variance of the GMM. The EM algorithm ensures the effectiveness of the model parameter estimation and provides closed-form solutions at each iteration. 3. **Local Consistency Constraint**: A local consistency constraint is introduced to further enhance the robustness of the method. This constraint limits the similarity of the posterior distributions of neighboring points, allowing the method to maintain high registration accuracy even in the presence of noise and outliers. ### Experimental Validation To verify the effectiveness of the proposed method, the paper conducts both simulation and real-world experiments. In the simulation experiments, the authors evaluate the robustness of the method to noise and outliers by adding different levels of these disturbances to the scanned point clouds. In the real-world experiments, real scanned data of blades and skin samples are used to validate the performance of the method in actual industrial environments. ### Conclusion In summary, this paper proposes a new point cloud registration method aimed at improving the accuracy and robustness of point cloud registration in robotic inspection. By introducing a probabilistic model framework and local consistency constraints, the method can achieve high-quality registration results even in the presence of noise and outliers.

Robust Point Cloud Registration in Robotic Inspection with Locally Consistent Gaussian Mixture Model

A Gmm Based Uncertainty Model for Point Clouds Registration

A Robust Probability-based Joint Registration Method of Multiple Point Clouds Considering Local Consistency

Non-rigid Point Set Registration Based on Gaussian Mixture Model with Integrated Feature Divergence

Robust Feature-Based Point Registration Using Directional Mixture Model

Robust Generalized Point Cloud Registration With Expectation Maximization Considering Anisotropic Positional Uncertainties

Unsupervised Point Cloud Registration by Learning Unified Gaussian Mixture Models

Point Cloud Registration Leveraging Structural Regularity in Manhattan World

A Robust Aerial Image Registration Method Using Gaussian Mixture Models.

Non-Rigid 3D Point Set Registration With Reliable Hybrid Mixture Model.

An Effective Point Cloud Registration Method Based on Robust Removal of Outliers

Gaussian Mixture Model-Signature Quadratic Form Distance Based Point Set Registration.

Density-adaptive Registration of Pointclouds Based on Dirichlet Process Gaussian Mixture Models.

Non-Rigid Point Set Registration Using Multi-Feature and Gaussian Mixture Model

Laser point cloud registration method based on iterative closest point improved by Gaussian mixture model considering corner features

Effective Multiview Registration of Point Clouds Based on Student’s-t Mixture Model

Robust point cloud registration using Hough voting-based correspondence outlier rejection

DeepGMR: Learning Latent Gaussian Mixture Models for Registration

Dependent landmark drift: robust point set registration with a Gaussian mixture model and a statistical shape model

Robust Non-Rigid Point Set Registration Using Student's-T Mixture Model.

Non-rigid point set registration based on local neighborhood information support