Enforcing essential boundary conditions on domains defined by point clouds

Frank Hartmann,Stefan Kollmannsberger

DOI: https://doi.org/10.48550/arXiv.2111.06502

2021-11-05

Abstract:This paper develops and investigates a new method for the application of Dirichlet boundary conditions for computational models defined by point clouds. Point cloud models often stem from laser or structured-light scanners which are used to scan existing mechanical structures for which CAD models either do not exist or from which the artifact under investigation deviates in shape or topology. Instead of reconstructing a CAD model from point clouds via surface reconstruction and a subsequent boundary conforming mesh generation, a direct analysis without pre-processing is possible using embedded domain finite element methods. These methods use non-boundary conforming meshes which calls for a weak enforcement of Dirichlet boundary conditions. For point cloud based models, Dirichlet boundary conditions are usually imposed using a diffuse interface approach. This leads to a significant computational overhead due to the necessary computation of domain integrals. Additionally, undesired side effects on the gradients of the solution arise which can only be controlled to some extent. This paper develops a new sharp interface approach for point cloud based models which avoids both issues. The computation of domain integrals is circumvented by an implicit approximation of corresponding Voronoi diagrams of higher order and the resulting sharp approximation avoids the side-effects of diffuse approaches. Benchmark examples from the graphics as well as the computational mechanics community are used to verify the algorithm. All algorithms are implemented in the FCMLab framework and provided at <a class="link-external link-https" href="https://gitlab.lrz.de/cie_sam_public/fcmlab/" rel="external noopener nofollow">this https URL</a>. Further, we discuss challenges and limitations of point cloud based analysis w.r.t. application of Dirichlet boundary conditions.

Numerical Analysis

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is how to impose Dirichlet boundary conditions on the domain defined by point clouds. Specifically, the paper proposes a new method to process point - cloud data obtained from laser or structured - light scanners. These data are usually used to scan existing mechanical structures, and these structures may not have corresponding CAD models, or there are deviations in shape or topology between the actual object and the CAD model.

### Main Problems
1. **Limitations of Traditional Methods**:
- The traditional processing method is to reconstruct the CAD model from the point cloud through surface reconstruction and then generate a boundary - conforming mesh. This requires a large number of pre - processing steps and is prone to problems such as surface topological defects and local over - refinement.
- Using the embedded domain finite element method allows direct analysis on the point cloud, but this method uses a non - boundary - conforming mesh, resulting in weak imposition of Dirichlet boundary conditions, increasing computational complexity and unnecessary side effects (such as the gradient problem of the solution).

2. **Shortcomings of Existing Methods**:
- Existing methods usually use the diffuse interface approach to impose Dirichlet boundary conditions. This method requires the calculation of domain integrals, resulting in significant computational overhead, and it is difficult to control the behavior of the solution and its gradient in the thin - limit case.

### Solution Proposed in the Paper
The paper develops a new sharp interface approach, aiming to overcome the above problems. Specifically:
- **Avoid Calculating Domain Integrals**: By implicitly approximating a higher - order Voronoi diagram, the need to calculate domain integrals is bypassed.
- **Avoid Side Effects of the Diffuse Interface Approach**: The new method provides a clear boundary separation, avoiding the undesired behavior of the solution and its gradient brought by the diffuse interface approach.

### Method Overview
1. **Local Boundary Approximation**: Calculate the local plane approximation based on the k - nearest neighbors in the point cloud.
2. **Diffuse Interface Approach**: Relax the boundary into a layer by introducing a thickness parameter ϵ, thereby converting the boundary integral into a domain integral.
3. **Sharp Interface Approach**: Integrate directly on ΓD,sharp without explicitly calculating the Voronoi diagram, and implement it through numerical integration schemes such as the bisection method.

### Conclusion
The paper verifies the effectiveness of the new method through multiple benchmark tests and discusses the challenges and limitations of imposing Dirichlet boundary conditions in point - cloud analysis. Although the new method performs well in handling smooth boundaries, it still has certain limitations when dealing with non - smooth or open boundaries.

### Formula Summary
- **Distance Function**: \[ d_{\Gamma_D}(x) \]
- **Regularized Dirac Delta Distribution in the Diffuse Interface Approach**: \[ \delta_\epsilon(x) = \begin{cases}
\frac{1}{2\epsilon}\left(1 + \cos\left(\frac{\pi x}{\epsilon}\right)\right) & \text{if } |x| \leq \epsilon \\
0 & \text{otherwise}
\end{cases} \]
- **Penalty Matrix in the Sharp Interface Approach**: \[ K^{ce}_P \approx \beta \sum_{vo} \sum_{sd} \sum_{i} \delta_{ce}^{cell}(x(\xi_i)) \delta_{vo}^{voronoi}(x(\xi_i)) (N^T N) \|J_{sd}\| |\xi_i v_i| \]

Through these improvements, the paper provides a more efficient and accurate method to handle Dirichlet boundary conditions on point - cloud data.