Generative Design and Topological Optimization of Climbing Holds

Abstract

The design and development of climbing holds has traditionally required significant time and hands-on effort, relying heavily on manual modeling, iterative refinement and expert knowledge. This thesis primarily investigates whether generative design, topology optimization and additive manufacturing methods can be effectively applied to the development of climbing holds. A secondary objective was to use these findings to establish a structured pipeline for climbing holds generation. A dataset of 3D-scanned climbing holds was created, representing predefined hold categories to enable unsupervised learning of geometric features. Two independent generative methods were developed and evaluated to assess their capability to generate novel within-category climbing hold geometries. These methods produced new hold designs as point clouds, which were subsequently reconstructed as CAD models for further refinement. The generated designs were topology optimized to improve material efficiency while maintaining structural integrity. Additional CAD refinement was performed to ensure manufacturability, validate tolerances and apply final surface textures. The results demonstrate that these design and manufacturing methods can be successfully adapted for climbing hold development, while also highlighting their potential to reduce many of the limitations associated with traditional design practices. In addition, this thesis establishes a functional pipeline that can serve as a foundation for future development in AI-assisted climbing hold design. This research contributes to the broader field of AI-assisted product development by demonstrating the feasibility of combining machine learning with engineering optimization in a specialized design context. Future work should focus on expanding the dataset, improving generative precision and exploring the commercial viability of the methodology.