Refine
Document Type
- Preprint (2)
- Conference Proceeding (1)
Language
- English (3)
Has Fulltext
- no (3)
Is part of the Bibliography
- no (3)
Keywords
- 3D deep learning (1)
- CAD (1)
- Computer Science - Artificial Intelligence (1)
- Graph neural networks (1)
- Machining feature recognition (1)
- Reinforcment Learning (1)
- Robotics (1)
- Sim2Real (1)
- Snthetic data generation (1)
- Surface models (1)
Institute
Cutting and Packing problems are occurring in different industries with a direct impact on the revenue of businesses. Generally, the goal in Cutting and Packing is to assign a set of smaller objects to a set of larger objects. To solve Cutting and Packing problems, practitioners can resort to heuristic and exact methodologies. Lately, machine learning is increasingly used for solving such problems. This paper considers a 2D packing problem from the furniture industry, where a set of wooden workpieces must be assigned to different modules of a trolley in the most space-saving way. We present an experimental setup to compare heuristics, constraint optimization, and deep reinforcement learning for the given problem. The used methodologies and their results get collated in terms of their solution quality and runtime. In the given use case a greedy heuristic produces optimal results and outperforms the other approaches in terms of runtime. Constraint optimization also produces optimal results but requires more time to perform. The deep reinforcement learning approach did not always produce optimal or even feasible solutions. While we assume this could be remedied with more training, considering the good results with the heuristic, deep reinforcement learning seems to be a bad fit for the given use case.
Driven by increasing customer demands, manufacturing processes encompass increasingly intricate workflows. The industry relies on computer-aided process planning software to effectively manage the intricacies of these complex manufacturing processes. This software is crucial to analyze computer-aided design data for a product and determining the required machining steps. However, a notable challenge arises, particularly in the case of custom products, where the machining steps can significantly vary depending on the available machinery and the employees' preferences.This study introduces a configurable automated feature recognition framework based on expert knowledge. Experts can encode their insights and expertise within this framework using a configurable synthetic data generator. Modern graph neural networks learn from the data generated by this data generator, achieving an average recognition accuracy up to 80% (F1-score) with a runtime performance of 21.98 milliseconds per model. Importantly, it accomplishes this even when confronted with highly intersecting machining features without requiring data conversion into alternative formats, such as voxel or pixel representations.
Randomization is currently a widely used approach in Sim2Real transfer for data-driven learning algorithms in robotics. Still, most Sim2Real studies report results for a specific randomization technique and often on a highly customized robotic system, making it difficult to evaluate different randomization approaches systematically. To address this problem, we define an easy-to-reproduce experimental setup for a robotic reach-and-balance manipulator task, which can serve as a benchmark for comparison. We compare four randomization strategies with three randomized parameters both in simulation and on a real robot. Our results show that more randomization helps in Sim2Real transfer, yet it can also harm the ability of the algorithm to find a good policy in simulation. Fully randomized simulations and fine-tuning show differentiated results and translate better to the real robot than the other approaches tested