Sketch2CAD: Sequential CAD Modeling by Sketching in Context

Abstract: We present a sketch-based CAD modeling system, where users create objects incrementally by sketching the desired shape edits, which our system automatically translates to CAD operations. Our approach is motivated by the close similarities between the steps industrial designers follow to draw 3D shapes, and the operations CAD modeling systems offer to create similar shapes. To overcome the strong ambiguity with parsing 2D sketches, we observe that in a sketching sequence, each step makes sense and can be interpreted in the \emph{context} of what has been drawn before. In our system, this context corresponds to a partial CAD model, inferred in the previous steps, which we feed along with the input sketch to a deep neural network in charge of interpreting how the model should be modified by that sketch. Our deep network architecture then recognizes the intended CAD operation and segments the sketch accordingly, such that a subsequent optimization estimates the parameters of the operation that best fit the segmented sketch strokes. Since there exists no datasets of paired sketching and CAD modeling sequences, we train our system by generating synthetic sequences of CAD operations that we render as line drawings. We present a proof of concept realization of our algorithm supporting four frequently used CAD operations. Using our system, participants are able to quickly model a large and diverse set of objects, demonstrating Sketch2CAD to be an alternate way of interacting with current CAD modeling systems.

• The paper introduces a deep-learning system that transforms sequential sketches into precise, editable CAD models.
• The methodology leverages a deep convolutional network for context-aware sketch interpretation and parameter estimation using synthetic training sequences.
• The system enhances user accessibility and design efficiency by enabling novices and experts to rapidly prototype complex CAD models through intuitive sketching.

Sketch2CAD: Sequential CAD Modeling by Sketching in Context

Introduction

"Sketch2CAD: Sequential CAD Modeling by Sketching in Context" presents a novel system that bridges sketch-based input with CAD modeling operations. The principal concept revolves around interpreting sequential sketches drawn by users into precise CAD models using deep neural networks. This methodology acknowledges the similar intent and progression present in traditional industrial design sketching and CAD operations despite their linguistic differences.

System Architecture

The Sketch2CAD system implements a pipeline composed of multiple stages:

1. Sketch Interpretation and Context Utilization:
   • Deep Network Architecture: A deep convolutional neural network is employed to interpret the user inputs. It takes both the current sketch and the context from the existing partial CAD model to determine the necessary operations.
   • Operation Recognition: The system supports four primary operations — extrusion, beveling, addition/subtraction of primitives, and sweeping shapes — and can accurately map user sketches to these operations.
2. Parameter Estimation:
   • Sketch Segmentation: Once an operation is identified, the system segments the sketch into parts representing different CAD features.
   • Optimization Process: Following segmentation, an optimization routine estimates the best-fitting parameters to modify the CAD model.
3. Training Data Generation:
   • Synthetic Sequences: Due to the lack of existing datasets, synthetic datasets of paired sketch-CAD sequences are generated, simulating realistic variations in design processes. This encompasses generating sequences by applying randomized operations to base geometries.

Implementation Considerations

• Operator Parameterization: Each operation is parameterized to encode its defining characteristics and interactions with the model’s existing geometry.
• Trade-offs: A significant challenge lies in the ambiguity of 2D sketches leading to multiple possible 3D interpretations. The system mitigates this by leveraging sequential context.
• Tool Interface and Real-time Feedback: The user interface supports efficient sketching and real-time visualization, allowing users to view immediate feedback on their inputs.

Practical Applications and Implications

• Industrial Design: Sketch2CAD simplifies the transition from initial design concepts to highly accurate and editable CAD models. This facilitates a fluid iteration process where design and CAD expertise are seamlessly integrated.
• User Accessibility: The system is particularly beneficial for novice users who find traditional CAD interfaces cumbersome. By focusing on sketch inputs, it democratizes access to precision modeling.

Performance Evaluation

In user studies, Sketch2CAD demonstrated significant ease of use, allowing participants with limited CAD experience to reproduce complex models effectively. Qualitative feedback emphasized the system's capacity to interpret approximate sketches into accurate procedures reliably.

Conclusion

Sketch2CAD provides a versatile and robust approach to CAD modeling via sketch input, merging intuitive sketching with the precision of CAD systems. This makes it a valuable tool for users across various skill levels, from novice to professional designers. The potential extension to support an even broader array of operations and more sophisticated geometrical contexts promises a wide range of applications in design and engineering. Future work will focus on expanding the operation library, incorporating semantic understanding, and refining the integration with traditional CAD workflows.