Proceedings of the 2020 DigitalFUTURES : : The 2nd International Conference on Computational Design and Robotic Fabrication (CDRF 2020).

• Author: Yuan, Philip F.
• TeilnehmendeR: Yao, Jiawei.; Yan, Chao.; Wang, Xiang.; Leach, Neil.
• Place / Publishing House: Singapore : : Springer Singapore Pte. Limited,, 2021.; ©2021.
• Year of Publication: 2021
• Edition: 1st ed.
• Language: English
• Physical Description: 1 online resource (327 pages)

Table of Contents

• Intro
• Preface
• Committees
• Honorary Advisors
• Organization Committees
• Scientific Committees (list by surname)
• Contents
• Machine Thinking
• Machinic Phylum and Architecture
• 1 Nips and Bites
• 2 Ducks and Rabbits
• References
• Pipes of AI - Machine Learning Assisted 3D Modeling Design
• 1 Principle of CNN
• 1.1 Principle and Applications of Style Transfer
• 1.2 Project Goal
• 2 2D Image Representation of 3D Volume
• 2.1 The Effect of Style Weight in Style Transfer
• 2.2 Transformation of Image to Geometry
• 2.3 Algorithm Analysis of Geometry Generation Between Adjacent Layers
• 3 Result of Section Plans
• 3.1 Result of Perspective View
• 4 Conclusion
• References
• Developing a Digital Interactive Fabrication Process in Co-existing Environment
• 1 Introduction
• 2 Related Work
• 2.1 Fabrication Process of Maker
• 2.2 Towards Co-existing Environment
• 2.3 Automation Digital Fabrication Tools
• 2.4 Summary
• 3 Methodology
• 4 The Experiment
• 5 Conclusions
• References
• Real-Time Defect Recognition and Optimized Decision Making for Structural Timber Jointing
• 1 Introduction
• 2 Defect Recognition and Removal
• 2.1 Pre-process the Image for Segmentation
• 2.2 Preparation of the Classifier
• 2.3 Preparation of the Classifier
• 3 Decision Making for Joining Timber Segments
• 4 User Interface
• 5 Discussion and Future Development
• 6 Conclusion
• References
• On-Site BIM-Enabled Augmented Reality for Construction
• 1 Introduction
• 1.1 Motivation
• 1.2 Related Work
• 1.3 Our Solution
• 2 AR Application
• 2.1 Model Overlay Using Augmented Reality
• 2.2 Model Interaction as Query System
• 2.3 Abstraction of Drawings
• 2.4 Additional Features
• 3 Data Pipeline
• 3.1 BIM Pre-processing, Custom Parameter Creation and Population.
• 3.2 Construction Document Export and Metadata Post-processing
• 4 Unity Reflect
• 5 Results
• 5.1 Case Study/User Testing
• 5.2 Future Development
• 5.3 Connections
• References
• Recycling Construction Waste Material with the Use of AR
• 1 Introduction
• 2 Aims
• 3 Method
• 3.1 Method 1 | Mass Customisation and Working to a Fixed Digital Model
• 3.2 Mass Customized Aggregation Geometry
• 3.3 Holographic Part Nesting
• 3.4 Mixed Reality Interface
• 3.5 Jointing
• 3.6 Fabrication and Fixing Methods
• 3.7 Method 2 | Working to a Flexible Digital Model
• 3.8 Results
• 4 Discussion
• 4.1 Future Development
• References
• Growing Shapes with a Generalised Model from Neural Correlates of Visual Discrimination
• 1 Introduction
• 2 Methods
• 3 Results
• 4 Conclusion
• References
• Cyborgian Approach of Eco-interaction Design Based on Machine Intelligence and Embodied Experience
• 1 Tracing Cyborgian Theory and Embodied Cognition
• 1.1 A Hybrid of Part Clock Part Swarm [10]
• 1.2 The Importance of the Presence and the Bodily Experience
• 2 How Cyborgian Approach Activates Plants?
• 2.1 How They Sense
• 2.2 How They Think
• 2.3 How They Actuate
• 3 How Cyborgian Approach Encourages Human Participation?
• 3.1 Experience Level
• 3.2 Experience Assessment
• 4 Design an Interactive Outdoor Environment
• 4.1 Challenges and Opportunities of Outdoor Interaction
• 4.2 A Cyborgian Eco-interaction Design Model
• 5 Conclusion and Outlook
• References
• Machine Seeing
• A Large-Scale Measurement and Quantitative Analysis Method of Façade Color in the Urban Street Using Deep Learning
• 1 Introduction
• 2 Literature Review
• 2.1 Urban Color Planning
• 2.2 Façade Color Measurements
• 2.3 Quantitative Analysis of Visual Quality in Urban Street
• 3 Methodology
• 3.1 Study Area and Workflow
• 3.2 Street View Data Acquisition.
• 3.3 Building Façade Segmentation and Data Cleaning
• 3.4 Façade Color Calculation
• 4 Results
• 5 Discussion and Conclusion
• References
• Suggestive Site Planning with Conditional GAN and Urban GIS Data
• 1 Introduction
• 2 Related Works
• 3 Methodology
• 3.1 Data Acquisition and Feature Engineering
• 3.2 Machine Learning
• 3.3 Visualization
• 4 Case Study: Taking Boston as Example
• 4.1 Data Acquisition and Feature Engineering
• 4.2 Model Building and Training
• 4.3 Results and Visualization
• 5 Summary
• References
• Understanding and Analyzing the Characteristics of the Third Place in Urban Design: A Methodology for Discrete and Continuous Data in Environmental Design
• 1 Background
• 2 Methodology
• 2.1 Data and Data Structure for Manipulation
• 2.2 Pixel Structure for Continuous Data and Blending Data with Neighbors
• 2.3 Graph Structure for Discrete Data
• 3 Case Study Implementation
• 3.1 Site Selection
• 3.2 Parse Third Place Data and Visualization
• 3.3 Generate Data Structures and Inspect with Visualizations
• 3.4 Comparisons and Results
• 4 Discussion
• 5 Conclusion
• 6 Future Work
• References
• Sensing the Environmental Neighborhoods
• 1 Introduction
• 1.1 Sensing Kit Design
• 1.2 Case Study
• 1.3 Summary
• References
• A Performance-Based Urban Block Generative Design Using Deep Reinforcement Learning and Computer Vision
• 1 Introduction
• 2 Methodology
• 2.1 DRL Based Generative Design Framework
• 2.2 DDPG Agent
• 2.3 Hough Transform
• 3 Case Study
• 3.1 Observation, Action and Reward
• 3.2 Site Information
• 4 Results
• 5 Conclusions and Future Work
• References
• The Development of 'Agent-Based Parametric Semiology' as Design Research Program
• 1 Theory Background
• 2 Why We Need Agent-Based Life-Process Crowd Simulation.
• 3 The Intelligence Upgrading of Agent-Based Crowd Simulation
• 3.1 Crowd Behaviour Pattern Analysis
• 3.2 Intelligent Agents
• 3.3 Semantic Virtual Environment
• 4 Quantitative Analysis, Evaluation, and Optimization
• 4.1 Methodology and Toolset
• 4.2 Scenario and Example
• 5 Discussion
• References
• Machine Learning
• Machine Learning Aided 2D-3D Architectural Form Finding at High Resolution
• 1 Introduction
• 2 Relative Work
• 3 Method
• 4 Results
• 4.1 Training Data Preparation
• 4.2 Main Network Training
• 4.3 Multiple Network Training
• 5 Conclusion
• References
• Exploration of Campus Layout Based on Generative Adversarial Network
• 1 Introduction
• 2 Related Work in the Field of Architectural Layout
• 3 Methods
• 4 Experimental Results and Analysis
• 5 Discussion
• Appendix
• References
• A Preliminary Study on the Formation of the General Layouts on the Northern Neighborhood Community Based on GauGAN Diversity Output Generator
• 1 Introduction
• 2 Research
• 2.1 AI Application in Architecture
• 2.2 Deep Learning Architectural Plan Generator Application
• 3 Methodology
• 3.1 GauGAN
• 3.2 Step Training
• 4 Machine Learning for the General Layout Shapes of the Northern Neighborhoods in China
• 4.1 Morphological Analysis
• 4.2 Data Conversion
• 4.3 Model Architecture
• 4.4 Vectorization and 3D Procedural Modeling
• 4.5 Experiment Result
• 5 Conclusion
• 5.1 GauGAN Is More in Line with Architectural Design Needs Than Pix2pix (Pix2pixHD)
• 5.2 The Use of Step Training Can Improve the Clarity of Generated Results and Allow the Later Vectorization to Be More Convenient
• References
• Artificial Intuitions of Generative Design: An Approach Based on Reinforcement Learning
• 1 Introduction
• 1.1 Contemporary Algorithmic Generative System
• 1.2 Artificial Intuitions
• 2 Background.
• 2.1 Machine Learning with Generative Design
• 2.2 Reinforcement Learning
• 3 Methodology
• 3.1 Intuitive Random Walk Formation
• 3.2 RL Actions Definition
• 3.3 RL Observations Definition
• 3.4 RL Reward Definition
• 4 Discussions
• 4.1 Training Process and Outcomes
• 4.2 Further Research
• 5 Conclusions
• References
• Collection to Creation: Playfully Interpreting the Classics with Contemporary Tools
• 1 Introduction: Generations to Generative
• 2 Process: Beyond Codified Interaction
• 3 User Analysis
• 4 Media Creation
• 5 Synthetic Text Descriptions
• 6 Thoughts
• 7 Conclusion
• References
• embedGAN: A Method to Embed Images in GAN Latent Space
• 1 Introduction
• 2 Related Work
• 2.1 Regenerating Data in GAN
• 2.2 GAN Latent Walk
• 3 Method
• 3.1 Principle
• 3.2 Architecture
• 3.3 Training Details
• 4 Application
• 5 Evaluation
• 6 Conclusion
• References
• Research on Architectural Form Optimization Method Based on Environmental Performance-Driven Design
• 1 Introduction
• 2 Performance-Driven Design and Its Advantages
• 2.1 Performance-Driven Design Theory
• 2.2 Performance-Driven Design Advantages Compared with Bionic Form Design
• 3 Performance-Driven Architectural Form Optimization Method
• 3.1 Combined with Parametric Design
• 4 Form Optimization Simulation Process Establishment
• 5 Design Practice
• 5.1 Project Background
• 5.2 Design Parameters Selection and Numerical Constraint
• 5.3 Setting Simulation Parameters
• 5.4 Form Optimization Process Diagram
• 5.5 Result Analysis
• 6 Conclusion
• References
• Optimization and Prediction of Design Variables Driven by Building Energy Performance-A Case Study of Office Building in Wuhan
• 1 Introduction
• 2 Research Method
• 2.1 Research Objectives
• 2.2 Research Method.
• 2.3 Multi-Island Genetic Algorithm (MIGA) and Radial Basis Functions Artificial Neural Networks (RBF-ANNs).