MAS.863/4.140/6.9020
How To Make (almost) Anything
Final Project: 3D Incense Diffuser with Genji Mon Pattern
Awu Chen
2025
Original HTMAA Pitch: Scent Sequencer
This was my original pitch for the HTMAA (How To Make (almost) Anything) class - the Scent Sequencer concept that evolved into the digital nose project
Project Overview
This final project combines traditional Japanese incense ritual with modern digital fabrication and advanced sensor technology to create a digital nose. The project integrates the Genji Mon (源氏紋) pattern from Week 1's CAD work with 3D printing techniques from Week 4, and incorporates a laser-induced graphene (LIG) sensor array for smell detection and AI-powered scent analysis.
Project Vision
• Cultural Integration: Combining traditional Japanese design with AI VOC gas detection
• Functional Art: Creating a beautiful object that serves a practical purpose
• Pattern Application: Using the Genji Mon geometry to create ventilation and aesthetic features
• Material Exploration: Testing different materials (resin, ABS) for optimal functionality
• Digital Craftsmanship: Demonstrating mastery of CAD design and 3D printing techniques
Historical Foundation: Genji Mon Pattern
Cultural Significance
The Genji Mon (源氏紋) is a traditional Japanese family crest pattern with deep historical significance. The pattern represents the Genji clan, one of the most influential families in Japanese history, particularly during the Heian period (794-1185).
Historical Context:
•
Origins: The Genji Mon pattern dates back to the Heian period and is associated with the Minamoto clan (源氏)
•
Cultural Impact: Featured prominently in "The Tale of Genji" (源氏物語), considered the world's first novel
•
Design Elements: The pattern typically features geometric arrangements representing different aspects of court life
•
Modern Usage: Still used in traditional Japanese arts, textiles, and ceremonial contexts
Historical Images
Historical Genji Mon pattern showing traditional geometric design elements and cultural significance
Variations of Genji Mon patterns demonstrating the geometric complexity and artistic tradition
Mathematical analysis of Genji Mon pattern showing the underlying geometric principles and bell number relationships
Final Project Concept
Detailed sketch for Genji 3D incense holder showing design development and functional considerations
3D rendering of incense diffuser concept incorporating Genji Mon patterns
3D Printing Implementation
Resin Printing (Formlabs Form 3+)
•
Material: Standard Clear Resin
•
Wall Thickness: 0.1mm (ultra-thin for maximum detail)
•
Advantages: High resolution, smooth surface finish, intricate detail preservation
•
Applications: Display pieces, detailed prototypes
Genji-ko pattern printing on Formlabs with support structures for overhangs
Ultra-thin genji-ko pattern in resin showing intricate fold details and 0.1mm wall thickness
Exfoliated graphene structure used in machine-learning-enabled electronic nose systems. Image from: Machine-learning-enabled graphene-based electronic nose for microplastic detection
Next-Generation AI Electronic Nose Technology
The final project incorporates advanced sensor technology inspired by research from Korea's Daegu Gyeongbuk Institute of Science and Technology (DGIST), which developed a porous laser-induced graphene (LIG) sensor array that functions as a "next-generation AI electronic nose." This technology enables the system to distinguish scents like the human olfactory system does and analyze them using artificial intelligence.
How LIG Sensors Work
The LIG sensor array converts scent molecules into electrical signals and trains AI models on their unique patterns. This technology holds great promise for applications in personalized health care, the cosmetics industry, and environmental monitoring. While conventional electronic noses (e-noses) have already been developed and used in areas such as food safety and gas detection in industrial settings, they struggle to distinguish subtle differences between similar smells or analyze complex scent compositions. For instance, distinguishing among floral perfumes with similar notes or detecting the faint odor of fruit approaching spoilage remains challenging for current systems. This gap has driven demand for next-generation e-nose technologies with greater precision, sensitivity, and adaptability.
Biological Inspiration: Combinatorial Coding
The research team was inspired by the biological mechanism known as combinatorial coding, in which a single odorant molecule activates multiple olfactory receptors to create a unique pattern of neural signals. By mimicking this principle, the sensors respond to scent molecules by generating distinct combinations of electrical signals. The AI system learns these complex signal patterns to accurately recognize and classify a wide variety of scents, resulting in a high-performance artificial olfaction platform that surpasses existing technologies.
Fabrication and Performance
The novel electronic nose uses a laser to process graphene and incorporates a cerium oxide nano catalyst to create a sensitive sensor array. This single-step laser fabrication method eliminates the need for complex manufacturing equipment and enables high-efficiency production of integrated sensor arrays. In performance tests, the device successfully identified nine fragrances commonly used in perfumes and cosmetics, with over 95% accuracy. It could also estimate the concentration of each scent, making it suitable for fine-grained olfactory analysis.
LIG Sensor Development for This Project
Laser-induced graphene sensor pattern test showing the porous graphene structure created by laser processing. This pattern forms the foundation of the sensor array that will detect scent molecules.
LIG sensor resistor testing demonstrating the electrical response characteristics of the graphene sensor array. The sensor array generates distinct electrical signal patterns when exposed to different scent molecules, which will be analyzed by AI to identify and classify scents.
LIG Fabrication Methodology
The LIG sensor fabrication process for this project builds upon the laser-induced graphene research and documentation by
Wedyan Babatain from the HTMAA 2022 class. Her comprehensive work on LIG fabrication using CO2 lasers at MIT CBA provides the foundational methodology for creating the porous graphene sensor arrays used in this project.
The fabrication process involves laser irradiation of polyimide (Kapton) sheets to create the conductive graphene structures that form the basis of the sensor array. The single-step laser fabrication method eliminates the need for complex manufacturing equipment and enables high-efficiency production of integrated sensor arrays.
For detailed information on LIG fabrication techniques, parameters, and applications, see:
Wedyan Babatain's LIG Documentation
Integration with Incense Diffuser
The LIG sensor array will be integrated into the incense diffuser to detect and analyze the scents being released. As different incense types are burned, the sensor array will:
• Detect Scent Molecules: Convert volatile organic compounds into electrical signals
• Generate Signal Patterns: Create unique electrical signatures for each scent type
• AI Classification: Use machine learning to identify and classify scents with high accuracy
• Concentration Estimation: Measure the intensity of scents being released
• Real-time Monitoring: Provide continuous feedback about the incense diffusion process
This integration transforms the incense diffuser from a passive device into an intelligent system that can recognize, analyze, and provide feedback about the scents it produces, creating a bridge between traditional incense ceremony practices and modern sensor technology.
Ideation 3: Tangible Scent Interface with MultiModal AI
Concept Overview
Inspired by Paul's MultiModal AI lecture, Ideation 3 explores a Tangible Scent Interface that creates a direct visual-to-scent mapping using Vision-Language Models (VLMs) and gaze tracking. This approach transforms images with strong scent associations into interactive scent experiences, where visual attention directly controls olfactory output.
Core Technology: Vision-Language Model Integration
The system uses a VLM to isolate objects from an image that have strong scent associations. Each pixel of the image is mapped to a corresponding pixel of scent, creating a direct spatial relationship between what the user sees and what they smell. When the user's gaze locks onto a specific pixel or object, the visuals transform to focus on the corresponding pixels (e.g., all the flowers in the image), while everything else fades away.
Gaze-Locked Visual Transformation
The interaction model works as follows:
• Initial State: The system displays an image with multiple scent-associated elements (flowers, fruits, spices, etc.)
• Gaze Detection: Eye-tracking technology monitors where the user is looking
• Object Isolation: The VLM identifies all similar objects in the image based on the gazed pixel
• Visual Transformation: The image transforms - all matching objects (e.g., all flowers) become highlighted and focused, while other elements fade away
• Scent Emission: The corresponding scent pixel activates, releasing the appropriate aroma for the identified object
• Audio Fading: Optionally, audio elements may also fade away to enhance the focus on the visual-scent connection
Pixel-to-Pixel Scent Mapping
Each pixel in the visual image is mapped to a corresponding pixel in the scent output array. When the user's gaze focuses on a pixel showing a flower, for example:
1. The VLM identifies that pixel as belonging to a "flower" object class
2. All other pixels in the image that belong to the same class are found
3. These pixels are visually enhanced while others fade
4. The corresponding scent pixels are activated to emit the flower's aroma
5. The user experiences a synchronized visual and olfactory experience
This creates a tangible, spatial relationship between vision and smell, making scent a visible, interactive medium.
Idea 3 Exploration
Initial concept visualization for the Tangible Scent Interface showing the integration of visual images with scent mapping
Exploration of the visual-scent mapping concept, demonstrating how gaze tracking and object isolation work together to create synchronized sensory experiences
AI Model Architecture
AI model architecture for the Tangible Scent Interface, showing how Vision-Language Models integrate with scent emission systems to create pixel-to-pixel visual-scent mapping
Technical Implementation
The system integrates multiple technologies:
• Vision-Language Model: Processes images to identify scent-associated objects
• Eye Tracking: Monitors user gaze position in real-time
• Spatial Mapping: Creates pixel-to-pixel correspondence between visual and scent arrays
• LIG Sensor Array: Detects and confirms emitted scents
• Display Output: Renders transformed visuals with enhanced focus on selected objects
• Scent Emitter Array: Releases appropriate aromas corresponding to visual pixels
Applications and Implications
This ideation opens up new possibilities for:
• Interactive Art: Creating artworks where viewing and smelling are intrinsically linked
• Educational Tools: Teaching scent identification through visual reinforcement
• Therapeutic Applications: Using gaze-controlled scent therapy for relaxation or focus
• Cultural Experiences: Recreating historical or cultural scent-visual associations
• Accessibility: Making scent experiences more tangible and controllable for users
Connection to Digital Nose Project
Ideation 3 extends the digital nose concept by adding bidirectional interaction: not only does the system detect scents (input), but it also creates scents based on visual attention (output). This creates a complete sensory feedback loop where the user's visual attention directly controls olfactory output, and the system confirms scent emission through its detection capabilities. The LIG sensor array can verify that the correct scent is being emitted, creating a closed-loop system that learns and improves over time.
References and Inspirations
This ideation draws inspiration from
Hsin-Chien Huang's "The Moment We Meet" interactive installation, which uses split-flap displays to form a 10x10 matrix of faces that can be controlled independently, creating endless combinations. The work explores how emotions and expressions spread through visual transformation, similar to how Ideation 3 maps visual attention to scent emission. Huang's use of pixel-like displays that independently transform based on interaction provides a foundation for understanding how individual sensory pixels can be mapped and controlled in a spatial array.
Reference: "The Moment We Meet" by Hsin-Chien Huang
"The Moment We Meet" interactive installation showing the 10x10 matrix of split-flap displays that can be controlled independently. Each display acts like a pixel that can transform based on interaction, providing inspiration for the pixel-to-pixel mapping concept in Ideation 3.
Detail view of "The Moment We Meet" showing how individual displays within the matrix create different facial expressions. This demonstrates how independent pixel-like elements can create complex patterns through coordinated transformation.
Another view of "The Moment We Meet" installation showing the spatial array of independently controllable displays. This spatial array concept directly informs Ideation 3's approach to mapping visual pixels to scent pixels in a coordinated spatial system.
Ideation 4: Ambient Scent Phone Booth
Concept Overview
Ideation 4 explores an ambient scent phone booth using a rotary phone as the central design metaphor. This concept is a continuation of prior work, particularly the "teleabsence moment" around a missed call, where communication occurs through scent and memory rather than direct conversation. The phone booth creates an intimate space for asynchronous scent-based communication, bridging physical distance through olfactory experiences.
The Core Concept: Missed Call, Lingering Scent
The installation centers on the theme of always missing the call. The phone is constantly ringing, but when a person picks it up, a voice message plays: "We're sorry to have missed your call." This sets up a communication loop where the user is perpetually too late, but the experience continues through scent and memory rather than conversation.
Design Metaphor and Space
• Rotary Phone: An old rotary phone serves as the central interaction point, evoking nostalgia and the concept of missed connections
• Phone Booth Environment: A small chamber or phone booth space creates an intimate, private environment for the scent experience
• Hardware Integration: The scent emitter device is integrated directly into the phone itself, possibly where the speaker/earpiece would be, transforming the phone into a smelling device
• Ambient Environment: The scent becomes the ambient environment, replacing the traditional dial tone with meaningful, non-existent places - connecting to the "phone call to heaven" concept
Interaction Flow
1. Constant Ringing: The phone is always ringing, creating anticipation and the feeling of being called
2. Picking Up the Phone: User answers the phone to hear "We're sorry to have missed your call"
3. Sequential Scent Delivery: A scent is delivered sequentially for the user to experience and smell
4. Memory Recording: The user is prompted: "When you finished smelling, please leave a memory and hang up"
5. Voice Memo Creation: The user records a voice memo (a "memory") in response to the scent they just experienced
6. Real-time Scent Generation: The voice memo is translated in real-time into a new scent using a real-time scent generator model
7. Scent as Response: The generated scent acts as a signal that the user's message/memory has been received, even though they missed the original call
Technical Implementation
The system integrates multiple technologies:
• Real-time Scent Generator: A system that converts voice memos into scents in real-time, building on existing scent generation models
• Hardware Integration: Scent emitter/device integrated directly into the rotary phone hardware, possibly using 3D printing for custom fabrication
• Voice Recording System: Captures user voice memos in response to scents
• Sequential Scent Delivery: Controls the timing and sequence of scent emissions
• LIG Sensor Array: Can detect and verify the scents being emitted, creating a closed-loop system
Asynchronous Scent Conversation
The concept extends to having similar setups in different locations, facilitating asynchronous conversations through scent. Users at one location can leave a voice memory, which generates a scent. This scent is then experienced by users at another location, who can leave their own memory in response, creating a scent-based conversation loop. The lingering trace of previous conversations becomes part of the ambient environment of the phone booth.
Collaboration and Development
This ideation emerged from collaborative discussions focused on:
• Real-time Scent Generation: Developing systems to deliver scents in real-time based on voice input, which can also provide more data for scent model training
• Hardware Fabrication: Using 3D printing and fabrication facilities to integrate scent emitters into the rotary phone design
• Model Development: Building and refining scent generation models that can translate voice memories into olfactory experiences
• Asynchronous Communication: Creating networked scent experiences that bridge physical distance
Ambient Scent Phone Booth Concept
Concept visualization for the Ambient Scent Phone Booth, showing the rotary phone design integrated with scent emission technology in an intimate booth environment
Connection to Digital Nose Project
Ideation 4 extends the digital nose project by creating a complete communication system based on scent. While the digital nose detects and identifies scents, Ideation 4 creates scents as responses to human input (voice memories). The LIG sensor array can verify the generated scents, creating a bidirectional system: the booth generates scents from voice, and the digital nose detects and identifies what was generated. This creates a closed-loop scent communication system that learns and improves from the interaction data.
The concept connects traditional communication technologies (the phone) with cutting-edge scent generation and detection, creating a new form of asynchronous communication that operates entirely through smell and memory. The "teleabsence moment" - the feeling of a missed call - becomes a design element that drives the interaction, making absence and missed connections the foundation for a new type of presence through scent.
Manufacturing the Scent Diffuser: Based on Scentrealm Neck-Scent-Player Technology
Technology Reference
The scent diffuser manufacturing approach draws inspiration from the
Scentrealm neck-scent-player device, which provides a wearable scent delivery system with Bluetooth control capabilities. This commercial technology demonstrates a practical approach to portable, programmable scent emission that can be adapted for stationary installation use.
The Scentrealm neck-scent-player device showing the wearable scent delivery system with 12 scent cartridges and Bluetooth control capabilities
Hardware Adaptation: 12-Cartridge Airflow System
The neck-scent-player technology uses a sophisticated multi-cartridge system with airflow control:
• 12 Scent Cartridges: The device incorporates 12 individual scent cartridges, each containing a different aroma, allowing for a wide variety of scent combinations
• Airflow Emission: The system sends airflow through selected cartridges to emit scents into the air. By directing air through specific cartridges, different scents can be released individually or in combination
• Bluetooth Control: The device uses Bluetooth SDK for wireless control of which cartridges to activate and when, enabling programmable and sequential scent emission
• Portable Form Factor: The neck-worn design demonstrates how scent emitters can be integrated into small, contained spaces while maintaining functionality
• Programmable Timing: Bluetooth control enables precise timing and sequencing of scent emissions, creating complex scent patterns and experiences
Alternative Approach: Ultrasonic Conductor
While the Scentrealm device uses an airflow-based approach with 12 cartridges, an alternative approach using ultrasonic conductors could offer several advantages:
• Ultrasonic Atomization: Ultrasonic conductors use high-frequency vibrations to atomize liquid scents into fine mists, creating more uniform scent distribution without requiring airflow mechanisms
• Faster Response Time: Ultrasonic systems can respond more quickly than airflow systems, enabling more immediate scent delivery in response to sensor input or user interaction
• Reduced Mechanical Complexity: Ultrasonic systems eliminate the need for air pumps and airflow channels, potentially simplifying the hardware design
• Energy Efficiency: Ultrasonic atomization typically requires less power than airflow systems, which could be advantageous for both portable and stationary applications
• Precise Control: Ultrasonic systems allow for very precise control of scent intensity through modulation of vibration frequency and amplitude
• Multiple Cartridge Compatibility: Ultrasonic conductors can be integrated with multiple scent cartridges, similar to the 12-cartridge system, but using vibration-based emission rather than airflow
Comparison: Airflow vs Ultrasonic Approaches
For the incense diffuser project, both approaches have merit:
Airflow Approach (Scentrealm Method):
• Proven technology with existing commercial implementation
• Familiar mechanism (similar to traditional air fresheners)
• Requires air pumps and channels
• Good for larger scent volumes
Ultrasonic Conductor Approach:
• Potentially faster response times
• Simpler mechanical design (no pumps needed)
• More energy efficient
• Better for precise, controlled emission
• May offer better integration with Genji Mon pattern ventilation
For this project, the ultrasonic conductor approach may be preferable because it offers faster response times for real-time scent generation (important for Ideation 4's voice-to-scent pipeline) and could integrate more elegantly with the Genji Mon geometric pattern's ventilation pathways.
Adaptation for Incense Diffuser Application
For the incense diffuser project, the neck-scent-player technology can be adapted in several ways:
• Stationary Installation: Adapting the portable neck-worn design for fixed installation within the Genji Mon incense diffuser housing
• Integration with LIG Sensors: Combining the scent emission technology with the LIG sensor array to create a closed-loop system that detects and responds to scents
• Bluetooth Integration: Using the Bluetooth SDK to enable remote control and programming of scent sequences
• Multi-Scent Array: Creating an array of scent emitters similar to the neck device's cartridge system, but integrated into the diffuser housing
• Voice-to-Scent Pipeline: Leveraging the programmable control system to implement real-time scent generation from voice input (for Ideation 4's phone booth application)
Manufacturing Considerations
Key manufacturing considerations when adapting this technology:
• Housing Integration: Designing the Genji Mon diffuser housing to accommodate scent emitters while maintaining aesthetic integrity
• Ventilation Design: Using the Genji Mon pattern's geometric structure to create controlled scent diffusion pathways
• Modular Components: Creating a modular system where scent cartridges or emitters can be easily replaced or upgraded
• Power Management: Adapting power requirements from portable battery operation to stationary power supply
• Control System: Integrating Bluetooth control with microcontroller systems (SAMD21) for coordinated operation with LIG sensors
Technical Implementation Path
The implementation path involves:
1. Hardware Study: Analyzing the neck-scent-player's hardware design and control mechanisms
2. SDK Integration: Utilizing or adapting the Bluetooth SDK for custom control applications
3. Housing Design: Integrating scent emitters into the 3D-printed Genji Mon diffuser housing
4. Sensor Integration: Connecting scent emission control with LIG sensor detection for feedback loops
5. Control Software: Developing software that coordinates scent emission, sensor detection, and user interaction
Advantages of This Approach
Building on existing commercial technology provides several advantages:
• Proven Technology: Leveraging tested and reliable scent emission mechanisms
• Open Source SDK: Access to Bluetooth SDK for custom development and integration
• Modular Design: Understanding how to create replaceable, maintainable scent systems
• Programmable Control: Learning how to sequence and time scent emissions precisely
• User Experience Insights: Understanding how people interact with wearable scent technology
Learning Outcomes
This project demonstrates the integration of traditional craft knowledge with modern digital fabrication techniques. Key learning outcomes include:
• Cultural Design Integration: Successfully incorporating traditional patterns into modern functional objects
• Material Science: Understanding material properties and their impact on design decisions
• Digital Fabrication: Mastering multiple 3D printing technologies and their applications
• Design Process: Developing systematic approach from concept to production
• Cross-Cultural Innovation: Bridging traditional Japanese design with contemporary manufacturing
Attachments
Links