Mobiot: Augmenting Everyday Objects into Moving IoT Devices Using 3D Printed Attachments Generated by Demonstration

Document Title

Mobiot: Augmenting Everyday Objects into Moving IoT Devices Using 3D Printed Attachments Generated by Demonstration

Document Information

• Research Area: Human-Computer Interaction (HCI), Personal Robotics, Internet of Things (IoT), 3D Printing, Home Automation
• Keywords: Personal fabrication, home automation, motion planning, human-computer interaction, smart IoT devices, 3D printing, task design

Research Background and Problem Statement

• Identified Issues or Challenges:
  • General users face difficulties in adding mobility functionalities to everyday objects using existing tools.
  • Current robotic platforms are often expensive and designed exclusively for experts, failing to meet the personalization needs of everyday users.
  • Capturing real-world design requirements and creating personalized robotic devices for different use scenarios is a complex task involving geometric, physical, and electronic considerations.
  • There is a lack of user-friendly solutions for path execution and motion programming.
• Significance:
  • Everyday objects with mobility capabilities can simplify household tasks, such as automatic plant watering and self-service kitchen assistance, and enhance home accessibility (e.g., providing services for the elderly).
  • Automated, easy-to-manufacture tools can inspire user creativity and promote the adoption of personal robotics.
• Research Motivation and Related Work:
  • Recent studies focus on personal fabrication and end-user programming to empower users to modify their surrounding objects.
  • Existing systems emphasize motion generation for robotic components or localized motion design but fail to address the dynamic capability needs of complete objects.
  • The core challenge lies in designing low-barrier, highly flexible tools that enable users to design and operate personal robots without technical expertise.

Solution

• Proposed Solution:
  • Develop an end-to-end toolkit named “Mobiot” to automatically generate attachments for adding mobility capabilities to objects based on single motion demonstrations.
  • The toolkit extracts information from motion trajectories demonstrated by smartphone users, including geometric requirements, motion implementation mechanisms, and motion planning.
  • Based on the captured data, it generates 3D printable models, electronic component lists, and relevant code to imbue traditional objects with intelligent mobility capabilities.
• Innovations:
  • One-stop design-to-manufacturing process: Users can easily generate models and control programs using images and motion demonstrations.
  • Low barrier, high flexibility: Provides options for designing new mechanisms or modifying existing ones, supporting various motion types and task combinations.
  • Task reusability: Manufactured mechanisms can be manually adjusted or reused for other objects.
• Implementation Steps and Techniques:
  1. Capturing Design Requirements: Motion demonstration data is collected via smartphone IMU sensors; geometric information is extracted from target objects using image recognition tools (YoloV4).
  2. Interactive Design and Task Planning: Users can select and adjust automatically detected motion blocks through the interface to design motion sequences.
  3. Output Generation: Based on motion and geometric requirements, 3D printable parts, electronic component lists, and control code are automatically generated.

Research Outcomes

• Specific Results:
  • The Mobiot toolkit was validated on eight everyday objects with different mobility functions, including an automatic plant watering pot, autonomous trash bin, and automatic cooking assistant.
  • Users were able to add motion functionalities such as “drifting,” “rotating,” and “lifting” to objects through simple motion demonstrations, while adjusting task complexity and parameters via the interface.
• Comparison and Advantages:
  • Compared to existing standalone robotic platforms, Mobiot significantly reduces manufacturing costs and technical barriers.
  • Provides flexibility for personalized design and task adjustments.
• Experimental or Evaluation Results:
  • Technical Validation: The motion precision of mechanisms was evaluated, including position estimation errors, rotation angles, and height dimension deviations. Overall precision performed well in daily user scenarios.
  • Expert Interviews: Experts praised the tool’s ease of use, particularly the innovative approach of capturing design requirements via smartphones, and highlighted potential for future enhancements such as integrating sensors and dynamic adaptation features.
• Limitations and Future Directions:
  • Current design supports limited motion types (three types); future work could expand to more complex motions.
  • Complex mechanical configurations, such as flipping motions, are not yet supported and require updates to hardware design capabilities.
  • Lack of real-time obstacle detection and position feedback functions; integrating IMU or sensors is recommended for improvement.
  • The user interface could be further optimized to support 3D scene composition and more dynamic motion planning.

Through the development of the Mobiot toolkit, this paper demonstrates the significant potential of providing tools for ordinary users to create personal robots, while highlighting directions for future improvements, such as reducing manufacturing costs, enhancing sensing capabilities, and expanding supported motion types.