Table of Contents
Fetching ...

SplatOverflow: Asynchronous Hardware Troubleshooting

Amritansh Kwatra, Tobias Weinberg, Ilan Mandel, Ritik Batra, Peter He, Francois Guimbretiere, Thijs Roumen

TL;DR

SplatOverflow addresses the lack of scalable end-user hardware troubleshooting by introducing a boundary-object workflow that binds a user-generated 3D Gaussian Splat to a CAD model, enabling asynchronous guidance from remote maintainers. The approach combines scene capture, CAD-aligned registration, and a gesture-driven interaction model to create editable, recontextualizable instructions anchored to the hardware in situ. The paper demonstrates design, workflow, and a usability study showing non-experts can generate scenes and follow guidance for common hardware issues, with strong usability scores and scalable indexing of solutions. This work offers a practical path toward scalable, community-driven hardware maintenance for small producers and maker communities, reducing the support burden while enhancing knowledge sharing across devices and contexts.

Abstract

As tools for designing and manufacturing hardware become more accessible, smaller producers can develop and distribute novel hardware. However, processes for supporting end-user hardware troubleshooting or routine maintenance aren't well defined. As a result, providing technical support for hardware remains ad-hoc and challenging to scale. Inspired by patterns that helped scale software troubleshooting, we propose a workflow for asynchronous hardware troubleshooting: SplatOverflow. SplatOverflow creates a novel boundary object, the SplatOverflow scene, that users reference to communicate about hardware. A scene comprises a 3D Gaussian Splat of the user's hardware registered onto the hardware's CAD model. The splat captures the current state of the hardware, and the registered CAD model acts as a referential anchor for troubleshooting instructions. With SplatOverflow, remote maintainers can directly address issues and author instructions in the user's workspace. Workflows containing multiple instructions can easily be shared between users and recontextualized in new environments. In this paper, we describe the design of SplatOverflow, the workflows it enables, and its utility to different kinds of users. We also validate that non-experts can use SplatOverflow to troubleshoot common problems with a 3D printer in a usability study. Project Page: https://amritkwatra.com/research/splatoverflow.

SplatOverflow: Asynchronous Hardware Troubleshooting

TL;DR

SplatOverflow addresses the lack of scalable end-user hardware troubleshooting by introducing a boundary-object workflow that binds a user-generated 3D Gaussian Splat to a CAD model, enabling asynchronous guidance from remote maintainers. The approach combines scene capture, CAD-aligned registration, and a gesture-driven interaction model to create editable, recontextualizable instructions anchored to the hardware in situ. The paper demonstrates design, workflow, and a usability study showing non-experts can generate scenes and follow guidance for common hardware issues, with strong usability scores and scalable indexing of solutions. This work offers a practical path toward scalable, community-driven hardware maintenance for small producers and maker communities, reducing the support burden while enhancing knowledge sharing across devices and contexts.

Abstract

As tools for designing and manufacturing hardware become more accessible, smaller producers can develop and distribute novel hardware. However, processes for supporting end-user hardware troubleshooting or routine maintenance aren't well defined. As a result, providing technical support for hardware remains ad-hoc and challenging to scale. Inspired by patterns that helped scale software troubleshooting, we propose a workflow for asynchronous hardware troubleshooting: SplatOverflow. SplatOverflow creates a novel boundary object, the SplatOverflow scene, that users reference to communicate about hardware. A scene comprises a 3D Gaussian Splat of the user's hardware registered onto the hardware's CAD model. The splat captures the current state of the hardware, and the registered CAD model acts as a referential anchor for troubleshooting instructions. With SplatOverflow, remote maintainers can directly address issues and author instructions in the user's workspace. Workflows containing multiple instructions can easily be shared between users and recontextualized in new environments. In this paper, we describe the design of SplatOverflow, the workflows it enables, and its utility to different kinds of users. We also validate that non-experts can use SplatOverflow to troubleshoot common problems with a 3D printer in a usability study. Project Page: https://amritkwatra.com/research/splatoverflow.

Paper Structure

This paper contains 66 sections, 12 figures.

Figures (12)

  • Figure 1: The components of SplatOverflow's web and mobile interface. (a) The palette of gestures SplatOverflow offers for selecting parts, guiding attention, and communicating actions. (b) The 3D Gaussian Splat of the hardware is aligned and registered onto the CAD model. (c) The timeline captures the troubleshooting interaction as a sequence of instructions and responses between users. (d) SplatOverflow's mobile interface allows local users to scan their hardware using a smartphone.
  • Figure 2: (a) Using the CAD model to visually query technical documentation and past issues. When the user clicks on the nozzle, SplatOverflow retrieves relevant sections from the assembly documentation and past issues referencing the part. Links to documentation are rendered in purple and previewed in the timeline as an iframe. Past issues are rendered in orange and can be recontextualized within the scene. (b) The local user recontextualizes a previous issue referencing the nozzle's poor suction; viewing the suggestions from a prior overlaid onto their machine.
  • Figure 3: (a) The local user's video is placed as a floating screen in the SplatOverflow scene and aligned to the same perspective it was filmed in. (b) The remote maintainer reviews the local user's issue by inspecting the vacuum nozzle in the splat and CAD model.
  • Figure 4: A reply from the remote maintainer asking the user to re-orient the machine to a new position and then update the scene with footage of the underside of the machine. (a) The wireframe corresponding to the final position the machine should be in. (b) The timeline element explaining why the movement needs to be made. When the local user clicks this element, the CAD model will animate to show how the machine should be moved. (c) The request for a new splat is rendered for the local user. The QR code links them to the mobile capture interface, where they can update the scene the QR code links to. (d) The local user performing the action specified by the maintainer and preparing to update the splat.
  • Figure 5: (a) The updated splat rendered in the browser. The machine is now standing upright as the maintainer requested, and the pneumatics in the staging plate are visible. (b) The timeline is updated to indicate that a new splat has been added, replacing the QR code with a success message. (c) A close-up of the vacuum pumps dislodged with their positions in the CAD model overlaid. (d) A close-up of the pneumatic seal that the maintainer is concerned about.
  • ...and 7 more figures