While troubleshooting hardware, a remote hardware maintainer may have to get on a video call to look at an employee’s broken laptop, malfunctioning factory machine, or other busted gear—scenarios leading to over-the-phone requests like, “Get closer to that part!” “Tilt the computer to the left!” or “Gimme your address. I’ll just send a new machine.”
Cornell researcher Amritansh Kwatra wants to provide an “asynchronous,” Google Docs-style approach to hardware fixes. With a name based on the software-help site StackOverflow, “SplatOverflow” aligns a phone’s video scan with its CAD, or digital design, file. The stacked rendering allows teams to clearly reference parts, provide suggestions, and track data related to previous fixes.
“The person who is asking for feedback can see all of the suggestions, directly overlaid on top of exactly the machine or hardware that they’re facing a problem with,” Kwatra, a third-year PhD student and maker of SplatOverflow, told IT Brew.
How it works:
- A user scans hardware with a standard, no-lidar-needed phone camera. With well-placed QR stickers on the physical gear, a 3D CAD model aligns with the hardware, allowing end-users to see technical documentation and actionable steps (“Remove this bolt,” for example) attached to relevant parts.
- A rendering called “3D Gaussian Splatting” lets maintainers remotely and independently compare as-built hardware to the as-designed CAD model.
- An end-user with an issue can attach a video or write a message to the browser’s rendering of the relevant part. A maintainer, on their own time, can provide suggestions, perhaps clicking and dragging, for emphasis, which bolt needs to go.
- Notations remain in the system, so previous maintenance lessons can be recorded, not lost forever once the Zoom call is over, Kwatra said.
See a demo here.
Maybe the problem is a busted operational amplifier. A non-expert user can learn this by clicking on the part and viewing related technical documentation and previous issues.
Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
“You can now address it without having to ever know the vocabulary of ‘Oh, this is an op amp,’” Kwatra told us.
Stacking up. The “Splat” idea is based on Stack Overflow, a community software site allowing developers to submit complicated code questions—often with code samples—like how to assign “a null variable to two different blocks.”
“In code, troubleshooting relies on being able to reference code samples in questions, suggestions, and answers. The Splat aims to capture a similar means of referencing bits and parts on physical hardware,” Kwatra wrote in a follow-up email to us.
Kwatra hopes to build a similar open-source ecosystem for hardware troubleshooters. The Cornell researcher said one major problem he discovered, when speaking with manufacturers, was scaling support.
“Software has solved this by having this dedicated digital infrastructure. What would that digital infrastructure look like for hardware?” Kwatra said.
Staying out of troubleshooting. If you thought troubleshooting via videoconferencing was difficult, VikingCloud Cybersecurity Evangelist Jon Marler used to help customers fix their custom-built computers, over the far-less-smart phone in the 1990s.
Marler remembered jobs that sent him to Denver and then a four-hour drive into Nebraska; a tool like SplatOverflow could save an IT technician a long trip, he said.
Marler sees SplatOverflow as potentially “transformative,” he said, “because it would allow people who don’t have those skills to be able to quickly solve problems.”
Kwatra envisions the tool helping farmers as they repair their own machines. “If something breaks during harvest season, this is mission-critical,” he said.
“I’m not a big agricultural person,” he added.
With tools like SplatOverflow, the idea is that you don’t have to be.