Re-introducing Tangram Vision

Hello, all. We’re Tangram Vision, the Reliable Perception company. We do calibration.

If you've ever worked on calibration, you know it touches nearly every system in robotics, automation, and physical AI. Everything using perception data eventually needs calibration, which means calibration engineers end up becoming experts in... well, everything.

This creates an interesting problem. Calibration experts can spot issues everywhere: navigation algorithms that will drift off course, deep learning pipelines making systematic errors, timing problems between sensors. With this broad knowledge comes an irresistible yearning to fix every problem they encounter.

For the past four years, we at Tangram have yearned. We always knew that calibration was a pain point for robotics, but what about everything calibration touches? Could we fix those on the way? We tinkered and toyed with big ideas to make the robotics world better, but ultimately they never found purchase:

• What if there was a universal data format for sensors? That was .tang, an idea we ditched in our first year (thanks, MCAP!).

• What if we could build a system that streamed and fused data from any sensor type in real-time? That’s TVMux, an API which we ultimately never deployed. It’s kept in-house as a streaming solution for our own future needs.

• What if we built a self-calibrating sensor? That was HiFi, a Kickstarter project that we shelved early this year. I’ve only recently finished refunding all of our pre-orders and backers for that effort (thanks to all who supported us, by the way!)

It’s not that these are bad ideas. Far from it; I think they’re all fantastic ideas, and I’d love to revisit these when we’re in a place to do so. However, time is a zero-sum commodity. By using time here, we were diverting time from the problem that we were uniquely capable of solving: calibration. Our core mission took a hit, along with our business.

So starting in 2025, we pulled plugs, accepted sunk costs, and focused up. All engineering time now goes directly to making our calibration software better: better user experiences, better mathematics, better integration. MetriCal is getting more modalities and faster runtimes. We’re iterating on sales and pricing to make our products affordable to everyone. NASA just awarded us a Phase II SBIR to build extraterrestrial calibration (which we’ll post about soon). Even our branding and messaging has been honed; you can check out the website for the new look. We’re laser-focused, and it feels amazing.

It’s true that we’ve been a little quiet. But we’re still here, still busy, and doing great. We're excited to share more about what we've been building - both in upcoming blog posts and in the product updates you'll start seeing again. We hope to make your life just a little bit better by making calibration suck a little bit less. Thanks for sticking with us.