Introducing Droplet

We started Droplet because we knew that there was a better way to get value out of robotics data.

This idea started with our frustration that robotics companies bog down on building internal data tooling rather than getting their robots to product market fit. It’s a bit of a running joke in the robotics industry that every company has their own, slightly-bespoke version of the same software stack. Everyone has their own data visualization tool. A customized data serialization system. Their own pub-sub middleware. Fragile data pipelines for unpacking robot logs. Sure, ROS exists, with plenty of users, but there are well-known issues using it at scale.

We’ll get deeper into why these problems exist later, but for now, suffice it to say that robotics is an extremely multidisciplinary field with a lot of technical surface area. The table stakes for a robotics company — to produce a robot that does anything at all — require many specialist engineering teams and large budgets. So we really feel for all those robotics engineers who do soldering iron surgery on Tuesday, embedded Linux on Wednesday, and data pipelines on Thursday, because we used to wear those same shoes.

Data has become a critical component of the robotics R&D process since almost every robot involves a machine learning model. But the vast amount of data that robots produce makes it hard to build a robot that improves its performance based on issues it sees in the real world. We believe that working with data in robotics should be as easy as any other data-intensive field.

That’s why we’re building Droplet: a data platform built by roboticists and software engineers designed for the unique needs of complex, timee-series sensor data. Our primary goal? To streamline data collection, enable efficient analysis, and help users identify and resolve potential problems in the field. 

Droplet is taking the hard fought data engineering lessons learned by the FAANGs of the world and applying them to an industry whose data models and cost constraints require something different.

Stay tuned as we share more about Droplet and the robotics industry in subsequent posts. If anything here resonates with you, share your info or get in touch!