As research fields mature, the most interesting work often transitions from happening in labs to happening in companies whose primary business is based on the technology. Twenty years ago if you wanted to advance the state of the art in search, Stanford was a great place to be. Today, if you want to advance the state of the art in search you’re much better off sitting on top of the data, infrastructure, tools, knowledge base, and centralized talent of Google or Baidu. Similarly, if you are interested in large scale distributed systems research, it is hard to work on more interesting challenges than those provided by the cloud infrastructure of Amazon, Facebook, Dropbox, etc.
Today, many people think that academia is the best place to be if you want to push the state of the art in robotics: computer vision, motion planning, control, and autonomy. Tomorrow, it will be clear that the best place to push the state of the art in these fields is at companies built around the core technologies.
Fundamentally, resources are meager in pure research settings compared to what’s available at great tech companies that reinvest in R&D. What’s been missing in robotics is a large market with problems that are in reach of the state of the art to spawn great companies. Today, there are three candidate markets coming into focus: manufacturing, self-driving cars, and drones. Of these, I believe that drones are the most interesting. Self-driving cars and manufacturing robots will allow us to do existing tasks more efficiently, while drones open up a world of possibilities that simply doesn’t exist today.
The success of manually piloted consumer and industrial inspection drones has shown us that “dumb drones” are already a billion dollar market. It seems a pretty safe bet that smart drones that are easy and intuitive for anyone to use (or operate fully autonomously) constitute a significantly larger market opportunity.
Today’s grad students in robotics spend half their time or more writing software “glue”, fixing broken robots, manually collecting datasets and other “distractions” in isolation. In tomorrow’s smart drone companies, researchers will have access to millions of flight hours of data collected in real world situations, uncovering untold interesting problems and opportunities for innovation. No more guessing what the research challenge is. They’ll work alongside exceptional hardware teams that build computation and sensor suites exactly to specification. They’ll use tools and infrastructure built by the best software engineers that make it easy to focus on getting the few lines of code that implement a new idea just right.
We’re very excited about the end customer experience we can deliver, but we’re also excited about the experience of doing product driven research at Skydio.