There are more than 10,000 robot companies in China.
Reportedly, less than 5% of them are doing well.
But what if they didn’t all have to reinvent the wheel? Or the servomotor. Or computer vision, spatial recognition, natural language processing, gripping, navigating, or any of a host of different problems that robots need to understand in order to be useful. What if they only had to learn the specific pieces of functionality that are unique to their specific set of tasks?
“70% [of each robot’s software stack] is generic; 30% is very specialized,” Blue Ocean Robotics CEO Claus Risager told me recently on the TechFirst podcast.
That means, theoretically, that robot startups which share code could develop software to make dumb metal, plastic, batteries, motors, and computer chips intelligent and capable at least twice as fast as…