Back in 2010, experts from the American University of Boston created a software platform capable of providing rovers and drones with high-quality computer vision and an object recognition system with modest energy and hardware resources. At Neurala, they were sure that the unmanned vehicle is quite capable of seeing pedestrians, obstacles and recognizing various objects using conventional devices comparable to the power of smartphones.
That is why the American Space Agency initially drew attention to the development and offered its cooperation, as they needed to provide rover with inexpensive equipment capable of recognizing objects under different conditions, including with poor visibility and without expensive optics – then NASA decided that the development can be used to study Mars. Now inNeurala are going to use some of their ideas on the Earth – the startup has already concluded a series of deals with the manufacturers of drones, drones and autopilots, because they all need inexpensive stand-alone computer vision systems and flexible artificial intelligence.
“Neurala is currently engaged in software optimization,” writes IEEE Spectrum, “because in order for autonomous transport and unmanned aerial vehicles to use on-board AI to identify objects, software needs to be thoroughly refined. For this, the project has already attracted about 16 million dollars of investment. “
Experts Neurala argue that after they bring everything to the mind, in the sky and on the ground will be a lot of apparatus working on their software. Neural network Neurala Brain is undemanding to resources and it is easy to adapt – with it even an ordinary robotic vacuum cleaner will recognize its owners.