Neural network software startup Neurala is preparing for the commercial roll-out of deep-learning technology originally designed to help NASA rovers operate autonomously on Mars.
The Neurala Brain software employs artificial neural networks that can train themselves to perform tasks like object recognition using multiple processing layers, according to IEEE Spectrum. The company’s software can run on relatively low-performance processing chips, such as those used in mobile devices like smartphones and tablets. This contrasts with many existing deep-learning approaches that require extensive cloud-based processing resources to operate.
The Neurala Brain’s computing-power stinginess arises from its original intended use in Mars rovers, which have limited computing capabilities. This makes the software suitable for autonomous vehicles like drones and self-driving cars that need to quickly recognize and react to surrounding objects.
“We want to provide the full, end-to-end solution to customers, including navigation, collision avoidance, and image processing, all small enough to run on a device versus doing the processing on a server,” said Massimiliano “Max” Versace, CEO of Neurala, speaking to IEEE Spectrum.
The Boston-based startup is testing a version of Neurala Brain and plans to start delivering it to exclusive customers within a few months.
Tyler Schulze is vice president, strategy & development at Veritone. He serves as general manager for developer partnerships, cognitive engine ecosystem, and media ingestion for the Veritone platform. Learn more about our platform and join the Veritone developer ecosystem today.