Tesla (TSLA.US) recently obtained a new patent that outlines a system and method for adapting neural network models on hardware platforms. The patent was filed on March 16, 2023, and raises speculation about the company potentially licensing its Full Self-Driving (FSD) technology to other manufacturers. Independent patent researcher SETI Park hinted on social media that this development might be related to implementing FSD on non-Tesla vehicles. However, Tesla has not disclosed any detailed plans regarding FSD licensing as of now.
The patent highlights how to acquire information on neural network models that include decision points linked to the network’s architecture. It also addresses a key challenge in machine learning applications: deploying and configuring neural networks on previously unsupported platforms. Mobile devices like smartphones and tablets increasingly require advanced machine learning implementations, stressing the need for adaptable neural network configurations that can work within the constraints of their hardware platforms.
Last year, Tesla CEO Elon Musk indicated in an earnings call that the company was considering licensing its FSD software and hardware to a major automotive industry player. Musk noted Tesla's openness to sharing its FSD technology under a model similar to the North American Charging Standard (NACS). Early discussions have already been initiated with a large Original Equipment Manufacturer (OEM) about the possibility of using Tesla’s FSD technology.
This announcement comes amidst rising global attention on Tesla’s advances in autonomous driving with its Robotaxi project. Despite technological strides, Tesla faces challenges such as regulatory scrutiny from entities like the National Highway Traffic Safety Administration (NHTSA) in the U.S., and national security concerns in China that could impede the approval of foreign autonomous vehicles.
The exploration of end-to-end learning models has paved the way for new prospects in autonomous driving, even as the technology faces obstacles. These models simplify the process by allowing the system to process data from sensors directly to generate driving commands. Tesla, along with several emerging players, is at the forefront of leveraging these models, striving for a system that can function seamlessly across various contexts by autonomously generating data-driven driving instructions.