Tesla acquires neural net startup to further goal of creating truly autonomous cars

Source - Tesla

Tesla’s acquisition of DeepScale, a computer vision startup, is the latest move towards Elon Musk’s vision of making Tesla’s into fully autonomous vehicles. Currently Tesla models are not considered fully autonomous or “level 4” by SAE International, meaning that they not able to fully operate and drive without any form of human intervention. On the SAE classification scale Tesla currently ranks at level 2, which is more akin to a advanced driver assistance.

Elon Musk said earlier this year during Tesla’s Autonomy Day that Tesla vehicles now have the hardware needed to become fully autonomous and the only thing remaining is to improve the soft-ware, which is exactly where DeepScale fits into the equation.

 “We are excited to announce that, as of today, all Tesla vehicles produced in our factory – including Model 3 – will have the hardware needed for full self-driving capability at a safety level substantially greater than that of a human driver”

Tesla Team

DeepScale will be integrated into Tesla’s AutoPilot team, which has suffered lossess over the last year according to The Information. This integration at least for now seems to include the co-founder of DeepScale Forrest Iandola.

DeepScale was founded in 2015 by Dr. Forrest Iandola and Prof. Kurt Keutzer and has since been focusing on creating smaller and efficient deep neural networks (DNN) to driver assistance systems and autonomous cars. In 2018 it raised $15M in a funding round led by Point72, Steven Cohen’s new personal fund as well as next47, a VC fund backed by Siemens. Furthermore, during 2018 DeepScale announced multiple strategic partnerships with various automotive suppliers, such as Viseton.