DeepScale's development of deep neural networks (DNNs), which can be fitted onto small, low-cost sensors in autonomous vehicles, has the potential to improve the accuracy of systems that interpret sensor data in real-time.
The software collects sensor data from the surrounding area as input and determines where everything is, what everything is called and how it is moving.
Because different sensors solve different perception problems -- for example, Lidar data to find the objects and Camera data to classify the objects -- a platform that integrates all sensors can improve accuracy and safety.
The DeepSpace software runs on a central processing platform commonly called an advanced driver assistance system (ADAS) domain controller, a type of processor designed by companies such as Nvidia, Qualcomm and Intel, among others.
"One of our core objectives is to drastically reduce the number of deaths and injuries on the road," Forrest Iandola, co-founder and CEO of DeepScale, wrote in a statement, announcing the funding round.
The company already has a number of strategic partnerships with OEMs and semiconductor suppliers to provide automated driving perception platforms, including Visteon and German automotive supplier Hella-Aglaia Mobile Vision.
Iandola received his doctorate at the University of California, Berkeley, working on deep neural networks and computer vision systems. While there he met Kurt Keutzer, his faculty advisor who would then become the company's co-founder.
Iandola's advances in scalable training and implementations of deep neural networks led to the founding of DeepScale, according to the company's statement.
Point72 Ventures, an early-stage venture capital firm specializing in artificial intelligence technology, and next47, an investment firm created by Siemens, joined existing investors Autotech Ventures and Trucks Venture Capital in the funding round.
DeepSpace is also making its perception software available for prototyping through development kits the company will offer in the second half of 2018, and will license its proprietary software to automakers and automotive suppliers.
The company plans to see its software embedded in vehicles that will begin selling around 2020, as well as autonomous taxis that will launch in the early 2020s.
— Nathan Eddy is a filmmaker and freelance journalist based in Berlin. Follow him on Twitter @dropdeaded209_LR.