Renesas Electronics and Fixstars Establish Automotive Software Platform Laboratory

Renesas Electronics and Fixstars Establish Automotive Software Platform Lab to Provide Users with Deep Learning Development Software and Operating Environment

As part of the collaboration, both parties launch cloud-based evaluation environment for R-Car SoCs

For simple initial evaluation of chip selection

March 2, 2021, Beijing, China - Renesas Electronics Corporation, a global supplier of semiconductor solutions, and Fixstars Corporation, a global leading supplier of multi-core CPU/GPU/FPGA acceleration technologies, announced today that the two parties have reached a cooperation intention in the field of automotive deep learning. Starting in April 2022, the two companies will establish an automotive software platform laboratory to provide development software and operating environment for Renesas automotive-grade chips. The new laboratory will support the early development and continuous operation of advanced driver assistance systems (ADAS) and autonomous driving (AD) systems. The two parties will also develop technologies for deep learning software development and build an operating environment for continuously updating learning network models to maintain and improve recognition accuracy and performance.

“Fixstars not only possesses advanced deep learning software technology, but also optimization technology that makes more efficient use of hardware,” said Ken Kataoka, Senior Vice President and Head of Automotive Solutions Business Unit at Renesas Electronics Corporation. “I believe that our collaboration will provide strong support for optimizing software development for in-vehicle applications, enabling our customers to fully utilize the superior performance of Renesas automotive-grade chips.”

“After developing a deep learning application, if it is not continuously updated with the latest learning data, high recognition accuracy and performance cannot be guaranteed,” said Satoshi Miki, CEO of Fixstars. “As we work with Renesas to develop a deep learning development platform optimized for their chips, Fixstars plans to focus more on machine learning operations (MLOps) in the automotive field.”

GENESIS for R-Car cloud-based evaluation environment

As part of the collaboration, Renesas and Fixstars today announced the launch of the GENESIS for R-Car evaluation environment, a cloud-based evaluation environment for R-Car that supports early development of ADAS and AD systems, helping customers perform immediate initial evaluations when selecting chips. It uses Fixstars’ cloud-based GENESIS chip evaluation environment.

Detailed specification research is time-consuming and inefficient. When selecting a device, evaluation based on actual use cases is essential. Therefore, users need to use evaluation boards and basic software to evaluate the device, but often require relevant technical expertise to build a relatively accurate evaluation environment. The new cloud-based GENESIS for R-Car evaluation environment does not require professional technical knowledge.

GENESIS for R-Car allows engineers to use common CNN models such as ResNet or MobileNet to confirm the processing time (in frames per second/fps) and recognition accuracy of the R-Car V3H's CNN accelerator on sample images. It also allows engineers to select the products and networks they want to evaluate and can be remotely operated on the actual development board. Engineers can use the GENESIS environment to confirm the evaluation results of tasks such as image classification and object detection, and can also choose to use their own image or video data for evaluation, which greatly simplifies the customer's initial evaluation process and makes it easier to determine whether the R-Car V3H is suitable for their system. Renesas plans to launch a service that allows customers to use their own CNN models for evaluation in the future.

Availability

GENESIS for R-Car is now available.