Semiconductor manufacturer Renesas Electronics, and Fixstars, a multi-core CPU/GPU/FPGA acceleration technology business, have announced the joint development of a tool suite that enables the optimization and efficient simulation of software for autonomous driving systems and advanced driver-assistance systems. The solution has been designed specifically for the R-Car system-on-chip (SoC) devices from Renesas.
These tools enable network models to be rapidly developed with highly accurate object recognition, right from the initial stage of software development, by taking advantage of the performance of the R-Car. As a result, post-development rework is reduced, shortening development cycles.
At present, AD and ADAS applications utilize deep learning to achieve highly accurate object recognition. Deep learning inference processing requires huge amounts of data calculations and memory capacity. The models and executable programs on automotive applications must be optimized for an automotive SoC, as real-time processing with limited arithmetic units and memory resources can sometimes be a challenging task. Additionally, the process from software evaluation to verification must be accelerated and updates need to be applied over and over again to improve accuracy and performance.
The tools that Renesas and Fixstars have developed to meet these needs include the R-Car Neural Architecture Search (NAS) tool for generating network models optimized for R-Car. This tool generates deep learning network models that efficiently utilize the CNN (convolutional neural network) accelerator, DSP and memory on the R-Car device. This allows engineers to rapidly develop lightweight network models that achieve highly accurate object recognition and fast processing time even without deep knowledge or experience of the R-Car architecture.
Next up is the R-Car DNN Compiler for compiling network models for R-Car. This converts optimized network models into programs that can make full use of the performance potential of R-Car. It converts network models into programs that can run quickly on the CNN IP and also performs memory optimization to enable high-speed, limited-capacity SRAM to maximize its performance.
Finally, the R-Car DNN Simulator can be used to rapidly verify the operation of programs on a PC rather than on the actual R-Car chip. Using this tool, developers can generate the same operation results that would be produced by R-Car. If the recognition accuracy of inference processing is affected during the process of making models more lightweight and optimizing programs, engineers can provide immediate feedback to model development, thereby shortening development cycles.
“Renesas continues to create integrated development environments that enable customers to adopt the ‘software-first’ approach,” said Hirofumi Kawaguchi, vice president of the automotive software development division at Renesas. “By supporting the development of deep learning models tailored to R-Car, we help our customers build AD and ADAS solutions while also reducing the time to market and development costs.”
Satoshi Miki, CEO of Fixstars, said, “The GENESIS for R-Car, which is a cloud-based evaluation environment that we built jointly with Renesas, allows engineers to evaluate and select devices earlier in the development cycles and has already been used by many customers. We will continue to develop new technologies to accelerate machine learning operations (MLOps) that can be used to maintain the latest versions of software in automotive applications.”