Overview

AI Workbench enables developers to easily set up a tailored development environment in the cloud to fine-tune AI models, develop AI-enabled ADAS applications for the R-Car SoC platform, and then seamlessly validate on simulators or evaluation boards connected to the cloud. AI Workbench is built on the Microsoft Azure Cloud and will be supported by other cloud vendors such as AWS and GCP in the near future, allowing customers to integrate AI Workbench functionality into their development workflow.

AI Workbench
AI Workbench
  • R-Car AI model development
    • Select and compile AI models on real SOC boards in the cloud (Hardware In the Loop) through the web browser, using a predefined set of optimized pre-trained deep neural networks or a custom model.
    • Track and manage experiments efficiently for improved performance and accuracy via self-guided multi-parameter search, execution and testing of the models.
    • Gain insights from the experiments, re-train the models with custom datasets and re-deploy the new models to the NNPerf machine learning pipeline.
  • R-Car application development
    • Access development environment and source code management from the cloud through the web browser.
    • Configure the development environment to meet your application needs (simulator, hardware device, SDK version etc.), and then automatically provision a virtual machine in the cloud configured to your application workload needs.
    • Enable multiple developers to work on similar functions in parallel in either simulation (Software In the Loop) or on real SOC hardware, enabling early software development.

Design & Development

Related Pages
Hybrid Compiler
Common interface across generations of SoCs.
 Compiler/Assembler Renesas
1 item