In the demonstration, the combination of the Yocto Project–based Wind River Linux and NVIDIA Jetson Orin™ provides a groundbreaking, CUDA-optimized solution for real-time object recognition in dashcam videos.

Wind River®, a global leader in delivering software for mission-critical intelligent systems, today announced that it will be demonstrating the use of Wind River Linux with the NVIDIA Jetson™ platform for advanced generative AI at-the-edge applications at NVIDIA GTC, a global AI conference running March 18-21 at the San Jose Convention Center.

In the demonstration, the combination of the Yocto Project-based Wind River Linux and NVIDIA Jetson Orin™ provides a groundbreaking, CUDA-optimized solution for real-time object recognition in dashcam videos. This can significantly enhance the safety and efficiency of urban mobility by providing high levels of precision for AI-driven object identification on the road.

This commercially supported Yocto Project-based Linux solution can be leveraged for any AI application with data models running on NVIDIA Jetson across multiple mission-critical applications, such as voice or gesture recognition in the use of robotics, factory safety, intelligent machines, or drone applications.

"The potential for generative AI to transform industries and improve everyday life is immense. When NVIDIA technology is combined with Yocto Project open source Linux, next-generation AI is enabled," said Paul Miller, chief technology officer, Wind River. "The combination of Wind River and NVIDIA technology can help deliver a foundational shift by empowering developers to rapidly create new innovations in AI and machine learning and contribute to the development of smarter, safer applications. We are excited to highlight this breakthrough technology at NVIDIA GTC."

Through the integration of advanced video processing algorithms and deep learning expertise, the robust and scalable solution improves the accuracy and speed of object recognition. Using the Yocto Project­-based Linux environment, the solution can offer unparalleled customization and control, tailor-made for the unique needs of AI for dashcam use cases. The demonstration showed a 25% improvement in computational efficiency and application performance in the Yocto Project-based environment compared to other Linux configurations previously tested, leveraging CUDA optimizations.

To thoroughly assess the capabilities of the AI-powered object detection system, the test utilized custom input videos provided by users from their personal dashcams. This approach allowed the evaluation of the system's performance in real-life scenarios, ensuring that the solution was tested against the unpredictability and diversity of real-world conditions.

The AI model successfully identifies and classifies various elements in a live traffic situation, including multiple cars (with their corresponding varying confidence levels) and a pedestrian. Key highlights from the image analysis include:

Accuracy in complex environments: The system demonstrates high accuracy, detecting vehicles and pedestrians with a high degree of confidence, despite the challenges presented by a busy urban street.
Real-time performance: The AI model processes the video stream efficiently, maintaining a high frame rate (188.8 FPS), which is crucial for real-time applications. This performance indicates the model's potential for integration into active advanced driver assist systems.
Practical application: By using footage from a user's dashcam, the demonstration validates the system's effectiveness in a true-to-life context, a step forward in ensuring that the technology can be reliably deployed in everyday scenarios.
Additionally, the combination of Wind River Linux and Jetson is also being demonstrated at NVIDIA GTC by Yaskawa to showcase an industrial robot using AI technology to make judgements and automate movements in unstructured environments.

A platform for edge AI, embedded, and robotics applications, NVIDIA Jetson provides scalable software, a modern AI stack, production-ready ROS packages, and application-specific AI workflows. It can share AI software and cloud-native workflows while delivering the power-efficient performance required for building software-defined autonomous machines and breakthrough edge AI solutions.

As the embedded industry's leading and most advanced Linux platform,1 Wind River Linux helps teams develop, deploy, and operate robust, reliable, and secure embedded solutions running on a purpose-built Linux operating system. Wind River Linux offers high stability and security to meet the high-performance needs of mission-critical applications.

To view Wind River technologies firsthand at NVIDIA GTC, taking place March 18-21, visit Wind River at booth #140.

For more information about Wind River Linux, visit www.windriver.com/products/linux.

About Wind River

Wind River is a global leader in delivering software for mission-critical intelligent systems. For more than four decades, the company has been an innovator and pioneer, powering billions of devices and systems that require the highest levels of security, safety, and reliability. Wind River software and expertise are accelerating digital transformation across industries, including automotive, aerospace, defense, industrial, medical, and telecommunications. The company offers a comprehensive portfolio supported by world-class global professional services and support and a broad partner ecosystem. To learn more, visit Wind River at www.windriver.com.

1 VDC Research, The Global Market for IoT & Embedded Operating Systems

Wind River is a trademark or registered trademark of Wind River Systems, Inc., and its affiliates. Other names may be the trademarks of their respective owners.