Unleashing the Power of Automotive AI/ML with VxWorks Using Texas Instruments Processors

In the realm of technological advancements, Artificial Intelligence (AI) and Machine Learning (ML) have emerged as transformative forces, re-shaping industries and pushing the boundaries of what was once deemed possible. The convergence of AI/ML with embedded systems has unlocked a plethora of opportunities, particularly in the automotive and industrial sectors. Embedded systems developers can benefit from the deterministic high performance of the VxWorks real-time operating system (RTOS), with its efficient AI/ML integration capabilities and options for cloud-enabled development, testing, and deployment – for faster time-to-market. VxWorks runs on Texas Instruments’ TDA4VH-Q1 processor offering developers valuable onboard functionality including integrated graphics capability, AI acceleration and video co-processing. This powerful combination is set to deliver intelligent edge computing for advanced driver assistance systems (ADAS). 

The Rise of AI/ML in Embedded Systems in Automotive

Embedded systems, characterized by their compact size and dedicated functionality, have traditionally relied on deterministic algorithms to perform tasks. However, the proliferation of sensors (radar, lidar and ultrasonic) and cameras in vehicles, coupled with the demand for real-time insights and decision-making, has fuelled the integration of AI/ML in automotive applications. These applications vary from ADAS to other functions including video processing for object detection, and pre-processing of vehicle sensor data for edge-to-cloud use cases such as predictive maintenance, tailored vehicle insurance, autonomous navigation, and other use cases yet to be defined.

Wind River VxWorks: The Foundation of Real-Time Intelligence

VxWorks can help to drive the seamless integration of AI/ML into embedded systems. As the most widely deployed and trusted RTOS, renowned for its deterministic performance, scalability, safety and security certification, and design quality, VxWorks provides a robust foundation for deploying mission-critical applications in diverse environments. Its Amazon cloud enablement, allied to DevSecOps pipeline integration and support for containerized deployment with Kubernetes orchestration serve to bring a modern dynamic – enabling embedded systems development teams to significantly reduce time-to-market. 

But what of AI/ML support? VxWorks has integrated the TensorFlow Lite open-source deep learning framework – speeding the implementation of machine learning models in resource-constrained environments. And for handling vast amounts of data, the Pandas and NumPy python-based libraries enable complex mathematical operations, as well as data manipulation and analysis.

Texas Instruments TDA4VH-Q1: Empowering Intelligent Edge Computing

Semiconductors play a crucial role in modern automotive technology, contributing to various aspects of vehicle functionality and safety. These include Electronic Control Units (ECUs) – managing and controlling various systems; vehicle domain controllers – handling multiple ECU functions in modern vehicle architectures; passive safety systems – such as antilock braking, and traction control; Infotainment and connectivity; and ADAS.

Advanced high-performance semiconductors, such as the Texas Instruments (TI) TDA4VH-Q1 automotive system-on-chip (SoC) features integrated graphics, AI acceleration and video co-processing - enabling sensor fusion and autonomous layer 2/3 domain control in ADAS applications.

Part of TI’s high-performance processor portfolio and engineered to meet the stringent requirements of automotive applications, the TDA4VH-Q1 features eight Arm® Cortex®-A72 CPU cores for application processing and six Arm® Cortex®-R5F co-processors for real-time processing. Integrated deep learning accelerators provide efficient machine learning inference (applying a trained ML model to make predictions/decisions on new, unseen data). Additionally, this SoC offers configurable vision acceleration for computer vision tasks such as edge detection, and high-speed interfaces for seamless connectivity with sensors and peripherals.

VxWorks on TDA4VH-Q1: Driving automotive AI/ML innovation

The VxWorks update cadence sees several major releases per year, to regularly add state-of-the-art SoCs to the stable of supported hardware along with innovative RTOS functionality. With the latest version of VxWorks released in early March 2024, support was added for deep learning applications with TI’s Deep Learning (TIDL) Library – TI’s software ecosystem for deep learning algorithm acceleration. TIDL allows users to run inference for pre-trained models on TI devices, using Convolutional Neural Networks (CNN) – a class of deep neural networks commonly used for analyzing visual imagery. TIOVX – an implementation of the OpenVX standard for cross-platform, optimized, power-efficient acceleration for vision applications is also supported.

Bringing VxWorks to automotive technology enables numerous and varied use cases, such as:

• Complex algorithms – for ADAS and autonomous driving.
• Advanced image processing – for high-resolution cameras and surround view systems, enabling accurate detection and recognition of objects and obstacles around the vehicle.
• Peripheral integration – such as CAN (Controller Area Network), Ethernet, PCIe (Peripheral Component Interconnect Express), USB, and others, providing connectivity options for automotive systems.
• Automotive applications features – ensuring the reliability and security of the system, such as functional safety mechanisms and secure boot.
• Automotive-grade reliability – meeting the rigorous requirements of the automotive industry, including temperature, reliability, and quality standards.

VxWorks on TDA4VH-Q1 empowers developers to create AI-powered solutions that deliver high performance, reliability, and safety.

For more information about VxWorks, visit: windriver.com/products/vxworks.

For details on Texas Instruments TI TDA4VH, visit: TI TDA4VH-Q1

About the author

Alan Stranaghan is a Senior Product Marketing Manager at Wind River