Enabling edge AI in automotive applications

Artificial intelligence at the edge is a key enabler for enhancing the overall driving experience while increasing safety for drivers, passengers, and other people on the road. By processing data from sensors in real time, automotive edge AI solutions provide driver-assist features and the latest services regarding vehicle status.

Edge AI solutions in vehicles are especially useful for:

  • Advanced driver assistance systems (ADAS), which can detect objects on the road, such as pedestrians, vehicles, and obstacles, and alert drivers of potential hazards.
  • In-cabin monitoring systems in ADAS that can warn drivers if they are showing signs of fatigue or distraction, reducing the risk of accidents.
  • Improving driving efficiency and enabling predictive maintenance. Monitoring the performance of vehicle components can help reduce downtime and improve the overall car reliability.
  • Protecting the car's critical functions. As cars become more connected, the complexity of data flows increases, making them more vulnerable to cyberattacks. However, edge AI can provide real-time analysis and anomaly detection to safeguard critical functions and enhance cybersecurity.

ST enables edge AI in automotive applications with the SPC5 and Stellar automotive microcontrollers and software solutions, which support machine learning and artificial intelligence algorithms at the edge.

Running tiny neural networks on the SPC58 microcontroller series for body, networking, and security applications


The SPC58 series features powerful processing capabilities, advanced peripherals, and dedicated hardware accelerators, which can be leveraged for running edge AI algorithms.


Thanks to the SPC5 plugin (SPC5-STUDIO-AI) running on the SPC5-STUDIO development environment, you can convert and import pretrained neural networks.

  • The tool supports the most popular edge AI frameworks and training tools, including TensorFlow Lite, Keras, Caffe, Lasagne.
  • ONNX supports highly used frameworks, including PyTorch, MATLAB®, PaddlePaddle, MXNET, Chainer, Caffe2 and many more.



Running tiny neural networks on the Stellar E microcontrollers for software-defined vehicles


The Stellar E series of actuation MCUs meets the advanced digital control and high-performance analog requirements of the latest power technologies, silicon carbide and gallium nitride. They can be used in power conversion applications such as on-board chargers, DC/DC converters, and advanced motor control applications, like traction inverters.


The Stellar plugin (STELLAR-STUDIO-AI) for the STELLAR-STUDIO development environment, provides a seamless platform for generating, executing, and validating pretrained neural network models.
The Stellar AI plugin enables developers to quickly compile, install, and run pre-trained neural networks on Stellar E microcontrollers. The plugin can automatically generate efficient "Ansi C" libraries, reducing design cycle time and cost.

Use case: detecting driving conditions with edge AI on automotive microcontrollers

This advanced driver assistance system (ADAS) detects vehicle movement and road conditions. This solution can warn the driver of uneven road surfaces or if the vehicle is skidding or sliding for improved safety for those both inside and outside the vehicle.

This flexible AutoDevKit system solution kit (AEKD-AICAR1) is based on a long-short term memory (LSTM) recurrent neural network (RNN) running on an SPC58 C line automotive microcontroller.