For best experience this site requires Javascript to be enabled. To enable on your browser, follow our accessibility instructions.

Roll over image to zoom in

This page contains information on a preproduction product. Specifications and information herein are subject to change without notice. For additional information contact support or your sales representative.

Discover AI for NXP Automotive Explore the Toolkit Now.

eIQ® Auto ML Toolkit provides high performance and fast deployment of ML algorithms on NXP automotive microcontrollers and microprocessors including S32G vehicle network processors, S32K general-purpose microcontrollers, and S32Z and S32E real-time processors.

The toolkit offers large compatibility of AI container formats from diverse popular AI training frameworks. It features a wide range of evaluation models within its release package and numerous model preparation services for model optimization, maximizing the usage of S32 computing resources. The Toolkit supports a variety of AI backends including an automotive-quality compliant option.

Application integration happens through eIQ Auto runtime API, without dependencies on specific models, backends, and S32 hardware, providing full control of where and how the application is executed.

Software Details

Diagram

Choose a diagram:

eIQ Auto ML Toolkit
eIQ Auto Development

eIQ Auto ML Toolkit

eIQ Auto Development

Get diagram PDF

Features

Key Features

Seamless deployment of ML algorithms to embedded devices
Automotive-qualified machine learning (ML) runtime software stack, including accelerator backends
Heterogeneous multi-core processing
Portability across S32 Automotive Processing Platform
User-defined backend capabilities
Model Zoo

Integrated Model Zoo in eIQ Auto release package

Offline Tools

Import models trained from PyTorch, TensorFlow/Keras, MATLAB®
Hardware target selection and model optimization tools specifically for S32 devices
Integrated model compilation support
x86 simulation and model evaluation

Runtime API

Single API independent of backend configuration
Heterogeneous computation flexibility

Runtime Backend

eIQ Auto automotive-quality backend
ONNX® runtime
LiteRT
LiteRT Micro
User-defined backend

Supported Devices

S32E2: S32E2 Safe and Secure High-Performance Real-Time Processors with Actuation Support

S32G2: S32G2 Processors for Vehicle Networking

S32G3: S32G3 Processors for Vehicle Networking

S32K3: S32K3 Microcontrollers for Automotive General Purpose

S32K5: S32K5 Automotive Microcontrollers

S32Z2: S32Z2 Safe and Secure High-Performance Real-Time Processors

System Requirements

Development environment

x86 Linux®
x86 Ubuntu LTS

Supported AI model format

ONNX®
Protobuf
Keras H5
Saved Model

Embedded Operating Systems

Linux®
FreeRTOS^™
QNX® Neutrinos
AUTOSAR®
Zephyr®
BareMetal
NXP RTOS

Model Zoo Applications

eIQ® Auto ML toolkit includes a model zoo for fast evaluation. Customers can utilize this open-source model zoo tailored to NXP devices to develop, test, and deploy models. The model zoo includes:

Application	Task	Input	Algorithm
Battery Management	State of Charge Estimation	Voltage Temperature	SoC LSTM, SoC GRU, SoC MLP
Driver Monitoring	Driver identification Driver behavior classification	CAN bus data	Drive2Vec GRU, Drive2Vec LSTM
Driver Monitoring	Driver awareness classification	Driver facing image	CNN
Emergency vehicle detection	Acoustic siren recognition	Microphone data	CNN (ResNet) + LSTM
Intrusion detection	Identify malicious network activity	IP records	LSTM
Machine health	Anomaly detection	Microphone data	LSTM
Machine health	Predictive maintenance	Sensor readings	TCN, LSTM
Road surface classification	Image classification	Road image	CNN
Data analytic	Driver Behavior Analysis	Data analytic	RNN
Data Analytic	Engine predictive maintenance	Data Analytic	RNN
Data Analytic	BMS (SOC, battery temperature prediction)	Data Analytic	RNN
Audio Processing	Siren detection	Audio Processing	CNN + LSTM hybrid
Data Analytic	Anomaly detection	Data Analytic	Transformer (multi-head attention network)
Autonomous Driving	Vision based road surface classification	Autonomous Driving	CNN (Resnet-18)

Model Zoo is included in eIQ® Auto release package. To access. Please download the eIQ® Auto toolkit. These evaluation models are not found on other online sources.

For additional information contact support or your sales representative.