Artificial intelligence (AI) has a fairly wide range of application fields. When AI combines popular automobile and IoT applications, there will be brand-new opportunities. This article will introduce AI development trend and relevant solutions of NXP Semiconductor in AI applications for you.
AI is the key factor of automobile, industrial and IoT applications
It was around 2015 when multi-core application processors and Graphics Processing Units (GPUs) started to be widely used, these tools can be adopted for processing vast amounts of data. Together with parallel processing capability, it can become faster and cheaper with more powerful functions. Combining rapid and abundant storage and more powerful algorithm, it can conduct data sorting and structuring. These factors made AI flourish.
In 2018, AI voice recognition software through neural network training became an indispensable part of various consuming and industrial applications. Its computing power increased about tenfold every year, which was mainly driven by new classes custom hardware and processor architecture. This computing boom was the key component of AI improvement and also helped promote AI to become the mainstream technology. And that was just a beginning.
Along with the gradual maturation of AI technology, more and more applications use AI to improve product functions and value, such as adopting voice recognition in automobile entertainment information system control, applying AI in automobile active safety system or even importing automatic drive functions and adding more intelligence for “sensing”, “thinking” and “acting” processes of autonomous vehicles in order to help solve global traffic challenges and achieve zero fatalities, zero emission and zero time waste.
In addition, due to the import of AI technology, popular IoT applications not only make smart connected devices talk to each other but also use AI to replace us in mutual interaction. This new global artificial intelligence things structure is called the Artificial Intelligence of Things (AIoT). The model transformation brought by the integration of AI and IoT will be bigger than that when introducing personal computer or mobile phone.
AI launches brand-new opportunities for automobile industry and IoT applications. In order to satisfy the industrial demand, NXP collaborates closely with leading academic institutions, research institutions and technology companies, making it take the leading position in developing automobile AI and AIoT solutions. Covering a series of IC optimization from MCU to applications and network processors, NXP can provide relevant devices and software required by AI applications for customers and provide extensible processing performance to support three-axis trade-off among accuracy, inference time (user experience) and cost. The text below will introduce various AI solutions provided by NXP.
Provide automatic drive solutions for automobile industry
In the aspect of automobile industry, NXP launches eIQ™ Auto kit which is an inference engine neural network compiler according with A-Spice and an optimized deep learning library used in automobile applications. NXP eIQ Auto deep learning kit helps developers introduce deep learning algorithm into its applications and continue satisfying automobile standards.
BlueBox development platform series can provide the performance, functional security and automobile reliability required by engineers to develop autonomous vehicles. BLBX2-xx series, the latest member of this series, integrates the S32V234 automobile vision and sensor fusion processor, LS2084A embedded computer processor and S32R27 radar microcontroller. The NXP BlueBox platform can provide the performance required by analyzing driving environment, evaluating risk factors and guiding automobile behavior.
The SBC-S32V234 vision evaluation board is an evaluation and development board with cost competitiveness, which is specifically designed for high-performance, safe, computation-intensive front vision, surround vision and sensor fusion applications. The SBC-S32V234 is collaboratively developed by MicroSys, which is the S32V processor based on 32-bit Arm® Cortex®-A53 with high-efficiency form factor and many use cases of the S32V234.
The S32V234 processor is the APEX accelerator integrating accelerating deep neural network for optimizing front and surround view camera, machine learning and sensor fusion. The S32V234 MPU provides image signal processor (ISP), powerful 3D GPU and dual APEX-2 vision accelerator as well as automotive-grade reliability, functional security and safety functions in order to support applications like computation intensive ADAS, NCAP front camera, object detection and recognition, surround view, automobile and industrial image processing, machine learning and sensor fusion.
Layerscape® communication processor has conducted optimization for virtualized network and embedded system, which is built based on Arm® technology. The processor provides wide depth and breadth, which can expand the performance to the minimum form factor from power-constrained network and industrial applications to new-type virtualized network and embedded system requiring advanced data path and network peripheral interface.
Make IoT applications more intelligent
In the aspect of industrial and IoT applications, NXP launches eIQ™ machine learning software development environment with scalable, robust and reliable inference engine, which is used for deploying machine learning applications and supports machine learning algorithm used on NXP MCU, i.MX RT crossover processor and i.MX series SoC. eIQ software includes the library of inference engine, neural network compiler and optimization, which is completely integrated in the development environment of MCUXpresso SDK and Yocto and can easily develop complete system-level applications.
In addition, the NXP sensor software development kit (ISSDK) is the embedded software framework supporting Sensor Toolbox ecosystem as well as the digital and analog sensor platform supporting NXP IoT applications. The ISSDK provides a set of unified sensor supporting models aiming at NXP sensor product portfolio including NXP microcontrollers (like NXP LPC, Kinetis and i.MX RT crossover platform) based on Arm® Cortex®-M core.
Aiming at voice and audio applications, NXP launches the i.MX 8M development kit used for Amazon Alexa voice service, which makes developers easily integrate Alexa into their next generation designs. In order to provide high quality audio experience, NXP cooperates with DSP Concepts to develop audio algorithm including echo cancellation, beam forming and noise reduction. Using TechNexion Voice Hat 2-Mic Kit as the audio front end and PICO-PI-IMX8M development board as the processor, it processes “Alexa” wake word recognition and connects to Amazon Alexa voice service.
EdgeScale kit is a scalable and safe device management solution suitable for edge computing applications. Being a set of software tool and service, it provides secure manufacturing, enrollment of IoT and remote management of edge computing devices. This solution also provides a kind of secure mechanism for developers that they can deploy and operate various edge applications through popular cloud frameworks (including AWS Greengrass, Microsoft Azure IoT, Google Cloud IoT, Alibaba and private cloud frameworks).
NXP and its cooperative partners provide wide software development tools to start embedded software development including design tools, applications and sample codes, supporting OpenCV, OpenCL and OpenVX™ etc. Design tools include CodeWarrior®, VortiQa®, S32 Design Studio IDE and MCUXpresso® software and tools, eIQ™ machine learning software development environment, S32 software development kit, Processor Expert® and embedded components and eIQ Auto toolkit deep learning enablement for Auto etc.
The Coral development board aims at the rapid prototyping making of those applications requiring rapid on-device machine learning inference. The Coral development board kit includes a system-on-module (SOM) and a baseboard. The SOM based on iMX 8M application processor also includes LPDDR4 memory, eMMC storage, dual-band Wi-Fi and Edge TPU. As a small ASIC designed by Google, Edge TPU can achieve high performance and local inference under the circumstance of low-power consumption, and thus transforms edge computing power of machine learning.
FEATURED PRODUCT
RELATED MANUFACTURER