STMicroelectronics accelerates the application of edge artificial intelligence to help equipment manufacturers complete the intelligent transformation of their products
▲ Click above to follow STM32
STMicroelectronics ( ST ) announced the launch of ST Edge AI Suite , a resource-rich embedded artificial intelligence ecosystem to help manufacturers add edge artificial intelligence to their products. ST Edge AI Suite is a set of integrated software tools that can be used with ST hardware for free. The release of this product takes the services provided to customers a step further and can help customers quickly develop autonomous IoT devices with locally embedded artificial intelligence and Deployment of billions of devices. The ST Edge AI Suite leverages STMicroelectronics’ various hardware (general-purpose microcontrollers, automotive microcontrollers and microprocessors, smart sensors) and related embedded AI optimization tools to simplify the development of customer AI solutions. development. The suite will extend and integrate a variety of software tools, evaluation kits and development kits that ST has introduced over the past decade, while better leveraging ST’s existing machine learning ( ML ) frameworks and support from key partners such as Nvidia and Amazon Cloud Technology) artificial intelligence ecosystem.
We are moving toward an IoT world of tens of billions of connected, autonomous devices that bring practical value and convenience to users in every aspect of consumer life and business productivity. To achieve this goal, it is necessary not only to run artificial intelligence algorithms in the cloud, but also to deploy AI on a wide range of edge devices , including smartphones, personal networked devices, smart homes, building control systems, industrial machines, automobiles, and more. ST products are already at the heart of all of these devices, but to do even more, we need the industry-leading free software suite we are launching today. This combination of software and hardware solutions will bring us into a more intelligent era of edge computing, allowing enterprises of any size to more easily deploy edge artificial intelligence and build their future Internet vision with the help of ST 's hardware products.
STMicroelectronics briefly introduced the functions of the ST Edge AI suite at the recent Edge Artificial Intelligence Online Summit. This market-leading solution will help embedded developers optimize machine learning models, help data scientists run machine learning models on embedded devices, and help product designers and makers redefine important features of their products.
By providing free download and use services, STMicroelectronics will empower all types of customers, large, medium and small, and pool resources and knowledge into a community-based development model. This suite will focus on the broader edge artificial intelligence community, integrate tools, unite users, and further promote the communityization of edge artificial intelligence development.
The first version of the ST Edge AI suite will be available in the first half of 2024.
Edge artificial intelligence further shortens the distance between intelligence and decision-making capabilities and data sources, and is a key technology for enterprises to realize product intelligence in today's Internet era. The benefits of edge artificial intelligence are mainly reflected in response speed, energy efficiency and power consumption, privacy protection, network security and cost-effectiveness, allowing enterprises to create smarter, faster response and more energy-efficient
products to meet the fast pace of today's world. and data-driven needs.
一家大型家电制造商
目前正在采用该洗衣机解决方案升级他们的产品,预计在明年会看到
他们的新品上市。其中用到的第一个机器学习算法是创建一种虚拟传感器,根据电机电流测量值估计待洗衣服的重量。
第二个机器学习算法是利用六轴运动传感器数据计算滚筒是否会接触洗衣机的外壳,从而避免滚筒碰撞洗衣机外壳。根据算法的输入数据,洗衣程序提供洗这些衣服所需的电流驱动电机,并调整用水量和洗涤剂量,因此,洗一次衣服可以节省15%至40%的电能和水。这两个算法都是基于ST的NanoEdge AI开发的,并运行在STM32G0 MCU和ST六轴运动传感器上。
惠普 工程团 队与意法半导体密切合作,开发和训练活动识别人工智能模型,根据设备运动和用户动作来识别不同的用户活动。该合作面向多个使用场景,包括将笔记本电脑放在桌子上、用户腿上或放在包里和从包里取出的场景,帮助惠普创建了一个 运行在智能六轴运动传感器上的、惠普设备专用人工智能模型。其中最值得一提的是功耗,该传感器运行边缘AI算法的功耗仅为34微安。该解决方案使惠普计算机能够发现用户活动变化,并做出相应的响应,为更关键的任务节省电池电量。
意法半导体正与 HPE集团 合作,优化电动汽车电机的运行和维护,这方面的挑战是如何在实际使用中监测电动机转子的内部温度,以优化输出功率,提高电机运行的能效和安全。在实验室中,转子是裸露在外的,可以直接测量电机内部温度,但在车上测量正在运 行的电机的内部温度是行不通的。ST提出的解决方案是训练一个模型,并使用边缘人工智能构建一个虚拟温度传感器,根据外部温度测量值估算内部转子温度。该算法运行在控制电机的Stellar车规微控制器上。该微控制器还运行一个状态监测人工智能算法,通过分析振动来检测潜在的异常事件。这个解决方案也可以用于其他关键部件,例如,测量动力电池内部温度是不现实的,但边缘人工智能模型可以用外部测量数据模拟内部温度。
意法半导体
的人工智能
策略基于
一套
资源丰富的
集成
化开发
工具(其中一些
已上市
)、技术示例
、学习示例
,以及
名为
ST Edge AI Core Technology
,一款创新的
统一嵌入式人工智能解决方案优化器。
ST Edge AI
套件满足了不同用户的需求,如数据科学家、嵌入式软件开发人员和硬件系统工程师。该套件简单好用,界面简约直观,有多个不同版本(
Desktop, CLI, Web, API
)
。
❖
S
T Edge AI Suite
supports multiple hardware platforms of STMicroelectronics
:
STM32
general-purpose
MCU
, including
already released
products
with
AI
hardware acceleration capabilities, industrial-grade
STM32N6
and
STM32
MPU
, as well as
Stellar
automotive-grade microcontrollers and smart sensors. Among them, automotive-grade microcontrollers support automakers' transition to software-defined vehicles, with predictive maintenance functions for electric drive systems to extend vehicle life or battery management systems to maximize energy efficiency. Embedded smart sensors are based on the Intelligent Sensor Processing Unit
ISPU
, machine learning core
MLC
, and advanced imaging using time-of-flight sensors, making them ideal for all applications from industrial, automotive, and wearable accessories to high-end personal electronics. The above products come with various evaluation boards and development boards.
❖
S
T Edge AI Suite
integrates all software and tools required for each stage of project development,
of which
ST Edge AI Core
is a key component
.
ST Edge AI Core
can
export
ML
machine learning and
NN
neural network algorithms from
the most widely used machine learning frameworks
, and generate detailed algorithm analysis reports to
optimize algorithms for selected devices
such as sensors,
MCUs
,
MPUs
, etc., and
verify algorithms based on the original model , and finally, convert the final embedded
AI
solution into code suitable for the selected device. The same artificial intelligence algorithm can be benchmarked on different platforms using pure software or specific hardware accelerators, and algorithm accuracy and inference time can be evaluated with a few mouse clicks.
❖ The NanoEdge AI Studio autoML tool is free and open to STM32 users and is now available for all ARM Cortex-M based MCUs : At the same time, STMicroelectronics ’ NanoEdge AI Studio autoML tool is being upgraded and can be used free of charge by customers around the world . Now, software libraries created with NanoEdge AI Studio will no longer charge deployment fees, and there is no quantity limit for deployment on STM32 microcontrollers . Additionally, since NanoEdge AI Studio targets all ARM Cortex-M based microcontrollers, customers can now sign special license agreements to build and deploy model libraries on other brands of ARM Cortex-M microcontrollers , including exclusive devices Go to study.
© THE END
Follow STM32