In 1956, when McCarthy of Stanford University proposed "artificial intelligence", he certainly did not expect that this concept would be in full swing in China decades later. Artificial intelligence is not only expected to trigger a new industrial revolution, but also integrated into everyone's daily life, and is triggering social change. In fact, in April 2020, when determining the three aspects of new infrastructure, the national development and Reform Commission mentioned artificial intelligence in information infrastructure and integration infrastructure, but only to clearly inform the public of the ongoing changes.
With the development of computing power, data and Internet, artificial intelligence is at the node from quantitative change to qualitative change, especially at the edge. Gartner predicts that by 2025, at least 75% of data processing will be carried out in the cloud or outside the data center. The tide of artificial intelligence is both an opportunity and a challenge for semiconductor enterprises. Different from the cloud, the main demand for chips on the edge side still returns to the eternal topic of performance, cost and power consumption, and products can win only if they are available at the same time. In addition, because the development cycle of edge AI products is short and the iteration window is fast, a friendly development environment is also key.
What kind of edge AI system is successful? "Accurate perception, rapid decision-making, man-machine cooperation, energy efficiency, safety and reliability". As we all know, perception, decision-making and execution are the three links of edge artificial intelligence. With the development of edge artificial intelligence, the requirements for embedded perception and decision-making technology are more stringent and differentiated than non artificial intelligence.
Perception -- the data source of edge artificial intelligence
Data is the basis of edge AI, and perception is the source of data. Just as a person not only needs eyes to feel the world, but also ears are important organs to perceive the natural world, machines also need ears and eyes. Various sensors came into being with the development of technology. Ti's single-chip millimeter wave radar can avoid the disadvantages of traditional cameras in many applications and support multiple data fusion of the system, so that the machine can better obtain data and realize accurate perception of targets.
Using the transceiver of millimeter wave radar, the distance and relative velocity of objects and obstacles in its field of view can be measured with high accuracy. Compared with sensors based on vision and lidar, an important advantage of millimeter wave sensors is that they are not easily affected by environmental conditions such as rain, dust, smoke, fog or frost. In addition, millimeter wave sensors can work in complete darkness or in direct sunlight. These sensors can be installed directly behind the plastic housing without external lens, vent or sensor surface. They are very durable and can meet the protection level (IP) 69K standard.
Ti's single-chip millimeter wave radar adopts CMOS manufacturing technology to realize the high cost performance advantage that traditional radar does not have. At the same time, combined with ASIC back-end processing, it can directly reduce BOM cost, reduce product size, and reduce the dependence on processor. The product designed based on Ti millimeter wave radar is one-third of the volume and half of the weight of micro lidar rangefinder.
More importantly, in addition to the field of automatic driving, millimeter wave radar can also be applied to a wider range of industries and smart homes, smart buildings, medical treatment and other fields. For example, through the combination of millimeter wave radar and air conditioning, multiple intelligent functions such as wind moving with people, attitude perception of target human body, automatic switching and so on can be realized. In other applications, such as safety monitoring for manipulator operators, obstacle avoidance detection for logistics robots / UAVs, fall monitoring for the elderly, millimeter wave radar has the advantages of accurate and fast sensing that previous image sensors do not have, and meets the data desensitization requirements of many applications (it can be installed in bedrooms, toilets and other occasions).
In addition to millimeter wave radar, Ti also provides temperature sensor and DLP ® Technology, TOF and other products, which further enriches the interaction between machine and human.
Decision making -- the brain of edge artificial intelligence
Edge AI devices need a smart "brain" to process data and make decisions. Integrated SOC is usually a good choice in edge AI, because in addition to accommodating various processing elements that can perform deep learning reasoning, SOC also integrates many necessary components for the whole embedded application. Some integrated SOCS include display, graphics, video acceleration and industrial networking functions, so that the function of single-chip solution is not limited to running ml / AI.
Jacinto of Ti ™ 7 series processor is such a highly integrated SOC. The chip includes high-performance computing, deep learning engine and special accelerator for signal and image processing, which meets the functional safety asil-d / sil-3 standard. In addition to advanced driving assistance system (ADAS), the processor can also be used in robot, machine vision, radar and other fields.
The integrated special accelerator includes the "c7x" new generation DSP kernel, which combines ti's industry-leading DSP and Eve kernel, adds vector floating-point calculation function, and supports backward compatible code. With the rise of edge artificial intelligence, DSP can significantly improve the matrix operation efficiency because it is based on Harvard architecture, which is very suitable for neural network computing acceleration. At the same time, the new "MMA" deep learning accelerator can achieve the computing performance of 8 tops at low power under typical working conditions.
The general kernel includes multi-core arm cortex-a72, cortex-r5f and 8XE ge8430 GPU.
Jacinto 7 Series multi-core heterogeneous processor architecture design can maximize the selection and optimization of tasks, so as to achieve better performance improvement and cost control. In addition, Ti also hardware mature algorithms and the evolution of semiconductor process, so as to achieve the best cost performance and power consumption ratio. For example, TI's ISP can automatically realize wide dynamic adjustment, image pyramid scaling, stereo depth vision and dense optical flow algorithm acceleration based on the hardware acceleration unit embedded in the chip.
Jacinto 7 series processors provide a comprehensive security solution involving hardware and software, which is an important concern in the automotive and industrial markets. Jacinto 7 series processors use the hardware development process certified by an independent functional safety assessment organization (such as t Ü v s Ü d) to design the system for asil-d functions. In view of the new challenge of high bandwidth multi port brought by ADAS data fusion, Jacinto 7 series also integrates csi-2 and other multi-channel ports, which can ensure interconnection with multi-channel sensors and support high bandwidth data requirements. Jacinto 7 Series integrates both PCIe hubs and Gigabit Ethernet switches, which can be used for domain controllers to achieve a higher level of integration.
In order to facilitate users' development, Ti launched Ti edge AI cloud, which evaluates the cloud tool for AI reasoning on Jacinto processor and supports many industry-wide and popular deep learning frameworks (including tensorflow Lite, onnx runtime, OpenGL es, etc.), helping to easily compile and deploy models and accelerate reasoning.
In addition to CNN commonly used in visual recognition, Jacinto 7 processor also provides corresponding support for RNN required by edge artificial intelligence scenes such as predictive maintenance. In addition, the industrial application processor sitaratm series of Ti integrates the arm cortex-a series kernel. Similarly, it can also realize edge artificial intelligence applications with relatively low computing power requirements through arm NN, such as predictive maintenance in industrial applications.
From unrestrained to rapid landing
In addition to perception and decision-making, Ti also has many processors, motor drivers and various simulators in the execution link, which shows that Ti can support the key aspects required to realize edge AI through a wide product portfolio. Technology serves people. As a semiconductor company with a history of 90 years, TI's secret is to constantly study the changes of social life, insight and meet people's needs, and constantly change itself.
In the 35 years since Ti entered China, it has helped Chinese customers realize innovation again and again. In the field of artificial intelligence, Howard saw that China's development has been ahead in some places. For example, in the field of 4D imaging radar, foreign countries are still in the stage of laboratory concept verification, while domestic car manufacturers have proposed the specific time point of mass production. At the same time, Chinese customers are also actively introducing in the markets of intelligent industry, robot, smart home and medical image analysis. Sometimes Howard will inspire his American colleagues with the Innovations Adopted by Chinese customers.
When making product definition, ti will conduct in-depth cooperation with some customers to make the demand for innovation reflected in the product architecture, which is not only to meet the existing use, but also to jointly foresee the future demand. The rich imagination of Chinese customers also enables the demand from China to be quickly reflected in TI's next-generation product development and oriented to global innovative industries.
Thirty five years ago, Ti entered China and made joint innovation with Chinese local partners in many markets such as industry, consumption, communication and automobile. Now, with the expansion of new infrastructure projects represented by artificial intelligence and edge artificial intelligence, Ti and its partners have once again stepped into a new field to solve the new challenges brought by edge artificial intelligence through a full range of simulation and embedded processing products, strong local manufacturing and R & D capabilities, and product distribution and sales networks all over the country, Meet the new requirements of product design of Chinese customers. At Ti, there are a group of outstanding engineers like Howard, who strive to make the world a better place through semiconductor technology with the vision of "focusing on China and creating the world through science and technology".
Not long ago, the kitchen and water heater division of Midea Group and Ti jointly established the "Joint Laboratory of perception and interaction", which aims to help Midea accelerate the development of Midea's kitchen heat appliance applications by using TI's millimeter wave radar technology and a wide range of simulation and embedded processing products. Howard hopes Ti can help support more companies like Midea to turn their ideas into reality.
